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_ref_base_and_extent): Return the base, size and
6 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
7 (get_ref_base_and_extent_hwi): Declare.
8 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
9 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
10 (get_ref_base_and_extent_hwi): New function.
11 * cfgexpand.c (expand_debug_expr): Update call to
12 get_ref_base_and_extent.
13 * dwarf2out.c (add_var_loc_to_decl): Likewise.
14 * gimple-fold.c (get_base_constructor): Return the offset as a
15 poly_int64_pod rather than a HOST_WIDE_INT.
16 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
17 * ipa-polymorphic-call.c
18 (ipa_polymorphic_call_context::set_by_invariant)
19 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
20 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
21 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
22 rather than get_ref_base_and_extent.
23 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
25 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
27 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
28 call to get_ref_base_and_extent.
29 * tree-sra.c (create_access, get_access_for_expr): Likewise.
30 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
31 (stmt_kills_ref_p): Likewise.
32 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
33 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
35 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
36 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
37 when calling native_encode_expr.
38 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
39 call to get_ref_base_and_extent.
40 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
41 get_ref_base_and_extent.
42 * var-tracking.c (track_expr_p): Likewise.
44 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
45 Alan Hayward <alan.hayward@arm.com>
46 David Sherwood <david.sherwood@arm.com>
48 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
49 HOST_WIDE_INT to poly_int64_pod.
50 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
51 polynomail parameter offsets.
53 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
54 Alan Hayward <alan.hayward@arm.com>
55 David Sherwood <david.sherwood@arm.com>
57 * gengtype.c (main): Handle poly_int64_pod.
58 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
59 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
60 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
61 from HOST_WIDE_INT to poly_int64_pod.
62 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
63 (copy_cfa): New function.
64 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
65 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
66 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
67 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
68 (initial_return_save): Treat offsets as poly_ints.
69 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
70 offset is nonconstant.
71 (reg_save): Take the offset as a poly_int64. Fall back to
72 DW_CFA_expression for nonconstant offsets.
73 (queue_reg_save): Take the offset as a poly_int64.
74 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
76 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
77 Alan Hayward <alan.hayward@arm.com>
78 David Sherwood <david.sherwood@arm.com>
80 * rtl.h (operand_subword, operand_subword_force): Take the offset
81 as a poly_uint64 an unsigned int.
82 * emit-rtl.c (operand_subword, operand_subword_force): 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 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
89 'p' format code. Use INT_LIST rather than SUBREG as the example of
90 a code with an XINT and an XEXP. Remove the implication that
91 accessing an rtx field using XINT is expected to work.
92 * rtl.def (SUBREG): Change format from "ei" to "ep".
93 * rtl.h (rtunion::rt_subreg): New field.
94 (XCSUBREG): New macro.
95 (SUBREG_BYTE): Use it.
96 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
97 Update constructor accordingly.
98 (subreg_shape::operator ==): Update accordingly.
99 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
100 than an unsigned int.
101 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
102 a poly_uint64 rather than an unsigned int.
103 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
104 than an unsigned int.
105 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
106 (subreg_size_highpart_offset): Return a poly_uint64 rather than
107 an unsigned int. Take the sizes as poly_uint64s.
108 (subreg_offset_from_lsb): Return a poly_uint64 rather than
109 an unsigned int. Take the shift as a poly_uint64 rather than
111 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
112 as a poly_uint64 rather than an unsigned int.
113 (simplify_subreg_regno): Likewise.
114 (byte_lowpart_offset): Return the memory offset as a poly_int64
116 (subreg_memory_offset): Likewise. Take the subreg offset as a
117 poly_uint64 rather than an unsigned int.
118 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
119 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
120 poly_uint64 rather than an unsigned int.
121 * rtl.c (rtx_format): Describe 'p' in comment.
122 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
123 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
124 offset as a poly_uint64 rather than an unsigned int.
125 (byte_lowpart_offset): Return the memory offset as a poly_int64
127 (subreg_memory_offset): Likewise. Take the subreg offset as a
128 poly_uint64 rather than an unsigned int.
129 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
130 mode sizes as poly_uint64s rather than unsigned ints. Return a
131 poly_uint64 rather than an unsigned int.
132 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
133 (copy_insn_1): Handle 'p'.
134 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
135 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
136 an unsigned int. Return the shift in the same way.
137 (subreg_lsb): Return the shift as a poly_uint64 rather than an
139 (subreg_size_offset_from_lsb): Take the sizes and shift as
140 poly_uint64s rather than unsigned ints. Return the offset as
142 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
143 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
145 * rtlhash.c (add_rtx): Handle 'p'.
146 * genemit.c (gen_exp): Likewise.
147 * gengenrtl.c (type_from_format, gendef): Likewise.
148 * gensupport.c (subst_pattern_match, get_alternatives_number)
149 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
150 (subst_dup): Likewise.
151 * gengtype.c (adjust_field_rtx_def): Likewise.
152 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
153 (match_pattern_2): Likewise.
154 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
155 (rtx_test::subreg_field): New function.
156 (operator ==, safe_to_hoist_p, transition_parameter_type)
157 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
158 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
159 * genpeep.c (match_rtx): Likewise.
160 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
161 (rtx_writer::print_rtx_operand): Handle 'p'.
162 (print_value): Handle SUBREG.
163 * read-rtl.c (apply_int_iterator): Likewise.
164 (rtx_reader::read_rtx_operand): Handle 'p'.
165 * alias.c (rtx_equal_for_memref_p): Likewise.
166 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
167 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
169 * calls.c (expand_call): Likewise.
170 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
171 (make_extraction, gen_lowpart_for_combine): Likewise.
172 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
174 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
175 rather than an unsigned int. Treat subreg offsets as poly_ints.
176 (exp_equiv_p): Handle 'p'.
177 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
178 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
179 * dse.c (find_shift_sequence): Likewise.
180 * dwarf2out.c (rtl_for_decl_location): Likewise.
181 * expmed.c (extract_low_bits): Likewise.
182 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
183 (expand_expr_real_2): Likewise.
184 * final.c (alter_subreg): Likewise.
185 (leaf_renumber_regs_insn): Handle 'p'.
186 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
187 Treat subreg offsets as poly_ints.
188 * fwprop.c (forward_propagate_and_simplify): Likewise.
189 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
190 * ira.c (get_subreg_tracking_sizes): Likewise.
191 * ira-conflicts.c (go_through_subreg): Likewise.
192 * ira-lives.c (process_single_reg_class_operands): Likewise.
193 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
194 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
195 as a poly_uint64 rather than an unsigned int.
196 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
197 subreg offsets as poly_ints.
198 * lra-constraints.c (operands_match_p): Handle 'p'.
199 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
200 * lra-spills.c (assign_mem_slot): Likewise.
201 * postreload.c (move2add_valid_value_p): Likewise.
202 * recog.c (general_operand, indirect_operand): Likewise.
203 * regcprop.c (copy_value, maybe_mode_change): Likewise.
204 (copyprop_hardreg_forward_1): Likewise.
205 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
206 (record_subregs_of_mode): Likewise.
207 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
208 * reload.c (operands_match_p): Handle 'p'.
209 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
210 * reload1.c (alter_reg, choose_reload_regs): Likewise.
211 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
212 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
213 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
214 (simplify_const_poly_int_tests<N>::run): Likewise.
215 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
216 a poly_uint64 rather than an unsigned int.
217 * valtrack.c (debug_lowpart_subreg): Likewise.
218 * var-tracking.c (var_lowpart): Likewise.
219 (loc_cmp): Handle 'p'.
221 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
222 Alan Hayward <alan.hayward@arm.com>
223 David Sherwood <david.sherwood@arm.com>
225 * ira.c (get_subreg_tracking_sizes): New function.
226 (init_live_subregs): Take an integer size rather than a register.
227 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
228 to init_live_subregs.
230 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
231 Alan Hayward <alan.hayward@arm.com>
232 David Sherwood <david.sherwood@arm.com>
234 * expr.c (store_constructor_field): Change bitsize from a
235 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
236 HOST_WIDE_INT to a poly_int64.
237 (store_constructor): Change size from a HOST_WIDE_INT to
239 (store_field): Likewise bitsize and bitpos.
241 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
242 Alan Hayward <alan.hayward@arm.com>
243 David Sherwood <david.sherwood@arm.com>
245 * expmed.h (store_bit_field): Change bitregion_start and
246 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
247 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
248 (store_bit_field_1, store_integral_bit_field, store_bit_field)
249 (store_fixed_bit_field, store_split_bit_field): Likewise.
250 * expr.c (store_constructor_field, store_field): Likewise.
251 (optimize_bitfield_assignment_op): Likewise. Make the same change
252 to bitsize and bitpos.
253 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
254 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
255 same change in the constructor arguments.
256 (get_best_mode): Change bitregion_start and bitregion_end from
257 unsigned HOST_WIDE_INT to poly_uint64.
258 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
259 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
261 (bit_field_mode_iterator::next_mode): Update for new types
262 of m_bitregion_start and m_bitregion_end.
263 (get_best_mode): Change bitregion_start and bitregion_end from
264 unsigned HOST_WIDE_INT to poly_uint64.
266 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
267 Alan Hayward <alan.hayward@arm.com>
268 David Sherwood <david.sherwood@arm.com>
270 * rtl.h (simplify_gen_subreg): Add a temporary overload that
271 accepts poly_uint64 offsets.
272 * expmed.h (extract_bit_field): Take bitsize and bitnum as
273 poly_uint64s rather than unsigned HOST_WIDE_INTs.
274 * expmed.c (lowpart_bit_field_p): Likewise.
275 (extract_bit_field_as_subreg): New function, split out from...
276 (extract_bit_field_1): ...here. Take bitsize and bitnum as
277 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
278 extractions, check that BITSIZE matches the size of the extracted
279 value and that BITNUM is an exact multiple of that size.
280 If all else fails, try forcing the value into memory if
281 BITNUM is variable, and adjusting the address so that the
282 offset is constant. Split the part that can only handle constant
283 bitsize and bitnum out into...
284 (extract_integral_bit_field): ...this new function.
285 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
286 rather than unsigned HOST_WIDE_INTs.
288 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
289 Alan Hayward <alan.hayward@arm.com>
290 David Sherwood <david.sherwood@arm.com>
292 * expmed.h (store_bit_field): Take bitsize and bitnum as
293 poly_uint64s rather than unsigned HOST_WIDE_INTs.
294 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
295 that returns the byte size.
296 (store_bit_field_1): Take bitsize and bitnum as
297 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
298 to simple_mem_bitfield_p. Split the part that can only handle
299 constant bitsize and bitnum out into...
300 (store_integral_bit_field): ...this new function.
301 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
302 than unsigned HOST_WIDE_INTs.
303 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
305 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
306 Alan Hayward <alan.hayward@arm.com>
307 David Sherwood <david.sherwood@arm.com>
309 * lra-int.h (lra_reg): Change offset from int to poly_int64.
310 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
312 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
313 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
314 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
315 offset as a poly_int64 rather than an int.
316 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
317 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
318 * lra-constraints.c (equiv_address_substitution): Track offsets
320 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
321 (curr_insn_transform): Handle the new form of sp_offset.
322 * lra-eliminations.c (lra_elim_table): Change previous_offset
323 and offset from HOST_WIDE_INT to poly_int64.
324 (print_elim_table, update_reg_eliminate): Update accordingly.
325 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
326 (get_elimination): Update accordingly.
327 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
328 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
329 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
330 poly_int64 offsets generally.
331 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
332 (mark_not_eliminable, init_elimination): Update accordingly.
333 (remove_reg_equal_offset_note): Return a bool and pass the new
334 offset back by pointer as a poly_int64.
335 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
336 rather than a HOST_WIDE_INT.
337 (do_remat): Track offsets poly_int64s.
338 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
340 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
341 Alan Hayward <alan.hayward@arm.com>
342 David Sherwood <david.sherwood@arm.com>
344 * rtl.h (mem_attrs): Add a default constructor. Change size and
345 offset from HOST_WIDE_INT to poly_int64.
346 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
347 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
348 (widen_memory_access): Take the sizes and offsets as poly_int64s
349 rather than HOST_WIDE_INTs.
350 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
351 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
353 (adjust_offset_for_component_ref): Change the offset from a
354 HOST_WIDE_INT to a poly_int64.
355 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
356 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
357 * dce.c (find_call_stack_args): Likewise.
358 * dse.c (record_store): Likewise.
359 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
360 * print-rtl.c (rtx_writer::print_rtx): Likewise.
361 * read-rtl-function.c (test_loading_mem): Likewise.
362 * rtlanal.c (may_trap_p_1): Likewise.
363 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
364 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
365 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
366 (mem_attrs::mem_attrs): New function.
367 (set_mem_attributes_minus_bitpos): Change bitpos from a
368 HOST_WIDE_INT to poly_int64.
369 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
370 (clear_mem_offset, clear_mem_size, change_address)
371 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
372 initialize mem_attrs.
373 (set_mem_offset, set_mem_size, adjust_address_1)
374 (adjust_automodify_address_1, offset_address, widen_memory_access):
375 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
377 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
378 Alan Hayward <alan.hayward@arm.com>
379 David Sherwood <david.sherwood@arm.com>
381 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
382 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
383 rather than 0 to represent an unknown size. Assert that the size
384 is known when the mode isn't BLKmode.
385 (may_trap_p_1): Use -1 for unknown sizes.
386 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
388 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
389 Alan Hayward <alan.hayward@arm.com>
390 David Sherwood <david.sherwood@arm.com>
392 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
393 to poly_int64. Update commentary for positions_needed.large.
394 (read_info_type): Change offset and width from HOST_WIDE_INT
396 (set_usage_bits): Likewise.
397 (canon_address): Return the offset as a poly_int64 rather than
398 a HOST_WIDE_INT. Use strip_offset_and_add.
399 (set_all_positions_unneeded, any_positions_needed_p): Use
400 positions_needed.large to track stores with non-constant widths.
401 (all_positions_needed_p): Likewise. Take the offset and width
402 as poly_int64s rather than ints. Assert that rhs is nonnull.
403 (record_store): Cope with non-constant offsets and widths.
404 Nullify the rhs of an earlier store if we can't tell which bytes
406 (find_shift_sequence): Take the access_size and shift as poly_int64s
408 (get_stored_val): Take the read_offset and read_width as poly_int64s
409 rather than HOST_WIDE_INTs.
410 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
411 non-constant offsets and widths.
413 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
414 Alan Hayward <alan.hayward@arm.com>
415 David Sherwood <david.sherwood@arm.com>
417 * inchash.h (inchash::hash::add_poly_int): New function.
418 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
419 Use poly_int64 rather than HOST_WIDE_INT.
420 (ao_ref::max_size_known_p): New function.
421 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
422 rather than HOST_WIDE_INT.
423 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
424 to temporaries until its interface is adjusted to match.
425 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
426 (aliasing_component_refs_p, decl_refs_may_alias_p)
427 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
428 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
429 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
431 * alias.c (ao_ref_from_mem): Likewise.
432 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
433 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
434 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
435 (maybe_trim_complex_store, maybe_trim_constructor_store)
436 (live_bytes_read, dse_classify_store): Likewise.
437 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
438 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
439 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
440 (vn_reference_lookup_3): Likewise.
441 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
443 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
444 Alan Hayward <alan.hayward@arm.com>
445 David Sherwood <david.sherwood@arm.com>
447 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
448 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
449 instead of ranges_overlap_p.
451 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
452 Alan Hayward <alan.hayward@arm.com>
453 David Sherwood <david.sherwood@arm.com>
455 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
456 sizes as poly_int64s rather than HOST_WIDE_INTs.
458 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
459 Alan Hayward <alan.hayward@arm.com>
460 David Sherwood <david.sherwood@arm.com>
462 * gimple-fold.h (fold_ctor_reference): Take the offset and size
463 as poly_uint64 rather than unsigned HOST_WIDE_INT.
464 * gimple-fold.c (fold_ctor_reference): Likewise.
466 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
467 Alan Hayward <alan.hayward@arm.com>
468 David Sherwood <david.sherwood@arm.com>
470 * target.def (dwarf_poly_indeterminate_value): New hook.
471 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
472 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
473 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
474 * doc/tm.texi: Regenerate.
475 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
476 offset as a poly_int64.
477 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
478 offset as a poly_int64.
479 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
480 Take the offset as a poly_int64.
481 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
482 (frame_pointer_fb_offset): Change to a poly_int64.
483 (int_loc_descriptor): Take the offset as a poly_int64. Use
484 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
485 (based_loc_descr): Take the offset as a poly_int64.
486 Use strip_offset_and_add to handle (plus X (const)).
487 Use new_reg_loc_descr instead of an open-coded version of the
488 previous implementation.
489 (mem_loc_descriptor): Handle CONST_POLY_INT.
490 (compute_frame_pointer_to_fb_displacement): Take the offset as a
491 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
493 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
494 Alan Hayward <alan.hayward@arm.com>
495 David Sherwood <david.sherwood@arm.com>
497 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
498 (gen_rtx_REG_offset): Take the offset as a poly_int64.
499 * inchash.h (inchash::hash::add_poly_hwi): New function.
500 * gengtype.c (main): Register poly_int64.
501 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
502 offset as a poly_int.
503 (reg_attr_hasher::equal): Use must_eq to compare offsets.
504 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
505 offset as a poly_int64.
506 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
507 * print-rtl.c (print_poly_int): New function.
508 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
510 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
512 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
513 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
514 (vt_get_decl_and_offset): Return the offset as a poly_int64.
515 Enforce track_offset_p for parts of a PARALLEL.
516 (vt_add_function_parameter): Use const_offset for the final
517 offset to track. Use get_tracked_reg_offset for the parts
520 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
521 Alan Hayward <alan.hayward@arm.com>
522 David Sherwood <david.sherwood@arm.com>
524 * target.def (truly_noop_truncation): Take poly_uint64s instead of
525 unsigned ints. Change default to hook_bool_puint64_puint64_true.
526 * doc/tm.texi: Regenerate.
527 * hooks.h (hook_bool_uint_uint_true): Delete.
528 (hook_bool_puint64_puint64_true): Declare.
529 * hooks.c (hook_bool_uint_uint_true): Delete.
530 (hook_bool_puint64_puint64_true): New function.
531 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
532 instead of unsigned ints.
533 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
534 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
536 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
537 Alan Hayward <alan.hayward@arm.com>
538 David Sherwood <david.sherwood@arm.com>
540 * optabs.h (expand_operand): Add an int_value field.
541 (create_expand_operand): Add an int_value parameter and use it
542 to initialize the new expand_operand field.
543 (create_integer_operand): Replace with a declaration of a function
544 that accepts poly_int64s. Move the implementation to...
545 * optabs.c (create_integer_operand): ...here.
546 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
547 the mode preserves the value of int_value, instead of calling
548 const_int_operand on the rtx. Use gen_int_mode to generate
551 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
552 Alan Hayward <alan.hayward@arm.com>
553 David Sherwood <david.sherwood@arm.com>
555 * dumpfile.h (dump_dec): Declare.
556 * dumpfile.c (dump_dec): New function.
557 * pretty-print.h (pp_wide_integer): Turn into a function and
558 declare a poly_int version.
559 * pretty-print.c (pp_wide_integer): New function for poly_ints.
561 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
562 Alan Hayward <alan.hayward@arm.com>
563 David Sherwood <david.sherwood@arm.com>
565 * doc/generic.texi (POLY_INT_CST): Document.
566 * tree.def (POLY_INT_CST): New tree code.
567 * treestruct.def (TS_POLY_INT_CST): New tree layout.
568 * tree-core.h (tree_poly_int_cst): New struct.
569 (tree_node): Add a poly_int_cst field.
570 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
571 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
572 instead of a wide_int_ref.
573 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
575 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
576 instead of an unsigned HOST_WIDE_INT.
577 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
578 (ptrdiff_tree_p): Declare.
579 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
580 extern inline implementations if the target doesn't use POLY_INT_CST.
581 (poly_int_tree_p): New function.
582 (wi::unextended_tree): New class.
583 (wi::int_traits <unextended_tree>): New override.
584 (wi::extended_tree): Add a default constructor.
585 (wi::extended_tree::get_tree): New function.
586 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
587 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
588 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
589 (wi::tree_to_poly_wide_ref): New typedefs.
590 (wi::ints_for): Provide overloads for extended_tree and
592 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
593 (wi::to_wide): New functions.
594 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
595 * tree.c (poly_int_cst_hasher): New struct.
596 (poly_int_cst_hash_table): New variable.
597 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
598 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
600 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
601 (init_ttree): Initialize poly_int_cst_hash_table.
602 (build_int_cst, build_int_cst_type, build_invariant_address): Take
603 a poly_int64 instead of a HOST_WIDE_INT.
604 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
605 instead of an unsigned HOST_WIDE_INT.
606 (wide_int_to_tree): Rename to...
607 (wide_int_to_tree_1): ...this.
608 (build_new_poly_int_cst, build_poly_int_cst): New functions.
609 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
610 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
611 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
612 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
613 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
615 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
616 (streamer_read_tree_body): Likewise.
617 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
618 (streamer_write_tree_body): Likewise.
619 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
620 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
621 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
622 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
623 * gimple-expr.h (is_gimple_constant): Likewise.
624 * gimplify.c (maybe_with_size_expr): Likewise.
625 * print-tree.c (print_node): Likewise.
626 * tree-data-ref.c (data_ref_compare_tree): Likewise.
627 * tree-pretty-print.c (dump_generic_node): Likewise.
628 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
629 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
630 * tree-vrp.c (compare_values_warnv): Likewise.
631 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
632 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
633 (force_expr_to_var_cost): Likewise.
634 * tree-ssa-loop.c (for_each_index): Likewise.
635 * fold-const.h (build_invariant_address, size_int_kind): Take a
636 poly_int64 instead of a HOST_WIDE_INT.
637 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
638 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
640 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
641 (int_const_binop_2): New function, split out from...
642 (int_const_binop_1): ...here. Handle POLY_INT_CST.
643 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
644 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
645 * gimple-ssa-strength-reduction.c (slsr_process_add)
646 (slsr_process_mul): Check for INTEGER_CSTs before using them
648 * stor-layout.c (bits_from_bytes): New function.
649 (bit_from_pos): Use it.
650 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
651 by BITS_PER_UNIT to get the TYPE_SIZE.
652 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
653 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
655 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
656 Alan Hayward <alan.hayward@arm.com>
657 David Sherwood <david.sherwood@arm.com>
659 * doc/rtl.texi (const_poly_int): Document. Also document the
660 rtl sharing behavior.
661 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
662 * rtl.h (const_poly_int_def): New struct.
663 (rtx_def::u): Add a cpi field.
664 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
665 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
666 (wi::rtx_to_poly_wide_ref): New typedef
667 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
668 (poly_int_rtx_p): New functions.
669 (trunc_int_for_mode): Declare a poly_int64 version.
670 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
671 (immed_wide_int_const): Take a poly_wide_int_ref rather than
673 (strip_offset): Declare.
674 (strip_offset_and_add): New function.
675 * rtl.def (CONST_POLY_INT): New rtx code.
676 * rtl.c (rtx_size): Handle CONST_POLY_INT.
677 (shared_const_p): Use poly_int_rtx_p.
678 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
680 (gen_int_shift_amount): Likewise.
681 * emit-rtl.c (const_poly_int_hasher): New class.
682 (const_poly_int_htab): New variable.
683 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
684 (const_poly_int_hasher::hash): New function.
685 (const_poly_int_hasher::equal): Likewise.
686 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
687 (immed_wide_int_const): Rename to...
688 (immed_wide_int_const_1): ...this and make static.
689 (immed_wide_int_const): New function, taking a poly_wide_int_ref
690 instead of a wide_int_ref.
691 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
692 (gen_lowpart_common): Handle CONST_POLY_INT.
693 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
694 * cselib.c (cselib_hash_rtx): Likewise.
695 * dwarf2out.c (const_ok_for_output_1): Likewise.
696 * expr.c (convert_modes): Likewise.
697 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
698 * rtlhash.c (add_rtx): Likewise.
699 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
700 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
701 Handle existing CONST_POLY_INT rtxes.
702 * expmed.h (expand_shift): Take a poly_int64 instead of a
704 * expmed.c (expand_shift): Likewise.
705 * rtlanal.c (strip_offset): New function.
706 (commutative_operand_precedence): Give CONST_POLY_INT the same
707 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
709 * rtl-tests.c (const_poly_int_tests): New struct.
710 (rtl_tests_c_tests): Use it.
711 * simplify-rtx.c (simplify_const_unary_operation): Handle
713 (simplify_const_binary_operation): Likewise.
714 (simplify_binary_operation_1): Fold additions of symbolic constants
716 (simplify_subreg): Handle extensions and truncations of
718 (simplify_const_poly_int_tests): New struct.
719 (simplify_rtx_c_tests): Use it.
720 * wide-int.h (storage_ref): Add default constructor.
721 (wide_int_ref_storage): Likewise.
722 (trailing_wide_ints): Use GTY((user)).
723 (trailing_wide_ints::operator[]): Add a const version.
724 (trailing_wide_ints::get_precision): New function.
725 (trailing_wide_ints::extra_size): Likewise.
727 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
728 Alan Hayward <alan.hayward@arm.com>
729 David Sherwood <david.sherwood@arm.com>
731 * emit-rtl.h (gen_int_shift_amount): Declare.
732 * emit-rtl.c (gen_int_shift_amount): New function.
733 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
735 * calls.c (shift_return_value): Likewise.
736 * cse.c (fold_rtx): Likewise.
737 * dse.c (find_shift_sequence): Likewise.
738 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
739 (expand_shift, expand_smod_pow2): Likewise.
740 * lower-subreg.c (shift_cost): Likewise.
741 * optabs.c (expand_superword_shift, expand_doubleword_mult)
742 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
743 (expand_vec_perm_var): Likewise.
744 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
745 (simplify_binary_operation_1): Likewise.
746 * combine.c (try_combine, find_split_point, force_int_to_mode)
747 (simplify_shift_const_1, simplify_shift_const): Likewise.
748 (change_zero_ext): Likewise. Use simplify_gen_binary.
750 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
752 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
754 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
756 * doc/invoke.texi (ARM Options): Document accepted extension options
757 for -march=armv8.3-a.
759 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
762 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
763 or --with-float=softfp, set the default CPU to arm10e.
765 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
767 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
768 * config/visium/predicates.md (const_shift_operand): Likewise.
769 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
770 (visium_legitimize_reload_address): Likewise.
772 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
775 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
777 2017-12-20 Jakub Jelinek <jakub@redhat.com>
780 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
782 2017-12-20 Julia Koval <julia.koval@intel.com>
784 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
785 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
786 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
787 * config.gcc: Include vpclmulqdqintrin.h.
788 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
789 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
791 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
792 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
793 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
794 * config/i386/i386.c (isa2_opts): Add -mcx16.
795 (isa_opts): Add -mpclmulqdq, remove -mcx16.
796 (ix86_option_override_internal): Move mcx16 to flags2.
797 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
798 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
799 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
800 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
801 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
802 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
803 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
804 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
805 * doc/invoke.texi: Add -mvpclmulqdq.
807 2017-12-20 Tom de Vries <tom@codesourcery.com>
810 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
811 * calls.c (rtx_for_static_chain): ... here. New function.
812 * calls.h (rtx_for_static_chain): Declare.
813 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
814 instead of targetm.calls.static_chain.
815 * df-scan.c (df_get_entry_block_def_set): Same.
817 2017-12-19 Tom de Vries <tom@codesourcery.com>
819 PR tree-optimization/83493
820 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
823 2017-12-19 Jakub Jelinek <jakub@redhat.com>
825 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
826 inform with hardcoded english plural handling.
828 2017-12-18 Jeff Law <law@redhat.com>
830 PR tree-optimization/83477
831 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
832 a non-virtual PHI, always push a new range.
834 2017-12-19 Martin Sebor <msebor@redhat.com>
837 * builtins.c (compute_objsize): Handle non-constant offsets.
839 2017-12-19 Jakub Jelinek <jakub@redhat.com>
841 PR tree-optimization/83444
842 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
843 character load case, if get_stridx on MEM_REF's operand doesn't
844 look usable, retry with get_addr_stridx.
846 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
849 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
850 (variable_tracking_main_1): Keep markers even when VTA fails.
853 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
854 even if there are markers before it.
855 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
857 2017-12-19 Jakub Jelinek <jakub@redhat.com>
859 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
860 typical order conditions.
861 * sel-sched.c (extract_new_fences_from): Likewise.
862 * config/visium/constraints.md (J, K, L): Likewise.
863 * config/visium/predicates.md (const_shift_operand): Likewise.
864 * config/visium/visium.c (visium_legitimize_address,
865 visium_legitimize_reload_address): Likewise.
866 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
867 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
868 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
869 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
870 SET_DUMP_DETAIL): Likewise.
871 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
872 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
873 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
874 avr_set_core_architecture, avr_set_current_function,
875 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
876 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
877 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
878 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
879 avr_map_decompose, avr_fold_builtin): Likewise.
880 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
881 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
882 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
883 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
884 * config/m32c/m32c.c (m32c_conditional_register_usage,
885 m32c_address_cost): Likewise.
886 * config/m32c/predicates.md (shiftcount_operand,
887 longshiftcount_operand): Likewise.
888 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
889 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
890 can_use_cdx_ldstw): Likewise.
891 * config/nios2/nios2.h (CDX_REG_P): Likewise.
892 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
894 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
895 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
896 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
897 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
898 * config/vax/vax.c (adjacent_operands_p): Likewise.
899 * config/ft32/constraints.md (L, b, KA): Likewise.
900 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
902 * cfgexpand.c (expand_stack_alignment): Likewise.
903 * gcse.c (insert_expr_in_table): Likewise.
904 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
905 * cgraphunit.c (cgraph_node::expand): Likewise.
906 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
907 * emit-rtl.c (add_insn): Likewise.
908 * input.c (dump_location_info): Likewise.
909 * passes.c (NEXT_PASS): Likewise.
910 * read-rtl-function.c (parse_note_insn_name,
911 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
913 * sched-rgn.c (sched_rgn_init): Likewise.
914 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
915 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
916 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
917 * explow.c (eliminate_constant_term): Likewise.
918 * final.c (leaf_renumber_regs_insn): Likewise.
919 * cfgrtl.c (print_rtl_with_bb): Likewise.
920 * genhooks.c (emit_init_macros): Likewise.
921 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
922 * tree-data-ref.c (conflict_fn): Likewise.
923 * selftest.c (assert_streq): Likewise.
924 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
925 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
926 fold_binary_loc, multiple_of_p): Likewise.
927 * reload.c (push_reload, find_equiv_reg): Likewise.
928 * et-forest.c (et_nca, et_below): Likewise.
929 * dbxout.c (dbxout_symbol_location): Likewise.
930 * reorg.c (relax_delay_slots): Likewise.
931 * dojump.c (do_compare_rtx_and_jump): Likewise.
932 * gengtype-parse.c (type): Likewise.
933 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
934 simplify_const_relational_operation): Likewise.
935 * reload1.c (do_output_reload): Likewise.
936 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
937 * gengtype.c (type_for_name): Likewise.
938 * gimple-ssa-sprintf.c (format_directive): Likewise.
940 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
943 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
944 accessing it. Adjust comment.
946 2017-12-19 Jakub Jelinek <jakub@redhat.com>
949 * predict.c (zero_one_minusone): New function.
950 (apply_return_prediction): Avoid return prediction for functions
951 returning only -1, 0 and 1 values, unless they only return -1 and 0
954 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
956 * config/arc/arc.c (legitimate_scaled_address_p): Clean
957 fall-through warning.
958 (arc_compute_frame_size): Remove unused variables.
959 (arc_print_operand): Fix fprintif format.
960 (arc_can_follow_jump): Clean fall-through warning.
962 2017-12-19 Marek Polacek <polacek@redhat.com>
965 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
968 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
970 * config/arc/arc.c (overriderregs): New variable.
971 (arc_override_options): Track fixed/call saved/call options.
972 (arc_conditional_register_usage): Check against overrideregs
973 variable whenever we change register properties.
975 2017-12-19 Nathan Sidwell <nathan@acm.org>
977 * opts.c (finish_options): Don't prefix dump_base_name if it
978 already contains directories.
980 2017-12-19 Martin Liska <mliska@suse.cz>
982 PR rtl-optimization/82675
983 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
984 more element in sbitmap.
986 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
988 * gimplify.c (gimplify_expr): Use error_operand_p.
990 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
993 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
996 2017-12-19 Daniel Cederman <cederman@gaisler.com>
998 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
999 the jump is to a label.
1001 2017-12-19 Jakub Jelinek <jakub@redhat.com>
1003 PR tree-optimization/83444
1004 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
1009 * ipa-inline.c (flatten_remove_node_hook): New function.
1010 (ipa_inline): Keep only nodes with flatten attribute at the end of
1011 the array in the order from ipa_reverse_postorder, only walk that
1012 portion of array for flattening, if there is more than one such
1013 node, temporarily register a removal hook and ignore removed nodes.
1015 PR tree-optimization/80631
1016 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
1017 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
1018 IFN_REDUC_MAX or IFN_REDUC_MIN.
1020 2017-12-18 Jeff Law <law@redhat.com>
1022 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
1023 record symbolic equivalences from backedges in the CFG.
1026 2017-11-19 Jeff Law <law@redhat.com>
1028 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
1029 of degenerates resulting from ignoring an edge.
1031 2017-12-18 Martin Sebor <msebor@redhat.com>
1034 PR tree-optimization/78450
1035 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
1036 (handle_builtin_strlen): Call it.
1038 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
1040 PR rtl-optimization/83424
1041 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
1043 2017-12-18 Marek Polacek <polacek@redhat.com>
1046 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1047 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
1050 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1052 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
1053 (umaddsidi4, umaddsidi_split): Likewise.
1055 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
1057 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
1060 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1063 * config/s390/s390.c (s390_option_override): Avoid strncpy.
1065 2017-12-18 Richard Biener <rguenther@suse.de>
1067 PR tree-optimization/81877
1068 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
1069 (outermost_indep_loop): Adjust.
1070 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
1071 (can_sm_ref_p): Adjust.
1073 2017-12-18 Richard Biener <rguenther@suse.de>
1076 * tree.c (array_at_struct_end_p): Return true if the underlying
1077 object has space for at least one element in excess of what
1078 the array domain specifies.
1080 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1082 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
1083 example to avoid overfull hbox.
1084 * doc/invoke.texi (Option Summary): Add missing @gol.
1085 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
1088 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1089 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1091 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
1092 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
1093 (AArch64 Options): Document -mverbose-cost-dump.
1094 (ARM Options): Likewise, plus -mflip-thumb.
1096 2017-12-17 Martin Sebor <msebor@redhat.com>
1099 * gimple-ssa-warn-restrict.c
1100 (builtin_memref::offset_out_of_bounds): Correct the handling of
1103 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
1105 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
1107 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
1109 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
1111 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
1113 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
1116 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
1118 * doc/invoke.texi: Fix some typos.
1120 2017-12-16 Martin Sebor <msebor@redhat.com>
1122 PR tree-optimization/78918
1123 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
1124 * builtins.c (check_sizes): Rename...
1125 (check_access): ...to this. Rename function arguments for clarity.
1126 (check_memop_sizes): Adjust names.
1127 (expand_builtin_memchr, expand_builtin_memcpy): Same.
1128 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
1129 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
1130 (check_strncat_sizes, expand_builtin_strncat): Same.
1131 (expand_builtin_strncpy, expand_builtin_memset): Same.
1132 (expand_builtin_bzero, expand_builtin_memcmp): Same.
1133 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
1134 (maybe_emit_sprintf_chk_warning): Same.
1135 (expand_builtin_strcpy): Adjust.
1136 (expand_builtin_stpcpy): Same.
1137 (expand_builtin_with_bounds): Detect out-of-bounds accesses
1138 in pointer-checking forms of memcpy, memmove, and mempcpy.
1139 (gcall_to_tree_minimal, max_object_size): Define new functions.
1140 * builtins.h (max_object_size): Declare.
1141 * calls.c (alloc_max_size): Call max_object_size instead of
1142 hardcoding ssizetype limit.
1143 (get_size_range): Handle new argument.
1144 * calls.h (get_size_range): Add a new argument.
1145 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
1146 * doc/invoke.texi (-Wrestrict): Adjust, add example.
1147 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
1149 (gimple_fold_builtin_memory_chk): Same.
1150 (gimple_fold_builtin_stxcpy_chk): New function.
1151 * gimple-ssa-warn-restrict.c: New source.
1152 * gimple-ssa-warn-restrict.h: New header.
1153 * gimple.c (gimple_build_call_from_tree): Propagate location.
1154 * passes.def (pass_warn_restrict): Add new pass.
1155 * tree-pass.h (make_pass_warn_restrict): Declare.
1156 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
1158 (handle_builtin_strcat): Same.
1159 (strlen_optimize_stmt): Rename...
1160 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
1161 stpncpy, strncpy, and their checking forms.
1163 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
1165 PR rtl-optimization/82849
1166 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
1167 and get_max_loop_iterations_int.
1169 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1170 Alan Hayward <alan.hayward@arm.com>
1171 David Sherwood <david.sherwood@arm.com>
1173 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
1174 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
1175 size as a poly_uint64.
1176 (mode_for_vector, mode_for_int_vector): Take the number of vector
1177 elements as a poly_uint64.
1178 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
1179 size as a poly_uint64.
1180 (mode_for_vector, mode_for_int_vector): Take the number of vector
1181 elements as a poly_uint64.
1183 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1184 Alan Hayward <alan.hayward@arm.com>
1185 David Sherwood <david.sherwood@arm.com>
1187 * machmode.h (MACRO_MODE): New macro.
1188 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
1189 * targhooks.c (default_libcall_value, default_secondary_reload)
1190 (default_memory_move_cost, default_register_move_cost)
1191 (default_class_max_nregs): Likewise.
1193 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1194 Alan Hayward <alan.hayward@arm.com>
1195 David Sherwood <david.sherwood@arm.com>
1197 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
1198 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
1200 * genautomata.c (main): Likewise.
1201 * genconditions.c (write_header): Likewise.
1202 * genemit.c (main): Likewise.
1203 * genextract.c (print_header): Likewise.
1204 * genopinit.c (main): Likewise.
1205 * genoutput.c (output_prologue): Likewise.
1206 * genpeep.c (main): Likewise.
1207 * genpreds.c (write_insn_preds_c): Likewise.
1208 * genrecog.c (writer_header): Likewise.
1209 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
1210 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
1211 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
1212 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
1213 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
1214 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
1215 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
1216 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
1217 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
1218 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
1219 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
1220 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
1221 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
1222 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
1223 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
1224 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
1225 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
1226 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
1227 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
1228 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
1229 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
1230 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
1231 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
1232 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
1233 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
1234 * config/darwin.c (IN_TARGET_CODE): Likewise.
1235 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
1236 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
1237 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
1238 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
1239 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
1240 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
1241 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
1242 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
1243 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
1244 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
1245 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
1246 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
1247 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
1248 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
1249 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
1250 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
1251 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
1252 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
1253 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
1254 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
1255 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
1256 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
1257 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
1258 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
1259 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
1260 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
1261 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
1262 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
1263 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
1264 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
1265 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
1266 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
1267 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
1268 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
1269 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
1270 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
1271 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
1272 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
1273 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
1274 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
1275 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
1276 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
1277 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
1278 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
1279 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
1280 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
1281 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
1282 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
1283 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
1284 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
1285 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
1286 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
1287 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
1288 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
1289 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
1290 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
1291 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
1292 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
1293 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
1294 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
1295 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1296 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
1297 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
1298 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
1299 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
1300 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
1301 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
1302 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
1303 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
1304 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
1305 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
1306 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
1307 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
1308 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
1309 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
1310 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
1311 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
1312 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
1313 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
1314 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
1315 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
1316 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
1317 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
1318 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
1319 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
1320 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
1321 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
1322 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
1323 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
1324 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
1325 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
1326 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
1327 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
1328 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
1329 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
1330 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
1331 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
1332 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
1333 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
1334 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
1335 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
1336 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
1337 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
1338 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
1339 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
1340 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
1342 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1343 Alan Hayward <alan.hayward@arm.com>
1344 David Sherwood <david.sherwood@arm.com>
1346 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
1349 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1350 Alan Hayward <alan.hayward@arm.com>
1351 David Sherwood <david.sherwood@arm.com>
1353 * doc/generic.texi (VEC_SERIES_EXPR): Document.
1354 * doc/md.texi (vec_series@var{m}): Document.
1355 * tree.def (VEC_SERIES_EXPR): New tree code.
1356 * tree.h (build_vec_series): Declare.
1357 * tree.c (build_vec_series): New function.
1358 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
1359 * tree-pretty-print.c (dump_generic_node): Likewise.
1360 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
1361 * tree-inline.c (estimate_operator_cost): Likewise.
1362 * expr.c (expand_expr_real_2): Likewise.
1363 * optabs-tree.c (optab_for_tree_code): Likewise.
1364 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
1365 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
1366 * expmed.c (make_tree): Handle VEC_SERIES.
1367 * optabs.def (vec_series_optab): New optab.
1368 * optabs.h (expand_vec_series_expr): Declare.
1369 * optabs.c (expand_vec_series_expr): New function.
1370 * tree-vect-generic.c (expand_vector_operations_1): Check that
1371 the operands also have vector type.
1373 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
1374 Alan Hayward <alan.hawyard@arm.com>
1375 David Sherwood <david.sherwood@arm.com>
1377 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
1378 (VEC_COND_EXPR): Add missing @tindex.
1379 * doc/md.texi (vec_duplicate@var{m}): Document.
1380 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
1381 * tree.c (build_vector_from_val): Add stubbed-out handling of
1382 variable-length vectors, using VEC_DUPLICATE_EXPR.
1383 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
1384 * cfgexpand.c (expand_debug_expr): Likewise.
1385 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
1386 * tree-inline.c (estimate_operator_cost): Likewise.
1387 * tree-pretty-print.c (dump_generic_node): Likewise.
1388 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
1389 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
1390 (test_vec_duplicate_folding): New function.
1391 (fold_const_c_tests): Call it.
1392 * optabs.def (vec_duplicate_optab): New optab.
1393 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
1394 * optabs.h (expand_vector_broadcast): Declare.
1395 * optabs.c (expand_vector_broadcast): Make non-static. Try using
1396 vec_duplicate_optab.
1397 * expr.c (store_constructor): Try using vec_duplicate_optab for
1399 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
1401 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
1404 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
1405 div/mod latencies a bit.
1407 2017-12-15 Jeff Law <law@redhat.com>
1409 PR tree-optimization/36550
1410 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
1411 (mark_threaded_blocks): Rewrite code to avoid block copying when
1412 optimizing for size. Don't pessimize blocks which will be
1413 copied, but all the statements will be dead.
1415 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
1417 PR tree-optimization/81165
1418 * tree-ssa-threadupdate.c (uses_in_bb): New.
1419 (estimate_threading_killed_stmts): New.
1420 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
1421 * tree-ssa-threadedge.c
1422 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
1425 2017-12-15 Jeff Law <law@redhat.com>
1427 PR tree-optimization/83410
1428 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
1429 threads when parallelizing loops.
1431 2017-12-15 Jakub Jelinek <jakub@redhat.com>
1433 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
1435 * config/alpha/alpha.c (vms_attribute_table): Swap
1436 affects_type_identity and handler fields, adjust comments.
1437 * config/mips/mips.c (mips_attribute_table): Likewise.
1438 * config/visium/visium.c (visium_attribute_table): Likewise.
1439 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
1440 * config/microblaze/microblaze.c (microblaze_attribute_table):
1442 * config/spu/spu.c (spu_attribute_table): Likewise.
1443 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
1444 * config/arc/arc.c (arc_attribute_table): Likewise.
1445 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
1446 * config/v850/v850.c (v850_handle_interrupt_attribute,
1447 v850_handle_data_area_attribute): Formatting fixes.
1448 (v850_attribute_table): Swap affects_type_identity and handler
1449 fields, adjust comments.
1450 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
1451 * config/arm/arm.c (arm_attribute_table): Likewise.
1452 * config/avr/avr.c (avr_attribute_table): Likewise.
1453 * config/s390/s390.c (s390_attribute_table): Likewise.
1454 * config/sh/sh.c (sh_attribute_table): Likewise.
1455 * config/i386/i386.c (ix86_handle_cconv_attribute,
1456 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
1457 (ix86_attribute_table): Swap affects_type_identity and handler
1458 fields, adjust comments.
1459 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1460 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
1461 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
1462 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
1463 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
1464 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1465 * config/rx/rx.c (rx_attribute_table): Likewise.
1466 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
1467 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
1468 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
1469 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
1470 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1471 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
1472 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
1473 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
1474 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1475 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1476 * doc/plugins.texi (user_attr): Likewise. Add NULL for
1478 * attribs.c (empty_attribute_table): Swap affects_type_identity and
1480 (register_scoped_attributes, decl_attributes): Formatting fixes.
1482 PR tree-optimization/83269
1483 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
1484 subtraction in arg0's type if type is signed and arg0 is unsigned.
1488 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
1489 convert? on @0 instead of convert. Check type of @1, not @0.
1490 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
1491 convert? on @0 instead of convert. Check type of @1, not @0.
1492 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
1493 only optimize if either both @1 and @2 types are narrower
1494 precision, or both are wider or equal precision, and in the former
1495 case only if both have undefined overflow.
1497 2017-12-15 Richard Biener <rguenther@suse.de>
1500 * internal-fn.def (IFN_NOP): Add.
1501 * internal-fn.c (expand_NOP): Do nothing.
1502 * lto-streamer-in.c (input_function): Instead of removing
1503 sanitizer calls replace them with IFN_NOP calls.
1505 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
1506 Alan Hayward <alan.hayward@arm.com>
1507 David Sherwood <david.sherwood@arm.com>
1509 * dse.c (store_info, read_info_type): Replace begin and end with
1511 (print_range): New function.
1512 (set_all_positions_unneeded, any_positions_needed_p)
1513 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
1515 (record_store): Likewise. Optimize the case in which all positions
1517 (get_stored_val): Replace read_begin and read_end with read_offset
1519 (replace_read): Update call accordingly.
1521 2017-12-15 Bin Cheng <bin.cheng@arm.com>
1523 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
1524 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
1525 (loop_cand::loop_cand): Initialize above members.
1526 (loop_cand::supported_operations): Delete.
1527 (loop_cand::can_interchange_p): Inline above function.
1528 (loop_cand::classify_simple_reduction): Record number of constant
1529 initialized simple reductions.
1530 (should_interchange_loops): New parameters. Check stmt cost of loops
1532 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
1533 Update call to should_interchange_loops.
1534 (should_interchange_loop_nest): Update call to
1535 should_interchange_loops.
1537 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
1540 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
1541 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
1542 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
1544 2017-12-15 Julia Koval <julia.koval@intel.com>
1546 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
1547 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
1548 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
1549 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
1550 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
1552 2017-12-15 Julia Koval <julia.koval@intel.com>
1554 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
1555 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
1556 * config/i386/sse.md (vaesenc_<mode>): New pattern.
1557 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
1558 _mm_aesenc_epi128): New intrinsics.
1560 2017-12-15 Julia Koval <julia.koval@intel.com>
1562 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
1563 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
1564 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
1565 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
1566 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
1568 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1570 * gimple-ssa-strength-reduction.c (analyze_increments):
1571 Distinguish replacement costs for constant strides from those for
1574 2017-12-14 Jakub Jelinek <jakub@redhat.com>
1576 * var-tracking.c (variable_tracking_main_1): Formatting fix.
1578 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
1580 * doc/invoke.texi: Document -Wcast-function-type.
1581 * recog.h (stored_funcptr): Change signature.
1582 * tree-dump.c (dump_node): Avoid warning.
1583 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
1585 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
1588 * gimple-fold.c (get_range_strlen): Add the handling of non-member
1591 2017-12-14 David Malcolm <dmalcolm@redhat.com>
1593 PR tree-optimization/83312
1594 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
1595 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
1596 "val" param, and to cope with arbitrary basic blocks.
1597 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
1598 handle NULL_TREE for "val", dropping "bb" param.
1599 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
1600 drop "bb" param. Handle NULL_TREE for "val".
1601 (find_case_label_for_value): Make "switch_stmt" param const.
1602 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
1604 (vrp_prop::check_all_array_refs): Reimplement as...
1605 (check_array_bounds_dom_walker::before_dom_children): ...this new
1606 vfunc. Replace linear search through BB block list, excluding
1607 those with non-executable in-edges via dominator walk.
1609 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1611 * config/arm/arm.opt (mverbose-cost-dump): New option.
1612 * config/arm/arm.c (arm_rtx_costs): Use it.
1614 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
1617 * reload1.c (emit_input_reload_insns): Skip debug markers.
1619 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
1621 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
1625 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
1626 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
1627 * config/sh/sh.c: Skip debug insns besides notes.
1628 * config/sh/sh.md: Likewise.
1629 * config/sh/sh_treg_combine.cc: Likewise.
1630 * config/sh/sync.md: Likewise.
1632 2017-12-14 Tom de Vries <tom@codesourcery.com>
1634 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
1635 Add item for weak_undefined.
1637 2017-12-14 Richard Biener <rguenther@suse.de>
1639 PR tree-optimization/67842
1640 * sese.h (bb_in_region): Remove #if 0'ed code.
1642 2017-12-14 Richard Biener <rguenther@suse.de>
1644 PR tree-optimization/83326
1645 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
1646 may_be_zero parameter and handle it by not marking the first
1647 peeled copy as not exiting the loop.
1648 (try_peel_loop): Likewise.
1649 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
1650 to handle the case of constant or zero iterations and perform
1651 loop header copying on-the-fly.
1653 2017-12-14 Richard Biener <rguenther@suse.de>
1655 PR tree-optimization/83418
1656 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
1657 Instead of asserting we don't get unfolded comparisons deal with
1660 2017-12-14 Jakub Jelinek <jakub@redhat.com>
1663 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
1664 basic blocks. Assert debug bind insns don't appear outside of bbs,
1665 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
1668 PR tree-optimization/83198
1669 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
1670 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
1671 value if arg is a REAL_CST with incompatible type.
1673 2017-12-14 Sudakshina Das <sudi.das@arm.com>
1674 Bin Cheng <bin.cheng@arm.com>
1677 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
1679 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
1682 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1684 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
1685 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
1687 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
1688 Alan Hayward <alan.hayward@arm.com>
1689 David Sherwood <david.sherwood@arm.com>
1691 * poly-int.h: New file.
1692 * poly-int-types.h: Likewise.
1693 * coretypes.h: Include them.
1694 (POLY_INT_CONVERSION): Define.
1695 * target.def (estimated_poly_value): New hook.
1696 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
1697 * doc/tm.texi: Regenerate.
1698 * doc/poly-int.texi: New file.
1699 * doc/gccint.texi: Include it.
1700 * doc/rtl.texi: Describe restrictions on subreg modes.
1701 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
1702 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
1703 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
1704 * targhooks.h (default_estimated_poly_value): Declare.
1705 * targhooks.c (default_estimated_poly_value): New function.
1706 * target.h (estimated_poly_value): Likewise.
1707 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
1708 (wi::unary_traits): Delete.
1709 (wi::binary_traits::signed_shift_result_type): Define for
1710 offset_int << HOST_WIDE_INT, etc.
1711 (generic_wide_int::operator <<=): Define for all types and use
1712 wi::lshift instead of <<.
1713 (wi::hwi_with_prec): Add a default constructor.
1714 (wi::ints_for): New class.
1715 (operator <<): Define for all wide-int types.
1716 (operator /): New function.
1717 (operator %): Likewise.
1718 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
1719 (ASSERT_MAYBE_NE_AT): New macros.
1721 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
1722 Dominik Vogt <vogt@linux.vnet.ibm.com>
1725 * emit-rtl.c (init_emit): Remove ??? comment.
1726 * explow.c (get_dynamic_stack_size): Take known alignment of stack
1727 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
1728 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
1729 alignment of 3 virtual registers to BITS_PER_WORD.
1731 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
1733 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
1735 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
1736 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
1738 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
1741 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
1743 2017-12-13 Jakub Jelinek <jakub@redhat.com>
1745 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
1746 with the exception of debug begin stmt markers appear before
1750 * final.c (rest_of_handle_final): Call variable_tracking_main only
1751 if !flag_var_tracking.
1753 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
1754 Jakub Jelinek <jakub@redhat.com>
1758 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
1759 labels debug stmts that can only appear after labels.
1761 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
1763 PR rtl-optimization/82398
1764 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
1765 EXPR_USEFULNESS in priority comparison.
1767 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
1769 PR rtl-optimization/83393
1770 * combine.c (move_deaths): If reg_stat points to a too new insn in
1771 last_death, do not use it: find the proper insn instead.
1773 2017-12-12 Jeff Law <law@redhat.com>
1775 PR tree-optimization/83298
1776 PR tree-optimization/83362
1777 PR tree-optimization/83383
1778 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
1779 push_value_range a public interface. Add new argument to
1780 record_ranges_from_stmt.
1781 * gimple-ssa-evrp-analyze.c
1782 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
1783 Update comments. Handle recording temporary equivalences.
1784 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
1785 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
1786 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
1787 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
1788 gimple-ssa-evrp-analyze.h.
1789 (record_temporary_equivalences_from_phis): Add new argument. When
1790 the PHI arg is an SSA_NAME, set the result's range to the range
1792 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
1793 from statements too.
1794 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
1795 Pass it down to children as needed.
1796 (thread_outgoing_edges): Likewise.
1797 (thread_across_edge): Likewise. Push/pop range state as needed.
1798 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
1800 2017-12-12 Julia Koval <julia.koval@intel.com>
1802 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
1803 (PTA_CANNONLAKE): Remove PTA_CLWB.
1805 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1808 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
1809 (vcondv2div2df): Likewise.
1810 (vconduv2dfv2di): Likewise.
1812 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1814 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
1815 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
1816 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
1817 ATTR_MATHFN_FPROUNDING_ERRNO.
1819 2017-12-12 Richard Biener <rguenther@suse.de>
1821 PR tree-optimization/83385
1822 * graphite-scop-detection.c (get_order, order): Remove.
1823 (bb_to_rpo): New global.
1825 (build_scops): Sort pbbs in RPO order.
1827 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
1829 * combine.c (simplify_set): Do not transform subregs to zero_extends
1830 if the destination mode is a vector mode.
1832 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1834 PR tree-optimization/83359
1835 * tree-cfg.h (fold_loop_internal_call): Declare.
1836 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
1837 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
1838 (find_loop_dist_alias): New function.
1839 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
1840 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
1843 PR tree-optimization/80631
1844 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
1845 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
1846 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
1847 hardcoding zero as the value if COND_EXPR is never true. For
1848 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
1849 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
1850 hardcoding MAX_EXPR as the reduction operation.
1851 (is_nonwrapping_integer_induction): Allow negative step.
1852 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
1853 vect_create_epilog_for_reduction, if no value is suitable, don't
1854 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
1856 2017-12-12 Richard Biener <rguenther@suse.de>
1858 PR tree-optimization/81889
1859 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
1860 range info from the non-wrapping IV instead of just the range
1863 2017-12-12 Julia Koval <julia.koval@intel.com>
1865 * config.gcc: Add vaesintrin.h.
1866 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1868 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1869 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1871 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1872 * config/i386/immintrin.h: Include vaesintrin.h.
1873 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1874 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1875 _mm_aesdec_epi128): New intrinsics.
1877 2017-12-12 Julia Koval <julia.koval@intel.com>
1879 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
1880 OPTION_MASK_ISA_VAES_UNSET): New.
1881 (ix86_handle_option): Handle -mvaes.
1882 * config/i386/cpuid.h: Define bit_VAES.
1883 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
1884 * config/i386/i386-c.c (__VAES__): New.
1885 * config/i386/i386.c (ix86_target_string): Add -mvaes.
1886 (ix86_valid_target_attribute_inner_p): Ditto.
1887 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
1888 * config/i386/i386.opt: Add -mvaes.
1889 * doc/invoke.texi: Ditto.
1891 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
1893 * debug.h (gcc_debug_hooks): Add inline_entry.
1894 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
1895 * debug.c (do_nothing_debug_hooks): Likewise.
1896 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
1897 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
1898 (dwarf2_lineno_debug_hooks): Likewise.
1900 * common.opt (gstatement-frontiers): New, setting
1901 debug_nonbind_markers_p.
1902 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
1903 * toplev.c (process_options): Autodetect value for debug statement
1905 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
1906 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
1908 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
1909 markers. Integrate source bind into debug stmt expand loop.
1910 (pass_expand::execute): Check debug marker limit. Avoid deep
1911 TER and expand debug locations for debug bind insns only.
1912 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
1914 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
1915 * final.c (reemit_insn_block_notes): Take current block from
1916 nonbind markers. Declare note where it's first set.
1917 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
1918 begin stmt markers if enabled.
1919 (notice_source_line): Handle nonbind markers. Fail if their
1920 location is unknown or that of builtins.
1921 (rest_of_handle_final): Convert begin stmt markers to notes if
1922 var-tracking didn't run.
1923 (rest_of_clean_state): Skip begin stmt markers.
1924 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
1926 * function.c (allocate_struct_function): Set begin_stmt_markers.
1927 * function.h (struct function): Add debug_marker_count counter
1928 and debug_nonbind_markers flag.
1929 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
1930 * gimple-low.c (lower_function_body): Adjust
1931 debug_nonbind_markers.
1932 (lower_stmt): Drop or skip gimple debug stmts.
1933 (lower_try_catch): Skip debug stmts.
1934 * gimple.c (gimple_build_debug_begin_stmt): New.
1935 (gimple_copy): Increment debug_marker_count if copying one.
1936 * gimple.h (gimple_build_debug_begin_stmt): Declare.
1937 * gimplify.c (rexpr_location): New.
1938 (rexpr_has_location): New.
1939 (warn_switch_unreachable_r): Handle gimple debug stmts.
1940 (shortcut_cond_r): Call expr_location.
1942 (find_goto_label): New.
1943 (shortcut_cond_expr): Call expr_has_location, expr_location, and
1945 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
1947 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
1948 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
1949 (LANG_HOOKS_INITIALIZER): ... this.
1950 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
1951 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
1953 (update_ebb_live_info): Skip debug insn markers.
1954 * lra.c (debug_insn_static_data): Rename to...
1955 (debug_bind_static_data): ... this.
1956 (debug_marker_static_data): New.
1957 (lra_set_insn_recog_data): Select one of the above depending
1959 (lra_update_isn_regno_info): Don't assume debug insns have
1961 (push_insns): Skip debug insns.
1962 * lto-streamer-in.c (input_function): Drop debug stmts
1963 depending on active options. Adjust debug_nonbind_markers.
1964 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
1965 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
1966 begin stmt marker notes.
1967 (print_insn): Likewise.
1968 * recog.c (extract_insn): Recognize rtl for debug markers.
1969 * rtl.def (DEBUG_MARKER): New.
1970 * tree-inline.c: Include params.h.
1971 (remap_gimple_stmt): Handle nonbind markers.
1972 (maybe_move_debug_stmts_to_successors): Likewise.
1973 (copy_debug_stmt): Likewise.
1974 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
1975 markers regardless of no side effects.
1976 (tsi_link_before): Don't update container's side effects when adding
1977 a begin stmt marker.
1978 (tsi_link_after): Likewise.
1979 (expr_first): Skip begin stmt markers.
1980 (expr_last): Likewise.
1981 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
1982 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
1983 Disregard nonbind markers.
1984 * tree.c (make_node_stat): Don't set side effects for begin stmt
1986 (build1_stat): Likewise.
1987 * tree.def (DEBUG_BEGIN_STMT): New.
1988 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
1989 * var-tracking.c (delete_debug_insns): Renamed to...
1990 (delete_vta_debug_insns): ... this.
1991 (reemit_marker_as_note): New.
1992 (vt_initialize): Reemit markers.
1993 (delete_vta_debug_insns): Likewise.
1994 (vt_debug_insns_local): Reemit or delete markers.
1995 (variable_tracking_main_1): Likewise.
1996 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
1997 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
1998 (gimple_debug_nonbind_marker_p): New.
1999 (gimple_build_debug_bind): Adjust.
2000 (gimple_build_debug_begin_stmt): New.
2001 * doc/invoke.texi (max-debug-marker-count): New param.
2002 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
2003 (debug_parameter_ref, debug_marker): New.
2004 (NOTE_INSN_BEGIN_STMT): New.
2005 (DEBUG_INSN): Describe begin stmt markers.
2007 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
2008 without debug insns, we wouldn't, but clean up debug insns
2009 after a control flow insn nevertheless.
2011 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
2012 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
2013 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
2014 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
2015 (rtl_tidy_fallthru_edge): Likewise.
2016 (rtl_verify_fallthru): Likewise.
2017 (rtl_verify_bb_layout): Likewise.
2018 (skip_insns_after_block): Likewise.
2019 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
2020 * dwarf2out.c: Include print-rtl.h.
2021 (dwarf2out_next_real_insn): New.
2022 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
2023 Dump debug binds in asm comments.
2024 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
2025 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
2026 callers to use gsi_start_nondebug_bb instead.
2027 (gsi_after_labels): Skip gimple debug stmts.
2028 (gsi_start_nondebug): New.
2029 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
2030 (proper_loop_form_for_interchange): Adjust.
2031 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
2032 * gimple.h (gimple_seq_last_nondebug_stmt): New.
2033 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
2034 (collect_fallthrough_labels): Likewise.
2035 (should_warn_for_implicit_fallthrough): Likewise.
2036 (warn_implicit_fallthrough_r): Likewise.
2037 (expand_FALLTHROUGH_r): Likewise.
2038 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
2039 (graphite_copy_stmts_from_block): Skip nonbind markers.
2040 * haifa-sched.c (sched_extend_bb): Skip debug insns.
2041 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
2042 * jump.c (clean_barriers): Skip debug insns.
2043 * omp-expand.c (expand_parallel_call): Skip debug insns.
2044 (expand_task_call): Likewise.
2045 (remove_exit_barrier): Likewise.
2046 (expand_omp_taskreg): Likewise.
2047 (expand_omp_for_init_counts): Likewise.
2048 (expand_omp_for_generic): Likewise.
2049 (expand_omp_for_static_nochunk): Likewise.
2050 (expand_omp_for_static_chunk): Likewise.
2051 (expand_omp_simd): Likewise.
2052 (expand_omp_taskloop_for_outer): Likewise.
2053 (expand_omp_taskloop_for_inner): Likewise.
2054 (expand_oacc_for): Likewise.
2055 (expand_omp_sections): Likewise.
2056 (expand_omp_single): Likewise.
2057 (expand_omp_synch): Likewise.
2058 (expand_omp_atomic_load): Likewise.
2059 (expand_omp_atomic_store): Likewise.
2060 (expand_omp_atomic_fetch_op): Likewise.
2061 (expand_omp_atomic_pipeline): Likewise.
2062 (expand_omp_atomic_mutex): Likewise.
2063 (expand_omp_target): Likewise.
2064 (grid_expand_omp_for_loop): Likewise.
2065 (grid_expand_target_grid_body): Likewise.
2066 (build_omp_regions_1): Likewise.
2067 * omp-low.c (check_combined_parallel): Skip debug stmts.
2068 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
2069 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
2070 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
2072 * tree-cfg.c (make_blobs_1): Skip debug stmts.
2073 (make_edges): Likewise.
2074 (cleanup_dead_labels): Likewise.
2075 (gimple_can_merge_blocks_p): Likewise.
2076 (stmt_starts_bb_p): Likewise.
2077 (gimple_block_label): Likewise.
2078 (gimple_redirect_edge_and_branch): Likewise.
2079 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
2081 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
2083 * tree-pretty-print (print_declaration): Omit initializer in slim
2085 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
2087 (eliminate_unnecessary_stmts): Stabilize block removal order.
2088 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
2089 * var-tracking.c (get_first_insn): New.
2090 (vt_emit_notes): Call it.
2091 (vt_initialize): Walk any insns before the first BB.
2092 (delete_debug_insns): Likewise.
2094 * gimple.h (enum gimple_debug_subcode): Add
2095 GIMPLE_DEBUG_BEGIN_STMT.
2096 (gimple_debug_begin_stmt_p): New.
2097 (gimple_debug_nonbind_marker_p): New.
2098 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
2099 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
2100 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
2101 * insn-notes.def (BEGIN_STMT): New.
2102 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
2103 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
2104 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
2105 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
2106 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
2107 (INSN_DEBUG_MARKER_KIND): New.
2108 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
2109 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
2110 (INSN_VAR_LOCATION_PTR): New.
2111 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
2113 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
2114 deps only if debug bind insns are enabled.
2115 (pass_expand::execute): Avoid deep TER and expand
2116 debug locations for debug bind insns only.
2117 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
2118 debug stmts special handling down to debug bind stmts.
2119 * combine.c (try_combine): Narrow debug insns special handling
2120 down to debug bind insns.
2121 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
2122 Narrow debug insns preexisting special handling down to debug
2124 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
2125 handling down to debug bind insns.
2126 * function.c (instantiate_virtual_regs): Skip debug markers,
2127 adjust handling of debug binds.
2128 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
2129 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
2130 * haifa-sched.c (schedule_insn): Narrow special handling of debug
2131 insns to debug bind insns.
2132 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
2133 special handling of debug stmts to debug bind stmts.
2134 * ipa-split.c (split_function): Likewise.
2135 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
2136 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
2138 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
2139 * regrename.c (build_def_use): Likewise.
2140 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
2141 (pass_cprop_hardreg): Narrow special casing of debug insns to
2143 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
2144 * reload1.c (reload): Likewise.
2145 * sese.c (sese_insert_phis_for_liveouts): Narrow special
2146 casing of debug stmts to debug bind stmts.
2147 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
2148 * ssa-iterators.h (num_imm_uses): Likewise.
2149 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
2150 debug stmts to debug bind stmts.
2151 * tree-inline.c (tree_function_versioning): Narrow special casing
2152 of debug stmts to debug bind stmts.
2153 * tree-loop-distribution.c (generate_loops_for_partition):
2154 Narrow special casing of debug stmts to debug bind stmts.
2155 * tree-sra.c (analyze_access_subtree): Narrow special casing
2156 of debug stmts to debug bind stmts.
2157 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
2158 stmts to debug bind stmts.
2159 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
2160 casing of debug stmts to debug bind stmts.
2161 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
2162 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
2163 casing of debug stmts to debug bind stmts.
2164 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
2166 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
2167 debug stmts to debug bind stmts.
2168 (gimple_replace_ssa_lhs): Likewise.
2169 (insert_debug_temp_for_var_def): Likewise.
2170 (insert_debug_temps_for_defs): Likewise.
2171 (reset_debug_uses): Likewise.
2172 * tree-ssanames.c (release_ssa_name_fn): Likewise.
2173 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
2174 (adjust_debug_stmts): Likewise.
2175 (adjust_phi_and_debug_stmts): Likewise.
2176 (vect_do_peeling): Likewise.
2177 * tree-vect-loop.c (vect_transform_loop): Likewise.
2178 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
2179 * var-tracking.c (adjust_mems): Narrow special casing of debug
2180 insns to debug bind insns.
2181 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
2182 (compute_bb_dataflow, vt_find_locations): Likewise.
2183 (vt_expand_loc, emit_notes_for_changes): Likewise.
2184 (vt_init_cfa_base): Likewise.
2185 (vt_emit_notes): Likewise.
2186 (vt_initialize): Likewise.
2187 (vt_finalize): Likewise.
2189 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
2190 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
2191 (next_nonnote_nondebug_insn_bb): New.
2192 (prev_nonnote_nondebug_insn_bb): New.
2193 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
2194 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
2195 (prev_nonnote_nondebug_insn_bb): Declare.
2196 (next_nonnote_nondebug_insn_bb): Declare.
2197 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
2198 * cfgrtl.c (get_last_bb_insn): Likewise.
2199 * lra.c (push_insns): Likewise.
2201 2017-12-11 David Malcolm <dmalcolm@redhat.com>
2204 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
2206 * selftest.c (selftest::test_runner::test_runner): New ctor.
2207 (selftest::test_runner::~test_runner): New dtor.
2208 * selftest.h (class selftest::test_runner): New class.
2210 2017-12-11 Carl Love <cel@us.ibm.com>
2212 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
2213 vec_extract_fp32_from_shortl]): Add #defines.
2214 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
2215 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
2216 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
2217 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
2218 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
2219 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
2220 * doc/extend.texi: Add documentation for the added builtins.
2222 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
2224 PR rtl-optimization/80693
2225 PR rtl-optimization/81019
2226 PR rtl-optimization/81020
2227 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
2228 are not mentioned in i3. Place the REG_UNUSED note on i2,
2229 possibly modified to REG_DEAD, if it did not originate in i3.
2231 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2233 * recog.c (store_data_bypass_p_1): New function.
2234 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
2235 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
2237 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
2239 PR rtl-optimization/83361
2240 * ifcvt.c (if_convert): Call fixup_partitions.
2242 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
2244 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2245 early folding of splat_u{8,16,32}.
2247 2017-12-11 Jakub Jelinek <jakub@redhat.com>
2249 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
2250 output_operand_lossage first argument with capital letter.
2251 (aarch64_override_options): Don't start error and sorry first argument
2252 with capital letter.
2254 2017-12-11 Andi Kleen <ak@linux.intel.com>
2256 PR gcov-profile/83355
2257 * auto-profile.c (string_table::get_index_by_decl): Don't
2258 recurse when abstract origin points to itself.
2260 2017-12-11 Bin Cheng <bin.cheng@arm.com>
2262 PR tree-optimization/83320
2263 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
2264 (prune_datarefs_not_in_loop): Ditto.
2266 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
2268 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
2270 2017-12-10 Jakub Jelinek <jakub@redhat.com>
2272 PR tree-optimization/83337
2273 * gimple-loop-interchange.cc (compute_access_stride): Handle
2274 bitfield DRs properly.
2276 2017-12-09 Jakub Jelinek <jakub@redhat.com>
2278 PR tree-optimization/83338
2279 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
2280 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
2281 vector of unsigned integers to vector of signed integers.
2283 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
2285 PR rtl-optimization/83317
2286 * lra-constraints.c (process_address_1): Add insn code check.
2288 2017-12-08 Michael Matz <matz@suse.de>
2290 Fix PR tree-optimization/83323
2291 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
2292 head-controlled loops and loop BBs.
2293 * common.opt (funroll-and-jam): Remove, instead ...
2294 (floop-unroll-and-jam): ... reuse this option.
2295 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
2296 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
2297 (-floop-unroll-and-jam): ... this option.
2299 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2301 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
2302 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
2303 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
2305 2017-12-08 Richard Biener <rguenther@suse.de>
2308 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
2309 handle accesses outside of zero-sized vars.
2311 2017-12-08 Martin Jambor <mjambor@suse.cz>
2313 PR tree-optimization/83141
2314 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
2315 test for MEM_REFs implicitely changing types with padding. Remove
2317 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
2319 2017-12-08 Jakub Jelinek <jakub@redhat.com>
2321 * config/arc/arc.c (arc_attribute_table): Add exclusions to
2323 * config/avr/avr.c (avr_attribute_table): Likewise.
2324 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
2325 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
2326 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
2327 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
2328 of struct attribute_spec.
2329 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
2331 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
2334 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
2336 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
2339 * profile-count.c (profile_count::from_gcov_type): Move from
2340 profile-count.h; handle overflow.
2341 * profile-count.h (profile_count::from_gcov_type): Move offline.
2343 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
2345 PR rtl-optimization/83304
2346 * combine.c (move_deaths): If we do not know where a register died,
2349 2017-12-08 Richard Biener <rguenther@suse.de>
2351 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
2352 Provide -fopt-info-loop feedback when we interchange in a nest.
2354 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2356 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
2357 for armv6 ARM CPU IDs.
2359 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2361 * common/config/arm/arm-common.c: Include <algorithm>.
2362 (INCLUDE_VECTOR): Define.
2363 (compare_opt_names): New function.
2364 (arm_rewrite_selected_arch): Only strip out extensions that can be
2365 expressed through -mfpu. Sort the remaining extensions
2368 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
2370 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
2371 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
2372 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
2373 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
2375 2017-06-08 Tristan Gingold <gindold@adacore.com>
2378 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
2379 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
2381 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2383 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
2384 of struct attribute_spec.
2386 2017-12-08 Julia Koval <julia.koval@intel.com>
2388 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
2389 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
2390 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
2391 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
2392 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
2393 _mm_maskz_dpwssds_epi32): Ditto.
2395 2017-12-08 Richard Biener <rguenther@suse.de>
2397 PR tree-optimization/81303
2398 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
2399 conditions try to create a comparison vector type matching
2400 the data vector type.
2401 (vectorizable_condition): Adjust.
2402 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
2403 Leave invariant conditions alone in case we can vectorize those.
2405 2017-12-08 Julia Koval <julia.koval@intel.com>
2407 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
2408 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
2409 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
2410 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
2411 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
2413 2017-12-08 Julia Koval <julia.koval@intel.com>
2415 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
2416 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
2417 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
2418 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
2419 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
2420 _mm_maskz_dpbusds_epi32): New intrinsics.
2422 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2424 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
2426 (ld<bh>io_signed): Likewise.
2427 (st<bhw>io): Likewise.
2428 * config/nios2/predicates.md (ldstio_memory_operand): Allow
2429 SMALL_INT12 constant integer operand.
2431 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
2433 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
2435 (nios2_large_constant_allowed): ...this. Adjust uses.
2436 (nios2_plus_symbolic_constant_p): Rename to...
2437 (nios2_plus_large_constant_p): ...this. Adjust uses.
2438 (nios2_legitimate_address_p): Correct CONST_INT handling.
2439 (nios2_symbolic_memory_operand_p): Rename to...
2440 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
2441 (nios2_large_constant_p): Check for large constant integers too.
2442 (nios2_split_large_constant): Handle constant integers.
2443 (nios2_split_symbolic_memory_operand): Rename to...
2444 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
2445 (nios2_legitimize_constant_address): Handle constant integers.
2446 (r0rel_constant_p): Handle small constant integers.
2447 (nios2_print_operand_address): Handle r0-relative integer addresses.
2448 * config/nios2/nios2-protos.h: Adjust for renamed functions.
2449 * config/nios2/nios2.md: Adjust for renamed functions.
2451 2017-12-07 Andrew Waterman <andrew@sifive.com>
2453 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
2454 (TARGET_HAVE_SRODATA_SECTION): New define.
2455 (riscv_select_section): New function.
2457 2017-12-08 Joseph Myers <joseph@codesourcery.com>
2458 Alexander Monakov <amonakov@ispras.ru>
2459 Jakub Jelinek <jakub@redhat.com>
2462 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
2464 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2466 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
2467 a bool success value. Don't call output_operand_lossage here.
2468 (aarch64_print_ldpstp_address): Return a bool success value.
2469 (aarch64_print_operand_address): Call output_addr_const if
2470 aarch64_print_address_internal fails.
2471 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
2472 'y'; call output_operand_lossage instead. Call output_operand_lossage
2473 if aarch64_print_ldpstp_address fails.
2475 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2477 * tree-vector-builder.h
2478 (tree_vector_builder::binary_encoded_nelts): Declare.
2479 * tree-vector-builder.c
2480 (tree_vector_builder::binary_encoded_nelts): New function.
2481 * fold-const.c (negate_expr_p): Likewise.
2482 (operand_equal_p, fold_checksum_tree): Likewise.
2483 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
2484 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
2485 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
2486 (uniform_vector_p): Likewise.
2487 * varasm.c (const_hash_1, compare_constant): Likewise.
2488 * tree-ssa-ccp.c: Include tree-vector-builder.h.
2489 (valid_lattice_transition): Operate directly on the VECTOR_CST
2491 * ipa-icf.c: Include tree-vector-builder.h.
2492 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
2493 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
2495 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2497 * tree.c (build_vector): Delete.
2498 * tree.h (build_vector): Make static and move into the self-testing
2501 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2503 * vector-builder.h (vector_builder::derived): New const overload.
2504 (vector_builder::elt): New function.
2505 * tree-vector-builder.h (tree_vector_builder::type): New function.
2506 (tree_vector_builder::apply_step): Declare.
2507 * tree-vector-builder.c (tree_vector_builder::apply_step): New
2509 * gimple-fold.h (tree_vector_builder): Declare.
2510 (gimple_build_vector): Take a tree_vector_builder instead of a
2511 type and vector of elements.
2512 * gimple-fold.c (gimple_build_vector): Likewise.
2513 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
2515 (get_initial_defs_for_reduction): Likewise.
2516 (vectorizable_induction): Likewise.
2518 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2520 * tree-vector-builder.h
2521 (tree_vector_builder::new_binary_operation): Declare.
2522 * tree-vector-builder.c
2523 (tree_vector_builder::new_binary_operation): New function.
2524 * fold-const.c (fold_relational_const): Use it.
2525 (const_binop): Likewise. Check that both input vectors have
2526 the same number of elements, thus excluding things like WIDEN_SUM.
2527 Check whether it is possible to operate directly on the encodings
2530 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2532 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
2533 new_unary_operation, operating only on the encoded elements.
2534 (const_unop): Likewise.
2535 (exact_inverse): Likewise.
2536 (distributes_over_addition_p): New function.
2537 (const_binop): Use tree_vector_builder and new_unary_operation
2538 for combinations of VECTOR_CST and INTEGER_CST. Operate only
2539 on the encoded elements unless the encoding is strided and the
2540 operation does not distribute over addition.
2541 (fold_convert_const): Use tree_vector_builder and
2542 new_unary_operation. Operate only on the encoded elements
2543 for truncating integer conversions, or for non-stepped encodings.
2545 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2547 * config/sparc/sparc.c: Include tree-vector-builder.h.
2548 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
2549 * expmed.c: Include tree-vector-builder.h.
2550 (make_tree): Use tree_vector_builder instead of build_vector.
2551 * fold-const.c: Include tree-vector-builder.h.
2552 (const_binop): Use tree_vector_builder instead of build_vector.
2553 (const_unop): Likewise.
2554 (native_interpret_vector): Likewise.
2555 (fold_vec_perm): Likewise.
2556 (fold_ternary_loc): Likewise.
2557 * gimple-fold.c: Include tree-vector-builder.h.
2558 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
2560 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
2561 (simplify_vector_constructor): Use tree_vector_builder instead
2563 * tree-vect-generic.c: Include tree-vector-builder.h.
2564 (add_rshift): Use tree_vector_builder instead of build_vector.
2565 (expand_vector_divmod): Likewise.
2566 (optimize_vector_constructor): Likewise.
2567 * tree-vect-loop.c: Include tree-vector-builder.h.
2568 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
2569 of build_vector. Explicitly use a stepped encoding for
2571 * tree-vect-slp.c: Include tree-vector-builder.h.
2572 (vect_get_constant_vectors): Use tree_vector_builder instead
2574 (vect_transform_slp_perm_load): Likewise.
2575 (vect_schedule_slp_instance): Likewise.
2576 * tree-vect-stmts.c: Include tree-vector-builder.h.
2577 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
2578 (vect_gen_perm_mask_any): Likewise.
2579 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
2580 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
2582 (build_vector_from_val): Likewise. Explicitly use a duplicate
2585 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
2587 * doc/generic.texi (VECTOR_CST): Describe new representation of
2589 * vector-builder.h: New file.
2590 * tree-vector-builder.h: Likewise.
2591 * tree-vector-builder.c: Likewise.
2592 * Makefile.in (OBJS): Add tree-vector-builder.o.
2593 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
2594 * tree-core.h (tree_base): Add a vector_cst field to the u union.
2595 (tree_vector): Change the number of elements to
2596 vector_cst_encoded_nelts.
2597 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
2598 (VECTOR_CST_ELTS): Delete.
2599 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
2600 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
2601 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
2602 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
2603 (VECTOR_CST_ENCODED_ELT): Likewise.
2604 (vector_cst_encoded_nelts): New function.
2605 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2606 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2607 (vector_cst_int_elt, vector_cst_elt): Declare.
2608 * tree.c: Include tree-vector-builder.h.
2609 (tree_code_size): Abort if passed VECTOR_CST.
2610 (tree_size): Update for new VECTOR_CST layout.
2611 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
2612 VECTOR_CST_NELTS_PER_PATTERN as arguments.
2613 (build_vector): Use tree_vector_builder.
2614 (vector_cst_int_elt, vector_cst_elt): New functions.
2615 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
2616 encoded elements and then create the vector in the canonical form.
2617 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
2618 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
2619 (tree_c_tests): Call test_vector_cst_patterns.
2620 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
2622 (hash_tree): Likewise.
2623 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
2624 (streamer_write_tree_header): Likewise.
2625 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
2626 (streamer_alloc_tree): Likewise. Update call to make_vector.
2627 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
2629 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
2631 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
2632 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
2633 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
2635 2017-12-07 Bin Cheng <bin.cheng@arm.com>
2636 Richard Biener <rguenther@suse.de>
2638 PR tree-optimization/81303
2639 * Makefile.in (gimple-loop-interchange.o): New object file.
2640 * common.opt (floop-interchange): Reuse the option from graphite.
2641 * doc/invoke.texi (-floop-interchange): Ditto. New document for
2642 -floop-interchange and mention it for -O3.
2643 * opts.c (default_options_table): Enable -floop-interchange at -O3.
2644 * gimple-loop-interchange.cc: New file.
2645 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
2646 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
2647 * passes.def (pass_linterchange): New pass.
2648 * timevar.def (TV_LINTERCHANGE): New time var.
2649 * tree-pass.h (make_pass_linterchange): New declaration.
2650 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
2651 interchange. Record IV before/after increment in new parameters.
2652 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
2653 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
2655 (check_reduction_path): ...New function here.
2656 * tree-vectorizer.h (check_reduction_path): New declaration.
2658 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
2661 PR rtl-optimization/80818
2662 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
2663 always early clobbered.
2664 * lra-lives.c (process_bb_lives): Check input hard regs for early
2665 clobbered non-operand hard reg.
2667 2017-12-07 Jakub Jelinek <jakub@redhat.com>
2670 * tree-cfg.c (verify_gimple_assign_binary): Don't require
2671 types_compatible_p, just that TYPE_MODE is the same.
2673 2017-12-07 Martin Sebor <msebor@redhat.com>
2676 * attribs.c (empty_attribute_table): Initialize new member of
2677 struct attribute_spec.
2678 (decl_attributes): Add argument. Handle mutually exclusive
2679 combinations of attributes.
2680 (selftests::test_attribute_exclusions): New function.
2681 (selftests::attribute_c_tests): Ditto.
2682 * attribs.h (decl_attributes): Add default argument.
2683 * selftest.h (attribute_c_tests): Declare.
2684 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
2685 * tree-core.h (attribute_spec::exclusions, exclude): New type and
2687 * doc/extend.texi (Common Function Attributes): Update const and pure.
2688 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
2689 of struct attribute_spec.
2690 * config/arc/arc.c (arc_attribute_table): Same.
2691 * config/arm/arm.c (arm_attribute_table): Same.
2692 * config/avr/avr.c ( avr_attribute_table): Same.
2693 * config/bfin/bfin.c (bfin_attribute_table): Same.
2694 * config/cr16/cr16.c (cr16_attribute_table): Same.
2695 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
2696 * config/h8300/h8300.c (h8300_attribute_table): Same.
2697 * config/i386/i386.c (ix86_attribute_table): Same.
2698 * config/ia64/ia64.c (ia64_attribute_table): Same.
2699 * config/m32c/m32c.c (m32c_attribute_table): Same.
2700 * config/m32r/m32r.c (m32r_attribute_table): Same.
2701 * config/m68k/m68k.c (m68k_attribute_table): Same.
2702 * config/mcore/mcore.c (mcore_attribute_table): Same.
2703 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
2704 * config/mips/mips.c (mips_attribute_table): Same.
2705 * config/msp430/msp430.c (msp430_attribute_table): Same.
2706 * config/nds32/nds32.c (nds32_attribute_table): Same.
2707 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
2708 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
2709 * config/rl78/rl78.c (rl78__attribute_table): Same.
2710 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
2711 * onfig/rx/rx.c (rx_attribute_table): Same.
2712 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
2713 * config/sh/sh.c (sh_attribute_table): Same.
2714 * config/sparc/sparc.c (sparc_attribute_table): Same.
2715 * config/spu/spu.c (spu_attribute_table): Same.
2716 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
2717 * config/v850/v850.c (v850_attribute_table): Same.
2718 * config/visium/visium.c (visium_attribute_table): Same.
2720 2017-12-07 Tamar Christina <tamar.christina@arm.com>
2723 * config/arm/arm.c (INCLUDE_STRING): Define.
2724 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
2725 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
2728 2017-12-07 Michael Matz <matz@suse.de>
2730 Add unroll and jam pass
2732 * gimple-loop-jam.c: New file.
2733 * Makefile.in (OBJS): Add gimple-loop-jam.o.
2734 * common.opt (funroll-and-jam): New option.
2735 * opts.c (default_options_table): Add unroll-and-jam at -O3.
2736 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
2737 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
2738 * passes.def: Add pass_loop_jam.
2739 * timevar.def (TV_LOOP_JAM): Add.
2740 * tree-pass.h (make_pass_loop_jam): Declare.
2741 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
2742 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
2743 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
2744 to flow_loop_tree_node_add.
2745 (duplicate_subloops, copy_loops_to): Append to sibling list.
2746 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
2747 * doc/invoke.texi (-funroll-and-jam): Document new option.
2748 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
2750 2017-12-07 Richard Biener <rguenther@suse.de>
2752 PR tree-optimization/83296
2753 PR tree-optimization/67769
2754 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
2755 flow sensitive info in an unrelated BB.
2756 (value_replacement): Use reset_flow_sensitive_info.
2757 (minmax_replacement): Reset flow sensitive info on the def
2758 we move. Do not reset flow sensitive info in the whole BB
2759 we move the stmt to.
2760 (abs_replacement): Likewise.
2762 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2765 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2766 rs6000_cpu to the given -mcpu=, or to the default processor.
2768 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2770 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
2771 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
2772 instead of rs6000_cpu_attr.
2773 (rs6000_adjust_cost): Ditto.
2774 (is_microcoded_insn): Ditto.
2775 (rs6000_adjust_priority): Ditto.
2776 (rs6000_issue_rate): Ditto.
2777 (rs6000_use_sched_lookahead): Ditto.
2778 (rs6000_use_sched_lookahead_guard): Ditto.
2779 (rs6000_sched_reorder): Ditto.
2780 (force_new_group): Ditto.
2781 * config/rs6000/rs6000.md (cpu attribute): Ditto.
2782 (group_ending_nop): Ditto.
2784 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2786 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
2787 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
2788 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
2789 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
2791 (rs6000_reassociation_width): Ditto.
2792 (rs6000_emit_epilogue): Ditto.
2793 (rs6000_adjust_cost): Ditto.
2794 (is_microcoded_insn): Ditto.
2795 (is_cracked_insn): Ditto.
2796 (rs6000_adjust_priority): Ditto.
2797 (rs6000_sched_reorder): Ditto.
2798 (rs6000_sched_reorder2): Ditto.
2799 (insn_must_be_first_in_group): Ditto.
2800 (insn_must_be_last_in_group): Ditto.
2801 (rs6000_register_move_cost): Ditto.
2802 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
2805 2017-12-07 Julia Koval <julia.koval@intel.com>
2807 * config.gcc: Add vaesintrin.h.
2808 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2810 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2811 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2813 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2814 * config/i386/immintrin.h: Include vaesintrin.h.
2815 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2816 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2817 _mm_aesdec_epi128): New intrinsics.
2819 2017-12-06 David Malcolm <dmalcolm@redhat.com>
2821 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
2822 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
2823 c-family/c-spellcheck.cc.
2824 (best_macro_match::best_macro_match): Likewise.
2826 (struct edit_distance_traits<cpp_hashnode *>): Move to
2827 c-family/c-spellcheck.h.
2828 (class best_macro_match): Likewise.
2830 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2832 PR tree-optimization/83293
2833 * gimple-ssa-strength-reduction.c (insert_initializers): Use
2834 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
2835 might insert into empty bb.
2838 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
2839 simplify for plus with :c added, and pointer_plus without that.
2840 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
2841 with undefined overflow and the conversion is not widening,
2842 perform negation in utype and only convert to type afterwards.
2843 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
2844 simplify for plus with :c added, and pointer_plus without that.
2845 If type is integral with undefined overflow and the conversion is
2846 not widening, perform minus in utype and only convert to type
2847 afterwards. Move the last pointer_diff_expr simplify into the
2850 2017-12-06 Martin Sebor <msebor@redhat.com>
2852 PR tree-optimization/82646
2853 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
2854 strncpy, not maxlen.
2856 2017-12-06 Martin Sebor <msebor@redhat.com>
2858 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
2861 PR tree-optimization/83075
2862 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
2863 strncat/strncpy don't change length of source string.
2865 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
2868 2017-11-29 Martin Aberg <maberg@gaisler.com>
2870 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
2871 to prevent b2bst errata sequence.
2872 (sqrtdf2_fix): Likewise.
2874 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2876 PR tree-optimization/81945
2877 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
2878 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
2879 to dest_cfun has orig_loop_num set, either remap it to the new
2880 loop number if the loop got moved too, or clear it.
2882 2017-12-05 Steve Ellcey <sellcey@cavium.com>
2884 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
2886 (thunderx2t99_nothing): New insn reservation.
2887 (thunderx2t99_mrs): New insn reservation.
2888 (thunderx2t99_multiple): New insn reservation.
2889 (thunderx2t99_alu_basi): Add bfx to reservation.
2890 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
2892 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2895 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
2897 2017-12-05 Bin Cheng <bin.cheng@arm.com>
2899 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
2900 tree-ssa-pre.c::remove_dead_inserted_code.
2901 * tree-ssa-dce.h: New file.
2902 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
2903 (remove_dead_inserted_code): Move and rename to function
2904 tree-ssa-dce.c::simple_dce_from_worklist.
2905 (pass_pre::execute): Update use.
2907 2017-12-05 Richard Biener <rguenther@suse.de>
2909 PR tree-optimization/83277
2910 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
2911 to code-gen liveout vars.
2913 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
2915 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
2916 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
2917 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
2920 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
2922 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
2924 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
2925 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
2928 2017-12-05 Richard Biener <rguenther@suse.de>
2930 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
2931 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
2932 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
2933 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
2935 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2937 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
2939 2017-12-05 Martin Liska <mliska@suse.cz>
2940 Jakub Jelinek <jakub@redhat.com>
2942 * doc/invoke.texi: Document the options.
2943 * flag-types.h (enum sanitize_code): Add
2944 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
2945 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
2946 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
2947 * opts.c: Define new sanitizer options.
2948 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
2949 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
2951 2017-12-05 Julia Koval <julia.koval@intel.com>
2953 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
2954 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
2955 (ix86_handle_option): Handle -mavx512vnni.
2956 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
2957 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
2958 * config/i386/i386-c (__AVX512VNNI__): New.
2959 * config/i386/i386.c (ix86_target_string): Handle new option.
2960 (ix86_valid_target_attribute_inner_p): Handle new option.
2961 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
2962 * config/i386/i386.opt (mavx512vnni): New option.
2964 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2967 * athlon.md: Disable for generic.
2968 * haswell.md: Enable for generic.
2969 * i386.c (ix86_sched_init_global): Add core hooks for generic.
2970 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
2972 (ix86_adjust_cost): Move generic to haswell path.
2974 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
2976 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
2977 instead of MEM_P in a couple more places. Fix formatting issues.
2979 2017-12-04 Jim Wilson <jimw@sifive.com>
2981 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
2982 instead of GP_REG_LAST-1.
2983 (riscv_adjust_libcall_cfi_prologue): Likewise.
2984 (riscv_adjust_libcall_cri_epilogue): Likewise.
2985 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
2988 2017-12-04 Luis Machado <luis.machado@linaro.org>
2990 * ipa-pure-const.c (check_decl): Add missing newline.
2991 (state_from_flags): Likewise.
2993 2017-12-04 Jeff Law <law@redhat.com>
2995 PR tree-optimizatin/78496
2996 * gimple-ssa-evrp-analyze.h
2997 (evrp_range_analyzer::get_vr_values): Simplify.
2998 * gimple-ssa-evrp-analyze.c: Corresponding changes.
2999 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
3000 and gimple-ssa-evrp-analyze.h.
3001 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
3002 (simplify_stmt_for_jump_threading): Copy a blob of code from
3003 tree-vrp.c to use ranges to simplify statements.
3004 (dom_opt_dom_walker::before_dom_children): Call
3005 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
3006 (dom_opt_dom_walker::after_dom_children): Similarly for
3007 evrp_range_analyzer::leave.
3008 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
3011 * gimple-ssa-evrp-analyze.c
3012 (evrp_range_analyzer::extract_range_from_stmt): Always use
3013 vr_values::update_value_range so preexisting range info is
3014 medged with new range info, even if the new range is VR_VARYING.
3016 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3018 * combine.c: Adjust comment.
3019 (use_crosses_set_p): Delete.
3020 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
3021 (try_combine): Ditto.
3023 2017-12-04 Richard Biener <rguenther@suse.de>
3025 PR tree-optimization/83255
3026 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
3027 Re-add zero-iteration check.
3029 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3031 PR rtl-optimization/83245
3032 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
3033 hard registers as earlyclobber, also if not in an asm.
3035 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
3039 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3042 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3043 rs6000_cpu based on cpu_index, not tune_index.
3045 2017-12-04 Richard Biener <rguenther@suse.de>
3047 PR tree-optimization/83238
3048 * graphite-scop-detection.c (scop_detection::merge_sese): Make
3049 code match comment, rejecting invalid SESE regions.
3051 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
3053 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
3054 require base operand is a REG_POINTER prior to reload on targets
3055 with non-equivalent space registers.
3057 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3059 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
3060 (update_specialized_profile): Fix updating of counts.
3061 (perhaps_add_new_callers): Likewise.
3063 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
3066 * x86-tune.def: Remove obsolette FIXMEs.
3067 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
3068 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
3069 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
3071 (X86_TUNE_PAD_RETURNS): Disable for generic.
3073 2017-12-02 Jakub Jelinek <jakub@redhat.com>
3075 PR tree-optimization/83170
3076 PR tree-optimization/83241
3077 * gimple-ssa-store-merging.c
3078 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
3079 gimple_vuse (ins_stmt) in case it has changed.
3080 (imm_store_chain_info::output_merged_store): Likewise.
3082 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
3086 * tree-cfg.c (pass_warn_function_return::execute): Handle
3087 __builtin_ubsan_handle_missing_return like __builtin_unreachable
3088 with BUILTINS_LOCATION.
3092 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
3093 is BLKmode for vector field with vector raw mode, use TYPE_MODE
3094 instead of DECL_MODE.
3096 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
3097 last argument to rtx pointer.
3098 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
3099 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
3100 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
3101 depending on the chosen ISAs.
3102 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
3103 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
3104 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
3106 * config/i386/sse.md (mov<mode>_internal): Likewise.
3107 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
3109 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3111 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
3113 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3115 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
3116 parameters from prototype.
3117 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
3118 parameters. Don't print a comment.
3119 (emit_fusion_gpr_load): Adjust.
3120 (emit_fusion_load_store): Adjust.
3121 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
3122 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
3123 comment on the second line.
3125 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3128 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
3129 rs6000_cpu based on cpu_index, not tune_index.
3131 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
3133 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
3134 which_alternative instead of which_alternative + 1.
3135 (output_asm_insn): Print an extra tab if the template is short.
3137 2017-12-01 Jim Wilson <jimw@sifive.com>
3139 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
3141 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
3142 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
3144 * doc/tm.texi: Regenerate.
3146 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3149 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
3150 whether we can allocate pseudos before trying to fix an address.
3151 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
3152 memory address is indexed or indirect.
3153 (floatuns_<mode>si2_hw2): Likewise.
3155 2017-12-01 Jason Merrill <jason@redhat.com>
3157 * Makefile.in (TAGS): Add c-family/*.cc.
3159 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
3161 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
3162 (call_value_insn): Likewise.
3163 (sibcall_insn): Likewise.
3164 (sibcall_value_insn): Likewise.
3165 (movsi_aarch64): Likewise.
3166 (movdi_aarch64): Likewise.
3167 (add_losym_): Likewise.
3168 (ldr_got_small_): Likewise.
3169 (ldr_got_small_sidi): Likewise.
3170 (ldr_got_small_28k_): Likewise.
3171 (ldr_got_small_28k_sidi): Likewise.
3172 * config/aarch64/aarch64.c (aarch64_print_address_internal):
3173 Move output_addr_const to symbolic case. Add error check.
3175 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3177 * config/s390/predicates.md (plus16_Q_operand): New predicate.
3178 * config/s390/s390.md: Disable MVC merging peephole if it would
3179 disable operand forwarding.
3180 (new peephole2): Split MVCs if it would turn them into up to 2
3183 2017-12-01 Richard Biener <rguenther@suse.de>
3185 PR tree-optimization/83232
3186 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
3187 detection of same access. Instead of breaking the group here
3188 do not consider the duplicate. Add comment explaining real fix.
3190 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
3192 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
3194 2017-12-01 Sudakshina Das <sudi.das@arm.com>
3196 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
3198 2017-12-01 Jakub Jelinek <jakub@redhat.com>
3200 * function.h (struct function): Remove cilk_frame_decl,
3201 is_cilk_function and calls_cilk_spawn fields.
3202 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
3204 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
3205 cilk_elemental field.
3206 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
3207 * target.def: Adjust comment.
3208 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
3209 Don't test cilk_elemental.
3211 PR tree-optimization/83233
3212 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
3213 bswap_stat name for the struct.
3216 * tree.h (SWITCH_BREAK_LABEL_P): Define.
3217 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
3218 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
3219 SWITCH_BREAK_LABEL_P set on the label.
3220 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
3221 added for default case if it was missing and not all cases covered.
3222 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
3223 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
3225 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
3226 to avoid -Wimplicit-fallthrough warning.
3227 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
3228 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
3230 PR tree-optimization/83221
3231 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
3233 (init_reassoc): Formatting fix.
3236 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
3237 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
3241 * tree-cfg.c: Include asan.h.
3242 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
3244 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3246 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
3248 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3250 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
3251 vpcmpeqd instruction.
3253 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
3255 * config/i386/i386.c (standard_sse_constant_opcode): Fix
3256 registers type for 128bit mode.
3258 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3260 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
3262 * gengtype-state.c (read_state_pair): Likewise.
3263 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
3264 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
3265 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
3266 * ifcvt.c (noce_try_store_flag_constants): Likewise.
3267 * tree-ssa-ccp.c (ccp_finalize): Likewise.
3268 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
3269 * builtins.c (fold_builtin_3): Likewise.
3270 * graphite-scop-detection.c
3271 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
3272 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
3275 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
3278 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
3279 parameter. Return true if ignore_decl_rtl_set_p is true and other
3280 conditions are satisfied.
3281 * asan.h (asan_protect_global): Add new parameter.
3282 * varasm.c (categorize_decl_for_section): Pass true as second parameter
3283 to asan_protect_global calls.
3285 2017-11-30 Jim Wilson <jimw@sifive.com>
3287 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
3288 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
3289 -mstrict-align, add info on default value. Delete redundant lines for
3290 -mabi. Add missing -mexplicit-relocs docs.
3292 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3294 * config/arc/arc.md (trap): New pattern.
3296 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3298 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
3299 instruction to end into a delay slot.
3300 * config/arc/arc.md (cond_delay_insn): Check if the instruction
3301 can be placed into a delay slot against reg_note.
3303 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3305 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
3306 labels number of usages.
3308 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
3310 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
3312 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
3314 2017-11-30 Jakub Jelinek <jakub@redhat.com>
3317 * internal-fn.c (expand_mul_overflow): Optimize unsigned
3318 multiplication by power of 2 constant into two shifts + comparison.
3320 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
3323 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
3325 2017-11-30 Richard Biener <rguenther@suse.de>
3327 PR tree-optimization/83202
3328 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3329 allow_peel argument and guard peeling.
3330 (canonicalize_loop_induction_variables): Likewise.
3331 (canonicalize_induction_variables): Pass false.
3332 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
3333 peeling from cunrolli.
3335 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3337 * combine.c (try_combine): Print a message to dump file whenever
3338 I0, I1, or I2 cannot be combined into I3.
3340 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3342 PR rtl-optimization/83156
3343 PR rtl-optimization/82621
3344 * combine.c (try_combine): Don't split an I2 if one of the dests is
3345 set again before I3. Allow unused dests.
3347 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
3349 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
3351 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
3353 PR rtl-optimization/80818
3354 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
3355 recursively. Use insn code for clobber.
3356 (lra_set_insn_recog_data): Pass the new arg to
3357 collect_non_operand_hard_regs.
3358 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
3360 (lra_update_insn_regno_info): Pass insn to
3361 add_regs_to_insn_regno_info.
3363 2017-11-29 Jim Wilson <jimw@sifive.com>
3364 Andrew Waterman <andrew@sifive.com>
3366 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
3367 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
3368 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
3369 use SINGLE_SHIFT_COST.
3370 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
3371 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
3373 2017-11-29 Julia Koval <julia.koval@intel.com>
3375 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
3376 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
3377 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
3378 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
3379 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
3380 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
3381 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
3382 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
3383 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
3384 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
3385 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
3386 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
3387 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
3388 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
3389 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
3390 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
3391 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
3392 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
3393 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
3394 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
3395 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
3396 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
3397 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
3398 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
3399 __builtin_ia32_vpshldv_v2di_mask,
3400 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
3401 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
3402 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
3404 2017-11-29 Julia Koval <julia.koval@intel.com>
3406 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
3407 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
3408 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
3409 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
3410 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
3411 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
3412 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
3413 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
3414 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
3415 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
3416 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
3417 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
3418 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
3419 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
3420 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
3421 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
3422 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
3423 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
3424 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3425 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
3426 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
3428 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3430 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
3431 movsi_pic_gotdata_op instruction as a load for the UT699 errata
3434 2017-11-29 Martin Aberg <maberg@gaisler.com>
3436 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
3437 to prevent b2bst errata sequence.
3438 (sqrtdf2_fix): Likewise.
3440 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3442 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
3443 (div_sqrt_insn_p): New function.
3444 (sparc_do_work_around_errata): Insert NOP instructions to
3445 prevent sequences that could trigger the TN-0013 errata for
3446 certain LEON3 processors.
3447 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
3448 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
3449 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
3450 (in_branch_delay): Prevent div and sqrt in delay slot if
3452 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
3454 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3456 * config/sparc/sparc.c (atomic_insn_p): New function.
3457 (sparc_do_work_around_errata): Insert NOP instructions to
3458 prevent sequences that could trigger the TN-0010 errata for
3460 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
3461 instruction referable in atomic_insns_p.
3463 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3465 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
3466 (atomic_compare_and_swap_leon3_1): Likewise.
3469 2017-11-29 Daniel Cederman <cederman@gaisler.com>
3471 * config/sparc/sparc.c (fpop_insn_p): New function.
3472 (sparc_do_work_around_errata): Insert NOP instructions to
3473 prevent sequences that could trigger the TN-0012 errata for
3475 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
3476 * config/sparc/sparc.md (fix_gr712rc): New attribute.
3477 (in_branch_annul_delay): Prevent floating-point instructions
3478 in delay slot of annulled integer branch.
3480 2017-11-29 Richard Biener <rguenther@suse.de>
3482 PR tree-optimization/83202
3483 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
3485 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
3486 nodes vectorized to the same stmts multiple times.
3487 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
3488 (vect_analyze_slp_instance): Adjust.
3489 (scalar_stmts_to_slp_tree_map_t): New typedef.
3490 (vect_schedule_slp_instance): Add a map recording the SLP node
3491 representing the vectorized stmts for a set of scalar stmts.
3492 Avoid code-generating redundancies.
3493 (vect_schedule_slp): Allocate map and pass it down.
3495 2017-11-29 Nathan Sidwell <nathan@acm.org>
3498 * tree.c (build_complex_type): Fix canonicalization. Only fill in
3501 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
3503 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
3504 cases for printing LDP/STP memory addresses.
3505 (aarch64_print_address_internal): Renamed from
3506 aarch64_print_operand_address, added parameter, add Pmode check.
3507 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
3508 (aarch64_print_operand_address): Indirect to
3509 aarch64_print_address_internal.
3510 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
3511 'y' operand output specifier.
3513 2017-11-29 Jakub Jelinek <jakub@redhat.com>
3516 * tree.c (build_simple_mem_ref_loc): Handle
3517 get_addr_base_and_unit_offset returning a MEM_REF.
3520 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
3523 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
3526 2017-11-28 David Malcolm <dmalcolm@redhat.com>
3528 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
3530 (layout::move_to_column): Likewise.
3532 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3535 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
3536 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
3538 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3539 Martin Jambor <mjambor@suse.cz>
3542 * tree.h (expr_type_first_operand_type_p): Declare
3543 * tree.c (expr_type_first_operand_type_p): New function.
3544 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
3545 (ipa_value_from_jfunc): Adjust declaration.
3546 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
3547 Use it as result type for arithmetics, unless it is NULL in which case
3548 be more conservative.
3549 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
3550 ipa_get_jf_pass_through_result.
3551 (propagate_vals_across_pass_through): Likewise.
3552 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
3553 is to propagate_vals_across_pass_through.
3554 (propagate_constants_across_call): Pass PARM_TYPE to
3555 propagate_scalar_across_jump_function.
3556 (find_more_scalar_values_for_callers_subset): Pass parameter type to
3557 ipa_value_from_jfunc.
3558 (cgraph_edge_brings_all_scalars_for_node): Likewise.
3559 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
3560 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
3561 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
3562 target_type, pass it to ipa_value_from_jfunc.
3563 (update_indirect_edges_after_inlining): Pass parameter type to
3564 try_make_edge_direct_simple_call.
3566 2017-11-28 Jeff Law <law@redhat.com>
3568 * gimple-ssa-evrp-analyze.c
3569 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
3570 refine ranges if scev_initialized_p returns true.
3571 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
3573 2017-11-28 Julia Koval <julia.koval@intel.com>
3575 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
3576 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
3577 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
3578 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
3579 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
3580 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
3581 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
3582 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
3583 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
3584 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
3585 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
3586 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
3587 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
3588 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
3589 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
3590 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
3591 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
3592 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
3593 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
3594 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
3595 __builtin_ia32_vpshrd_v2di_mask): New builtins.
3596 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
3598 2017-11-28 Julia Koval <julia.koval@intel.com>
3600 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
3601 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
3602 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
3603 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
3604 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
3605 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
3606 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
3607 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
3608 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
3609 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
3610 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
3611 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
3612 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
3613 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
3614 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
3615 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
3616 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
3617 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
3618 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
3619 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
3620 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
3621 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
3622 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
3623 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
3624 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
3625 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
3626 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
3627 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
3628 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
3629 __builtin_ia32_vpshld_v2di_mask): New builtins.
3630 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3631 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
3633 2017-11-28 Richard Biener <rguenther@suse.de>
3635 PR tree-optimization/80776
3636 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
3638 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
3640 (evrp_range_analyzer::record_ranges_from_incoming_edges):
3641 If the incoming edge is an effective fallthru because the other
3642 edge only reaches a __builtin_unreachable () then record ranges
3643 derived from the controlling condition in SSA info.
3644 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
3645 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3647 2017-11-28 Olivier Hainque <hainque@adacore.com>
3649 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
3650 on mingw build hosts.
3652 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3655 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
3657 2017-11-28 Richard Biener <rguenther@suse.de>
3660 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
3661 copies generated from memcpy use a character array as reference
3664 2017-11-28 Julia Koval <julia.koval@intel.com>
3665 Sebastian Peryt <sebastian.peryt@intel.com>
3667 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
3668 c-family/c-cilkplus.o, c-family/array-notation-common.o,
3669 cilk-common.o, cilk.h, cilk-common.c): Remove.
3671 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
3672 * builtins.c (is_builtin_name): Remove cilkplus condition.
3673 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
3674 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
3675 cilk-builtins.def, cilkplus.def): Remove.
3676 * cif-code.def (CILK_SPAWN): Remove.
3677 * cilk-builtins.def: Delete.
3678 * cilk-common.c: Ditto.
3680 * cilkplus.def: Ditto.
3681 * config/darwin.h (fcilkplus): Delete.
3682 * cppbuiltin.c: Ditto.
3683 * doc/extend.texi: Remove cilkplus doc.
3684 * doc/generic.texi: Ditto.
3685 * doc/invoke.texi: Ditto.
3686 * doc/passes.texi: Ditto.
3687 * gcc.c (fcilkplus): Remove.
3688 * gengtype.c (cilk.h): Remove.
3689 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
3691 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
3693 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
3695 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
3696 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
3697 cilkplus conditions.
3698 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
3699 inline_read_section): Ditto.
3700 * ipa-inline-analysis.c (cilk.h): Remove.
3701 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
3702 * lto-wrapper.c (merge_and_complain, append_compiler_options,
3703 append_linker_options): Remove condition for fcilkplus.
3704 * lto/lto-lang.c (cilk.h): Remove.
3705 (lto_init): Remove condition for fcilkplus.
3706 * omp-expand.c (expand_cilk_for_call): Delete.
3707 (expand_omp_taskreg, expand_omp_for_static_chunk,
3708 expand_omp_for): Remove cilkplus
3710 (expand_cilk_for): Delete.
3711 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
3712 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
3713 execute_lower_omp, diagnose_sb_0): Ditto.
3714 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
3715 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
3716 * tree-nested.c: Ditto.
3717 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
3718 (dump_generic_node): Ditto.
3719 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
3720 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
3722 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
3724 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3726 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
3727 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
3728 for vector mode and !TARGET_SIMD.
3730 2017-11-28 Jakub Jelinek <jakub@redhat.com>
3732 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
3734 * tree.h (SWITCH_LABELS): Remove.
3735 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
3736 assert SWITCH_BODY is non-NULL.
3737 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
3739 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
3741 PR tree-optimization/80788
3742 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
3743 has TREE_OVERFLOW set, call drop_tree_overflow.
3745 2017-11-28 Richard Biener <rguenther@suse.de>
3747 PR tree-optimization/83158
3748 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
3750 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
3753 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
3754 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
3756 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
3758 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
3759 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
3761 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3764 * varasm.c (compare_constant): Take the mode of the constants into
3765 account when comparing floating point constants.
3767 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
3769 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
3770 from explicit instantiation of debug_helper.
3771 * vec.h (DEFINE_DEBUG_VEC): Ditto.
3773 2017-11-27 Richard Biener <rguenther@suse.de>
3775 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
3778 2017-11-27 Richard Biener <rguenther@suse.de>
3780 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
3781 (type_hash_canon): Also clear int_cst_hash_table entry for
3783 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
3785 2017-11-27 Tamar Christina <tamar.christina@arm.com>
3787 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3788 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3789 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3791 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3793 * hash-map.h (gt_cleare_cache): Avoid UB.
3795 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
3797 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
3798 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
3799 (decide_unroll_runtime_iterations): Likewise.
3800 (decide_unroll_stupid): Likewise.
3802 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3805 * config/i386/i386.md: Add a loop with incssp.
3807 2017-11-27 Martin Jambor <mjambor@suse.cz>
3809 PR tree-optimization/81248
3810 * tree-sra.c (splice_param_accesses): Remove size check.
3811 (decide_one_param_reduction): Fix size check.
3812 * gimple-pretty-print.c (dump_profile): Silence warning.
3813 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
3815 2017-11-27 Jakub Jelinek <jakub@redhat.com>
3818 * dbxout.c (lastlineno): New variable.
3819 (dbx_debug_hooks): Use dbxout_switch_text_section as
3820 switch_text_section debug hook.
3821 (dbxout_function_end): Switch to current_function_section
3822 rather than function_section. If crtl->has_bb_partition,
3823 output just one N_FUN, depending on in_cold_section_p.
3824 (dbxout_source_line): Remember last lineno in lastlineno.
3825 (dbxout_switch_text_section): New function.
3826 (dbxout_function_decl): Adjust dbxout_block caller.
3827 (dbx_block_with_cold_children): New function.
3828 (dbxout_block): Return true if any LBRAC/RBRAC have been
3829 emitted. Use dbx_block_with_cold_children at depth == 0
3830 in second partition. Add PARENT_BLOCKNUM argument, pass
3831 it optionally adjusted to children. Output LBRAC/RBRAC
3832 around recursive call only if the block is in the current
3833 partition, if not and anything was output, emit empty
3835 * final.c (final_scan_insn): Compute cold_function_name
3836 before calling switch_text_section debug hook. Call
3837 that hook even if dwarf2out_do_frame if not emitting
3841 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
3842 TREE_READONLY decls.
3844 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3846 PR rtl-optimization/82488
3847 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
3849 2017-11-26 Julia Koval <julia.koval@intel.com>
3851 * config/i386/i386.c (processor_target_table): Add skylake_cost for
3853 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
3856 2017-11-26 Julia Koval <julia.koval@intel.com>
3858 * config/i386/driver-i386.c (host_detect_local_cpu):
3859 Detect skylake-avx512.
3861 2017-11-26 Julia Koval <julia.koval@intel.com>
3863 * config.gcc: Add -march=cannonlake.
3864 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
3865 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
3866 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
3867 (PTA_CANNONLAKE): New.
3868 (processor_target_table): Add cannonlake.
3869 (ix86_option_override_internal): Ditto.
3870 (fold_builtin_cpu): Ditto.
3871 (get_builtin_code_for_version): Handle cannonlake.
3872 (M_INTEL_COREI7_CANNONLAKE): New.
3873 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
3874 * doc/invoke.texi: Add -march=cannonlake.
3876 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
3878 * plugin.c (add_new_plugin): Use platform-specific library extensions.
3879 (try_init_one_plugin): Alternative implementation for MinGW.
3880 * Makefile.in (plugin_implib): New.
3881 (gengtype-lex.c): Fix broken AIX workaround.
3882 * configure: Regenerate.
3883 * doc/plugins.texi: Document support for MinGW.
3885 2017-11-25 Jakub Jelinek <jakub@redhat.com>
3887 PR rtl-optimization/81553
3888 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
3889 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
3890 is a shift where C1 has different mode than the whole shift, use C1's
3891 mode for MULT rather than the shift's mode.
3894 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
3895 builtins not enabled in the currently selected ISA.
3897 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
3899 PR tree-optimization/71026
3900 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
3901 (insert_reciprocals): Change to insert reciprocals before a division
3902 by a square and to insert the square of a reciprocal.
3903 (execute_cse_reciprocals_1): Change to consider division by a square.
3904 (register_division_in): Add importance parameter.
3906 2017-11-24 Richard Biener <rguenther@suse.de>
3908 PR tree-optimization/82402
3909 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
3910 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
3912 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
3914 * match.pd (0-ptr): New transformation.
3916 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
3919 * ipa-inline.c (inline_small_functions): Set current badnes correctly
3920 when skipping checking.
3922 2017-11-24 Richard Biener <rguenther@suse.de>
3924 PR tree-optimization/83128
3925 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
3926 (vn_reference_lookup_3): Likewise.
3928 2017-11-24 Jakub Jelinek <jakub@redhat.com>
3931 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
3932 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
3935 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
3936 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
3937 check for the latter with gimple_call_builtin_p. Do not handle
3938 BUILT_IN_STPNCPY_CHK which is not a pass through call.
3940 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
3942 * config/arm/arm_neon.h: Fix pragma GCC push_options before
3945 2017-11-23 Julia Koval <julia.koval@intel.com>
3947 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
3948 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
3949 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
3950 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
3951 _mm512_maskz_expandloadu_epi16): New intrinsics.
3952 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
3953 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
3954 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
3955 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
3956 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
3957 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
3958 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
3959 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
3960 _mm256_maskz_expandloadu_epi8): New intrinsics.
3961 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
3962 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
3963 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
3964 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
3965 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
3966 * config/i386/sse.md (VI248_VLBW): New iterator.
3967 (expand<mode>_mask, expand<mode>_maskz): New patterns.
3969 2017-11-23 Julia Koval <julia.koval@intel.com>
3971 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
3972 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
3973 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
3974 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
3975 _mm512_mask_compressstoreu_epi16): New.
3976 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
3977 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
3978 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
3979 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
3980 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
3981 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
3982 _mm256_mask_compressstoreu_epi8): New.
3983 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
3984 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
3985 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
3986 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
3987 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
3988 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
3989 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
3990 __builtin_ia32_compresshi128_mask,
3991 __builtin_ia32_compressstoreuqi512_mask,
3992 __builtin_ia32_compressstoreuhi512_mask,
3993 __builtin_ia32_compressstoreuqi256_mask,
3994 __builtin_ia32_compressstoreuqi128_mask,
3995 __builtin_ia32_compressstoreuhi256_mask,
3996 __builtin_ia32_compressstoreuhi128_mask): New builtins.
3997 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
3999 (ix86_expand_special_args_builtin): Handle new types.
4000 (s4fma_expand): Handle new builtin array.
4001 * config/i386/immintrin.h: Include new headers.
4002 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
4003 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
4005 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4007 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
4008 fixes. Declare temp and g variables at the top in order to avoid
4009 {} in most of the cases.
4011 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
4013 * match.pd (ptr-0): New transformation.
4015 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4017 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
4018 (struct addr_mode_cost_table): New.
4019 (struct tune_params): Add field addr_mode_costs.
4020 * config/arm/arm.c (generic_addr_mode_costs): New.
4021 (arm_slowmul_tune): Initialise addr_mode_costs field.
4022 (arm_fastmul_tune): Likewise.
4023 (arm_strongarm_tune): Likewise.
4024 (arm_xscale_tune): Likewise.
4025 (arm_9e_tune): Likewise.
4026 (arm_marvell_pj4_tune): Likewise.
4027 (arm_v6t2_tune): Likewise.
4028 (arm_cortex_tune): Likewise.
4029 (arm_cortex_a8_tune): Likewise.
4030 (arm_cortex_a7_tune): Likewise.
4031 (arm_cortex_a15_tune): Likewise.
4032 (arm_cortex_a35_tune): Likewise.
4033 (arm_cortex_a53_tune): Likewise.
4034 (arm_cortex_a57_tune): Likewise.
4035 (arm_exynosm1_tune): Likewise.
4036 (arm_xgene1_tune): Likewise.
4037 (arm_cortex_a5_tune): Likewise.
4038 (arm_cortex_a9_tune): Likewise.
4039 (arm_cortex_a12_tune): Likewise.
4040 (arm_cortex_a73_tune): Likewise.
4041 (arm_v7m_tune): Likewise.
4042 (arm_cortex_m7_tune): Likewise.
4043 (arm_v6m_tune): Likewise.
4044 (arm_fa726te_tune): Likewise.
4045 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
4047 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
4049 * config/arm/arm.c (arm_mem_costs): New function.
4050 (arm_rtx_costs_internal): Use arm_mem_costs.
4052 2017-11-23 Mark Wielaard <mark@klomp.org>
4054 * dwarf2out.c (init_sections_and_labels): Use generation to create
4055 unique ranges_section_label and ranges_base_label. Return generation.
4056 (output_rnglists): Add generation argument. Use generation to create
4057 unique ranges labels.
4058 (dwarf2out_finish): Get generation from init_sections_and_labels
4059 and pass generation to output_rnglists.
4061 2017-11-23 Mike Stump <mikestump@comcast.net>
4062 Eric Botcazou <ebotcazou@adacore.com>
4064 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
4065 * cfgloop.h (struct loop): Add unroll field.
4066 * function.h (struct function): Add has_unroll bitfield.
4067 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
4068 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
4069 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
4070 (pass_rtl_unroll_loops::gate): Likewise.
4071 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
4072 for which loop->unroll==1.
4073 (decide_unroll_constant_iterations): Use note for consistency and
4074 take loop->unroll into account. Return early if loop->unroll is set.
4075 Fix thinko in existing test.
4076 (decide_unroll_runtime_iterations): Use note for consistency and
4077 take loop->unroll into account.
4078 (decide_unroll_stupid): Likewise.
4079 * lto-streamer-in.c (input_cfg): Read loop->unroll.
4080 * lto-streamer-out.c (output_cfg): Write loop->unroll.
4081 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
4083 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
4084 (print_loop): Print loop->unroll if set.
4085 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
4086 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
4087 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
4089 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
4090 loop->unroll is set and smaller than the trip count. Otherwise bypass
4091 entirely the heuristics if loop->unroll is set. Remove dead note.
4092 Fix off-by-one bug in other note.
4093 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
4094 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
4096 (tree_unroll_loops_completely): Make static.
4097 (pass_complete_unroll::execute): Use correct type for variable.
4098 (pass_complete_unrolli::execute): Fix formatting.
4099 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
4101 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
4103 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
4104 enable when TARGET_PREFER_AVX128 is set.
4106 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4108 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
4110 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
4111 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
4113 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4115 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
4116 frequencies. Fix estimation of aggregate parameters.
4118 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
4120 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
4123 2017-11-23 Tom de Vries <tom@codesourcery.com>
4125 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
4127 2017-11-23 Tom de Vries <tom@codesourcery.com>
4129 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
4132 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
4135 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
4136 sibcall_value_pcrel_fdpic): Use local variable instead of
4138 (calli_tbr_rel): Add missing operand 2.
4139 (call_valuei_tbr_rel): Add missing operand 3.
4141 2017-11-23 Jakub Jelinek <jakub@redhat.com>
4144 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
4145 bitpos/bitsize covering the whole destination, use store_expr only if
4146 the complex mode is the same. Otherwise, use expand_normal and if
4147 it returns CONCAT, subreg each part separately instead of trying to
4148 subreg the whole result.
4150 2017-11-23 Richard Biener <rguenther@suse.de>
4152 PR tree-optimization/23094
4153 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
4154 come here from walking over backedges in the first iteration.
4155 (vn_reference_lookup_3): Skip clobbers that store the same value.
4157 2017-11-23 Richard Biener <rguenther@suse.de>
4159 PR tree-optimization/81403
4160 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
4161 a block we need a leader relative to.
4162 (phi_translate_1): For nary processing require a leader from
4163 get_representative_for given we run expression simplification
4164 using match-and-simplify. Remove previous fix.
4166 2017-11-22 Jeff Law <law@redhat.com>
4168 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
4169 Use new method allocate_value_range rather than accessing the
4170 vrp_value_range_pool data member directly.
4171 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
4172 to use extract_range_from_stmt method to avoid need for
4173 extract_range_from_assignment method.
4174 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
4175 method rather than setting values_propgated data member directly.
4176 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
4177 and values propagated data members and extract_range_from_assignment
4178 method. Reorder private data members to conform to standards.
4179 Add new methods set_lattice_propagation_complete and
4180 allocate_value_range.
4182 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
4184 PR rtl-optimization/83030
4185 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
4186 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
4187 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
4188 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
4189 CROSSING_JUMP_P flag is already set before setting it.
4190 * cfgrtl.c (fixup_partition_crossing): Likewise.
4191 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
4194 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4196 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
4197 Handle the case where both arguments are using gen_const_vec_series.
4199 2017-11-22 David Malcolm <dmalcolm@redhat.com>
4202 * pretty-print.c (pp_format): Move quoting implementation to
4203 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
4204 to pass address of "quote" local.
4205 (pp_begin_quote): New function.
4206 (pp_end_quote): New function.
4207 * pretty-print.h (printer_fn): Convert penultimate param from bool
4209 (pp_begin_quote): New decl.
4210 (pp_end_quote): New decl.
4211 * tree-diagnostic.c (default_tree_printer): Convert penultimate
4212 param from bool to bool *.
4213 * tree-diagnostic.h (default_tree_printer): Likewise.
4215 2017-11-22 Jeff Law <law@redhat.com>
4217 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
4218 Initialize vr_values.
4219 (evrp_range_analyzer::try_find_new_range): Call methods attached to
4220 vr_values via vr_values class instance rather than delegators.
4221 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
4222 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
4223 (evrp_range_analyzer::push_value_range): Likewise.
4224 (evrp_range_analyzer::pop_value_range): Likewise.
4225 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
4226 most delegators. Those remaining are exposed as public interfaces.
4227 Make vr_values a pointer and private.
4228 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
4230 (evrp_range_analyzer::get_vr_value): New method.
4231 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
4232 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
4233 (class evrp_dom_walker): Attach evrp_folder class, initialize
4234 it in the ctor. Remove temporary delegators.
4235 (evrp_dom_walker::before_dom_children): Call methods in attached
4236 evrp_range_analyzer class via class instance pointer. Use
4237 free value_range_constant_singleton to remove need for
4238 op_with_constant_singleton_value delegator method. Do not
4239 create a vrp_prop class instance for every call! Narrow
4240 scope of a couple variables.
4241 (evrp_dom_walker::cleanup): Call methods in attached
4242 evrp_range_analyzer class via class instance pointer.
4243 * vr-values.h (class vr_values): Privatize many methods and
4246 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4248 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
4249 * cfgexpand.c (expand_debug_expr): Remove handling for them.
4250 * expr.c (expand_expr_real_2): Likewise.
4251 * fold-const.c (const_unop): Likewise.
4252 * optabs-tree.c (optab_for_tree_code): Likewise.
4253 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
4254 * tree-inline.c (estimate_operator_cost): Likewise.
4255 * tree-pretty-print.c (dump_generic_node): Likewise.
4256 (op_code_prio): Likewise.
4257 (op_symbol_code): Likewise.
4258 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
4259 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
4260 * internal-fn.c (direct_internal_fn_optab): New function.
4261 (direct_internal_fn_array, direct_internal_fn_supported_p
4262 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
4263 * fold-const-call.c (fold_const_reduction): New function.
4264 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
4266 * tree-vect-loop.c: Include internal-fn.h.
4267 (reduction_code_for_scalar_code): Rename to...
4268 (reduction_fn_for_scalar_code): ...this and return an internal
4270 (vect_model_reduction_cost): Take an internal_fn rather than
4272 (vect_create_epilog_for_reduction): Likewise. Build calls rather
4274 (vectorizable_reduction): Use internal functions rather than tree
4275 codes for the reduction operation. Update calls to the functions
4277 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
4278 Use calls to internal functions rather than REDUC tree codes.
4279 * config/aarch64/aarch64-simd.md: Update comment accordingly.
4281 2017-11-22 Olivier Hainque <hainque@adacore.com>
4283 * config/vxworks.c (vxworks_override_options): Pick default
4284 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
4285 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
4286 DWARF_GNAT_ENCODINGS_DEFAULT.
4287 * config/vxworksae.h: Likewise.
4289 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
4291 PR tree-optimization/83104
4292 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
4295 2017-11-22 Marek Polacek <polacek@redhat.com>
4296 H.J. Lu <hongjiu.lu@intel.com>
4297 Jason Merrill <jason@redhat.com>
4302 * calls.c (initialize_argument_information): Call
4303 warn_parameter_passing_abi target hook.
4304 (store_one_arg): Use 0 for empty record size. Don't push 0 size
4305 argument onto stack.
4306 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
4307 * common.opt: Update -fabi-version description.
4308 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
4309 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
4311 (ix86_is_empty_record): New function.
4312 (ix86_warn_parameter_passing_abi): New function.
4313 (TARGET_EMPTY_RECORD_P): Redefine.
4314 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
4315 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
4316 * doc/tm.texi: Regenerated.
4317 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
4318 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
4319 * dwarf2out.c (get_ultimate_context): Move to tree.c.
4320 * explow.c (hard_function_value): Call arg_int_size_in_bytes
4321 instead of int_size_in_bytes.
4322 * expr.c (copy_blkmode_to_reg): Likewise.
4323 * function.c (aggregate_value_p): Return 0 for empty types.
4324 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
4326 (locate_and_pad_parm): Call arg size_in_bytes instead
4328 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
4329 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
4330 * target.def (empty_record_p, warn_parameter_passing_abi): New target
4332 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
4333 (std_gimplify_va_arg_expr): Skip empty records. Call
4334 arg_size_in_bytes instead size_in_bytes.
4335 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
4336 * tree-core.h (tree_type_common): Add empty_flag.
4337 (tree_decl_common): Update comments.
4338 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
4340 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4341 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
4343 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
4344 * tree.c (default_is_empty_type): New function.
4345 (default_is_empty_record): New function.
4346 (arg_int_size_in_bytes): New function.
4347 (arg_size_in_bytes): New function.
4348 (get_ultimate_context): New function.
4349 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
4350 TRANSLATION_UNIT_WARN_EMPTY_P.
4351 (default_is_empty_record, arg_int_size_in_bytes,
4352 arg_size_in_bytes, get_ultimate_context): Declare.
4354 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4356 * config/arm/arm.c (cmse_clear_registers): New function.
4357 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
4358 code by call to cmse_clear_registers.
4359 (cmse_nonsecure_entry_clear_before_return): Likewise.
4361 2017-11-22 Tamar Christina <tamar.christina@arm.com>
4363 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
4364 (vdot_s32, vdotq_s32): New.
4365 (vdot_lane_u32, vdotq_lane_u32): New.
4366 (vdot_lane_s32, vdotq_lane_s32): New.
4369 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4372 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
4373 for unsigned values with fewer HWIs than the precision.
4374 (test_overflow): New function.
4375 (wide_int_cc_tests): Call it.
4377 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
4379 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
4380 has the same mode class as Pmode.
4382 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4384 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
4385 padding_bits_to_clear_ptr.
4386 (cmse_nonsecure_entry_clear_before_return): Likewise.
4388 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
4390 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
4391 auto_sbitap instead of integer bitfield to control register needing
4394 2017-11-22 Jakub Jelinek <jakub@redhat.com>
4396 PR tree-optimization/83044
4397 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
4398 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
4399 that need the upper bound. Subtract offset from
4400 get_addr_base_and_unit_offset only if positive and subtract it
4401 before division by eltsize rather than after it.
4404 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
4405 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
4406 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
4410 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
4411 expand_binop with *mul_widen_optab, make sure at least one of the
4412 operands doesn't have VOIDmode.
4415 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
4417 PR rtl-optimization/82044
4418 PR tree-optimization/82042
4419 * dse.c (record_store): Check for overflow.
4420 (check_mem_read_rtx): Properly check for overflow if width == -1, call
4421 add_wild_read instead of clear_rhs_from_active_local_stores on
4422 overflow and log it into dump_file.
4424 2017-11-22 Richard Biener <rguenther@suse.de>
4426 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
4427 fake edges to exit when looking for a place to insert.
4428 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
4430 (insert_into_preds_of_block): Commit edge insertion immediately,
4431 assert that doesn't require new BBs.
4432 (fini_pre): Release expressions.
4433 (pass_pre::execute): Shuffle things around a bit, if the fn
4434 is too large do not compute AVAIL either as this is really the
4437 2017-11-22 Richard Biener <rguenther@suse.de>
4439 PR tree-optimization/83089
4440 * tree-if-conv.c (pass_if_conversion::execute): If anything
4441 changed reset SCEV and free the number of iteration estimates.
4443 2017-11-21 Martin Sebor <msebor@redhat.com>
4445 PR tree-optimization/82945
4446 * calls.h (warn_nonstring_bound): Remove unused function.
4448 2017-11-21 Martin Sebor <msebor@redhat.com>
4450 PR tree-optimization/82945
4451 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
4452 * calls.h (maybe_warn_nonstring_arg): Declare new function.
4453 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
4455 (initialize_argument_information): Call maybe_warn_nonstring_arg.
4456 * calls.h (get_attr_nonstring_decl): Declare new function.
4457 * doc/extend.texi (attribute nonstring): Update.
4458 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
4459 get_attr_nonstring_decl and handle it.
4460 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
4461 detection of nul-termination.
4462 (strlen_to_stridx): Change to a pointer.
4463 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
4464 (pass_strlen::execute): Same.
4466 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
4468 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
4469 for the new option -mprefer-vector-width=[none|128|256|512].
4470 * config/i386/i386.c (ix86_target_string): remove old style options
4471 -mprefer-avx256 and make -mprefer-avx128 as alias.
4472 (ix86_option_override_internal): Apply defaults for the
4473 -mprefer-vector-width=[128|256] option.
4474 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
4475 Implement macros to work with -mprefer-vector-width=.
4476 * config/i386/i386.opt: Implemented option
4477 -mprefer-vector-width=[none|128|256|512].
4478 * doc/invoke.texi: Documentation for
4479 -mprefer-vector-width=[none|128|256|512].
4481 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
4483 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
4485 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
4487 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
4489 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
4491 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
4493 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
4494 * expr.c (expand_expr_real_2): Likewise.
4495 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
4496 fold_binary_loc): Likewise.
4497 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
4498 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
4499 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
4500 MINUS_EXPR transformations.
4501 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
4502 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
4503 * tree-inline.c (estimate_operator_cost): Likewise.
4504 * tree-pretty-print.c (dump_generic_node, op_code_prio,
4505 op_symbol_code): Likewise.
4506 * tree-vect-stmts.c (vectorizable_operation): Likewise.
4507 * vr-values.c (extract_range_from_binary_expr): Likewise.
4508 * varasm.c (initializer_constant_valid_p_1): Likewise.
4509 * tree.def: New tree code POINTER_DIFF_EXPR.
4511 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
4513 * config/i386/i386.md (*bswap<mode>2_movbe): Add
4514 integer suffix to movbe mnemonic.
4515 (*bswaphi2_movbe): Ditto.
4516 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
4518 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4521 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
4522 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
4523 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
4525 2017-11-21 Martin Liska <mliska@suse.cz>
4527 * tree-inline.c (expand_call_inline): Remove not needed
4530 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
4531 Jakub Jelinek <jakub@redhat.com>
4534 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
4535 Remove static keyword from f variable.
4537 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4539 PR tree-optimization/83086
4540 * gimple-ssa-store-merging.c
4541 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
4542 rather than n.base_addr.
4544 2017-11-21 Martin Liska <mliska@suse.cz>
4546 PR rtl-optimization/82044
4547 PR tree-optimization/82042
4548 * dse.c (check_mem_read_rtx): Check for overflow.
4550 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4552 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
4555 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4557 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
4558 and its corresponding call arg location note.
4560 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
4561 Andrew Burgess <andrew.burgess@embecosm.com>
4563 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
4565 (arc_return_slot_offset): Likewise.
4566 (arc_eh_return_address_location): New declaration.
4567 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
4568 (MUST_SAVE_REGISTER): Add exception handler case.
4569 (MUST_SAVE_RETURN_ADDR): Likewise.
4570 (arc_frame_pointer_required): Likewise.
4571 (arc_frame_pointer_needed): New function.
4572 (arc_compute_frame_size): Changed.
4573 (arc_expand_prologue): Likewise.
4574 (arc_expand_epilogue): Likewise.
4575 (arc_initial_elimination_offset): Likewise.
4576 (arc_return_slot_offset): Delete.
4577 (arc_eh_return_address_location): New function.
4578 (arc_builtin_setjmp_frame_value): Likewise.
4579 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
4580 (EH_RETURN_STACKADJ_RTX): Define.
4581 (EH_RETURN_HANDLER_RTX): Likewise.
4582 * config/arc/arc.md (eh_return): Delete.
4584 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
4586 * print-tree.h (debug_vec_tree): Remove prototype.
4587 * gdbinit.in (pvt): Remove macro.
4589 2017-11-21 Jakub Jelinek <jakub@redhat.com>
4591 PR tree-optimization/83047
4592 * gimple-ssa-store-merging.c
4593 (imm_store_chain_info::output_merged_store): If the loads with the
4594 same vuse are in different basic blocks, for load_gsi pick a load
4595 location that is dominated by the other loads.
4598 * config/i386/i386.c (ix86_memmodel_check): Start
4599 -Winvalid-memory-model diagnostics with lowercase letter.
4602 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
4603 set in_cold_section_p to the partition containing loc_list->first.
4604 When seeing loc_list->last_before_switch node, update secname and
4605 perform range_across_switch second partition handling only after that.
4608 * run-rtl-passes.c: Include debug.h.
4609 (run_rtl_passes): Call debug_hooks->assembly_start.
4610 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
4614 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
4615 OPTAB_DIRECT in calls to expand_simple_binop.
4617 2017-11-20 David Malcolm <dmalcolm@redhat.com>
4620 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
4622 2017-11-20 David Malcolm <dmalcolm@redhat.com>
4625 * spellcheck.h (best_match::blithely_get_best_candidate): New
4628 2017-11-20 Jakub Jelinek <jakub@redhat.com>
4630 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
4631 with lower case letter.
4633 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
4635 * config/i386/i386.md (bswaphi2): New expander.
4636 (*bswaphi2_movbe): New insn pattern.
4637 (bswaphi -> rorhi pepehole2): New peephole pattern.
4639 2017-11-20 Jeff Law <law@redhat.com>
4641 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
4642 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
4643 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
4644 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
4645 gimple-ssa-evrp-analyze.h.
4647 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
4648 set BB_VISITED here.
4649 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
4651 2017-11-20 Bin Cheng <bin.cheng@arm.com>
4653 * tree-predcom.c: Add general comment on Store-Store chains.
4654 (split_data_refs_to_components): Postpone clearing eliminate_store_p
4656 (get_chain_last_ref_at): Rename into...
4657 (get_chain_last_write_at): ...this.
4658 (get_chain_last_write_before_load): New function.
4659 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
4660 CT_STORE_STORE when write reference is added.
4661 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
4662 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
4663 (initialize_root_vars_store_elim_1): Ditto.
4664 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
4665 definition is created.
4666 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
4667 chain by replacing it with dominant stored value.
4669 2017-11-20 Bin Cheng <bin.cheng@arm.com>
4671 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
4673 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
4675 * vr-values.c (extract_range_from_binary_expr): Use a full range
4678 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
4680 * config/arm/arm.md (R4_REGNUM): Define constant.
4681 (nonsecure_call_internal): Remove r4 clobber.
4682 (nonsecure_call_value_internal): Likewise.
4683 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
4684 clobber and resequence match_operands.
4685 (nonsecure_call_value_reg_thumb1_v5): Likewise.
4686 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
4687 (nonsecure_call_value_reg_thumb2): Likewise.
4689 2017-11-20 Jakub Jelinek <jakub@redhat.com>
4691 PR tree-optimization/78821
4692 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
4693 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
4694 to the address of the base rather than the base itself.
4695 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
4696 (find_bswap_or_nop_finalize): New function.
4697 (find_bswap_or_nop): Use it.
4698 (bswap_replace): Return a tree rather than bool, change first
4699 argument from gimple * to gimple_stmt_iterator, allow inserting
4700 into an empty sequence, allow ins_stmt to be NULL - then emit
4701 all stmts into gsi. Fix up MEM_REF address gimplification.
4702 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
4703 (struct store_immediate_info): Add N and INS_STMT non-static
4705 (store_immediate_info::store_immediate_info): Initialize them
4706 from newly added ctor args.
4707 (merged_store_group::apply_stores): Formatting fixes. Sort by
4709 (stmts_may_clobber_ref_p): For stores call also
4710 refs_anti_dependent_p.
4711 (gather_bswap_load_refs): New function.
4712 (imm_store_chain_info::try_coalesce_bswap): New method.
4713 (imm_store_chain_info::coalesce_immediate_stores): Use it.
4714 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
4715 (imm_store_chain_info::output_merged_store): Fail if number of
4716 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
4717 and NOP_EXPR rhs_code.
4718 (pass_store_merging::process_store): Compute n and ins_stmt, if
4719 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
4720 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
4722 (pass_store_merging::execute): Calculate dominators.
4724 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
4725 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
4726 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
4727 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
4728 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
4729 class pass_optimize_bswap, bswap_replace,
4730 pass_optimize_bswap::execute): Moved to ...
4731 * gimple-ssa-store-merging.c: ... this file.
4732 Include optabs-tree.h.
4733 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
4734 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
4735 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
4736 anonymous namespace, remove static keywords.
4737 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
4738 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
4740 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
4743 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
4745 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
4747 * vec.h (debug_helper): New function.
4748 (DEFINE_DEBUG_VEC): New macro.
4749 * hash-set.h (debug_helper): New function.
4750 (DEFINE_DEBUG_HASH_SET): New macro.
4751 * cfg.c (debug_slim (edge)): New function.
4752 Call DEFINE_DEBUG_VEC for edges.
4753 Call DEFINE_DEBUG_HASH_SET for edges.
4754 * cfghooks.c (debug_slim (basic_block)): New function.
4755 Call DEFINE_DEBUG_VEC for basic blocks.
4756 Call DEFINE_DEBUG_HASH_SET for basic blocks.
4757 * print-tree.c (debug_slim): New function to handle trees.
4758 Call DEFINE_DEBUG_VEC for trees.
4759 Call DEFINE_DEBUG_HASH_SET for trees.
4760 (debug (vec<tree, va_gc>) &): Remove.
4761 (debug (<vec<tree, va_gc>) *): Remove.
4762 * print-rtl.c (debug_slim): New function to handle const_rtx.
4763 Call DEFINE_DEBUG_VEC for rtx_def.
4764 Call DEFINE_DEBUG_VEC for rtx_insn.
4765 Call DEFINE_DEBUG_HASH_SET for rtx_def.
4766 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
4767 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
4768 (debug (vec<rtx_insn *> *ptr): Remove.
4769 (debug_insn_vector): Remove.
4770 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
4772 2017-11-20 Tom de Vries <tom@codesourcery.com>
4774 PR rtl-optimization/82020
4775 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
4776 IF_THEN_ELSE condition.
4778 2017-11-19 Jeff Law <law@redhat.com>
4780 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
4781 of degenerates resulting from ignoring an edge.
4783 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4786 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
4788 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4791 * profile-count.c (profile_count::to_sreal_scale): Fix return value
4792 for uninitialied counts.
4794 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4797 * tree-inline.c (estimate_num_insns): Set to 1 at least.
4799 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4802 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
4804 2017-11-19 Tom de Vries <tom@codesourcery.com>
4806 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
4809 2017-11-19 Tom de Vries <tom@codesourcery.com>
4811 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
4814 2017-11-19 Tom de Vries <tom@codesourcery.com>
4816 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4819 2017-11-19 Tom de Vries <tom@codesourcery.com>
4821 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
4823 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
4825 2017-11-19 Tom de Vries <tom@codesourcery.com>
4827 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
4829 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
4830 semicolon after MCORE_EXPORT_NAME call.
4832 2017-11-19 Tom de Vries <tom@codesourcery.com>
4835 * vmsdbgout.c (vmsdbgout_early_finish): New function.
4836 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
4838 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
4840 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
4841 (cgraph_edge::clone): Cleanup updating of profile.
4842 * ipa-cp.c (update_profiling_info): Likewise.
4843 * ipa-inline-transform.c (inline_transform): Likewise.
4844 * ipa-inline.c (inline_small_functions): Add missing space to dump.
4845 * ipa-split.c (execute_split_functions): Do not split when function
4847 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
4848 * profile-count.c (profile_count::dump): Add global0.
4849 (profile_count::to_cgraph_frequency): Do not ICE when entry is
4851 (profile_count::to_sreal_scale): Likewise.
4852 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
4853 (profile_count::combine_with_ipa_count): New function.
4854 * profile-count.h (profile_guessed_global0adjusted): New.
4855 (profile_count::adjusted_zero): New.
4856 (profile_count::global0adjusted): New.
4857 (profile_count::combine_with_ipa_count): New.
4858 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
4859 correct profile of return block of split functions.
4860 (copy_cfg_body): Remove unused profile_count.
4861 (copy_body): Likewise.
4862 (expand_call_inline): Update.
4863 (tree_function_versioning): Update.
4865 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
4867 * hash-set.h (hash_set::empty): New.
4868 * tree-ssa-threadbackward.h: Delete.
4869 * tree-ssa-threadbackward.c (class thread_jumps): New.
4870 Move max_threaded_paths into class.
4871 (fsm_find_thread_path): Remove arguments that are now in class.
4872 (profitable_jump_thread_path): Rename to...
4873 (thread_jumps::profitable_jump_thread_path): ...this.
4874 (convert_and_register_jump_thread_path): Rename to...
4875 (thread_jumps::convert_and_register_current_path): ...this.
4876 (check_subpath_and_update_thread_path): Rename to...
4877 (thread_jumps::check_subpath_and_update_thread_path): ...this.
4878 (register_jump_thread_path_if_profitable): Rename to...
4879 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
4880 (handle_phi): Rename to...
4881 (thread_jumps::handle_phi): ...this.
4882 (handle_assignment): Rename to...
4883 (thread_jumps::handle_assignment): ...this.
4884 (fsm_find_control_statement_thread_paths): Rename to...
4885 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
4886 (find_jump_threads_backwards): Rename to...
4887 (thread_jumps::find_jump_threads_backwards): ...this.
4888 Initialize path local data.
4889 (pass_thread_jumps::execute): Call find_jump_threads_backwards
4890 from within thread_jumps class.
4891 (pass_early_thread_jumps::execute): Same.
4893 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4895 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
4897 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4899 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
4900 to canonical location.
4902 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4905 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
4907 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
4909 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
4911 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
4912 rcrt1.o%s/grcrt1.o%s for -static-pie.
4914 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4916 * i386.c (ix86_multiplication_cost, ix86_division_cost,
4917 ix86_shift_rotate_cost): Break out from ...
4918 (ix86_rtx_costs): ... here.
4919 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
4922 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4924 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
4925 when entry block was promoted unlikely.
4926 (estimate_bb_frequencies): Increase frequency scale.
4927 * profile-count.h (profile_count): Export precision info.
4929 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4931 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
4932 disturbin profile of entry block.
4934 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4936 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
4939 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4941 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
4942 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
4944 2017-11-17 Jeff Law <law@redhat.com>
4946 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
4947 from evrp_dom_walker class. Various methods moved into new class.
4948 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
4949 (evrp_range_analyzer::enter): New method.
4950 (evrp_range_analyzer::leave): New method.
4951 (evrp_dom_walker): Remove delegators no longer needed by this class.
4952 Replace vr_values data member with evrp_range_analyzer
4954 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
4955 method extracted from evrp_dom_walker::before_dom_children.
4956 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
4957 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
4959 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
4960 Add private copy constructor and move assignment operators.
4961 Privatize methods and class data where trivially possible.
4962 (evrp_dom_walker::cleanup): New function, extracted from
4963 execute_early_vrp. Simplify access to class data.
4965 * vr-values.h (get_output_for_vrp): Prototype.
4966 * vr-values.c (get_output_for_vrp): New function extracted from
4967 vrp_visit_assignment_or_call and extract_range_from_stmt.
4968 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
4970 2017-11-17 Luis Machado <luis.machado@linaro.org>
4972 * config/aarch64/aarch64.c
4973 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
4974 (qdf24xx_tunings) <autoprefetcher_model>: Set to
4975 tune_params::AUTOPREFETCHER_WEAK.
4977 2017-11-17 Tamar Christina <tamar.christina@arm.com>
4980 * config/arm/arm.c (arm_valid_target_attribute_rec):
4981 Parse "arch=" and "+<ext>".
4982 (arm_valid_target_attribute_tree): Re-init global options.
4983 (arm_option_override): Make non-static.
4984 (arm_options_perform_arch_sanity_checks): Make errors fatal.
4985 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
4986 (__ARM_FEATURE_CRC32): Support undef.
4987 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
4988 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
4990 2017-11-17 David Malcolm <dmalcolm@redhat.com>
4992 * gdbinit.in (break-on-diagnostic): New command.
4994 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4996 * config/i386/i386.c (ix86_expand_epilogue): Change simple
4997 return to indirect jump for EH return if control-flow
4998 protection is enabled. Change explicit 'false' argument in
4999 pro_epilogue_adjust_stack with a value of
5001 * config/i386/i386.md (simple_return_indirect_internal):
5002 Remove SImode restriction to support 64-bit.
5004 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5006 * combine.c (added_notes_insn): New.
5007 (try_combine): Handle added_notes_insn like added_links_insn.
5008 Rewrite return value code.
5009 (distribute_notes): Set added_notes_insn to the earliest insn we added
5012 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
5014 PR rtl-optimization/82621
5015 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
5016 dest of one of those SETs is unused.
5018 2017-11-17 Richard Biener <rguenther@suse.de>
5021 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
5022 * tree-pretty-print.c (dump_generic_node): Handle
5023 annot_expr_parallel_kind.
5024 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
5025 * gimplify.c (gimple_boolify): Likewise.
5027 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5029 * config.gcc (extra_headers): Add cet.h for x86 targets.
5030 * config/i386/cet.h: New file.
5031 * doc/install.texi: Add --enable-cet/--disable-cet.
5033 2017-11-17 Richard Biener <rguenther@suse.de>
5035 PR tree-optimization/83017
5036 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
5037 (gen_parallel_loop): Properly count iterations.
5038 (parallelize_loops): Handle loop->can_be_parallel independent
5039 of flag_loop_parallelize_all. Make static profitability test match
5041 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
5042 * invoke.texi (parloops-min-per-thread): Document.
5044 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
5046 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
5047 upstreaming review comments.
5049 2017-11-17 Tamar Christina <tamar.christina@arm.com>
5051 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
5052 with fast unaligned access.
5053 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
5055 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
5057 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
5058 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
5060 2017-11-17 Richard Biener <rguenther@suse.de>
5062 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
5063 folding of references.
5065 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
5068 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
5069 of replacing call to strncmp with corresponding call to strcmp when
5072 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
5074 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
5075 option prefer-avx256 for skylake-avx512 configuration.
5076 * config/i386/i386.c (ix86_option_override_internal): Ditto.
5077 (get_builtin_code_for_version): Ditto.
5079 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5080 Monk Chiang <sh.chiang04@gmail.com>
5082 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
5083 (FIXED_REGISTERS): Reserve more register numbers.
5084 (CALL_USED_REGISTERS): Likewise.
5085 (REG_ALLOC_ORDER): Likewise.
5086 (REG_CLASS_CONTENTS): Likewise.
5087 (REGISTER_NAMES): Likewise.
5089 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
5090 Kito Cheng <kito.cheng@gmail.com>
5092 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
5094 * config/nds32/iterators.md: Add vector mode iterators and attributes.
5096 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
5098 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
5099 parameter list for vec_splats.
5101 2017-11-16 Joseph Myers <joseph@codesourcery.com>
5103 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
5105 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
5107 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
5110 * doc/install.texi: Document incompatibility of Perl >=5.6.26
5111 with the required version of automake 1.11.6.
5113 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
5115 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
5117 (DU_C2_3_power9): Correct reservation combinations.
5118 (FP_DIV_power9, VEC_DIV_power9): New.
5119 (power9-alu): Split out rotate/shift...
5120 (power9-rot): ...to here, correct dispatch resource.
5121 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
5123 (power9-fp): Correct latency.
5124 (power9-sdiv): Add div/sqrt resource.
5125 (power9-ddiv): Correct latency, add div/sqrt resource.
5126 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
5127 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
5129 (power9-qpdiv, power9-qpmul): Adjust resource usage.
5131 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
5133 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
5134 switch statement mapping KF built-ins to TF built-ins if we don't
5135 have the proper ISA 3.0 assembler support.
5137 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5139 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
5140 (lower_emutls_function_body): Do not compute it.
5142 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5144 * ipa-split.c (split_bb_info): Turn time to sreal.
5145 (split_point): Likewise.
5146 (dump_split_point): Likewise.
5147 (fine_split_points): Likewise.
5148 (execute_split_functions): Only zero split_bbs; turn time to sreals.
5150 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5152 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
5155 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5157 * predict.c (combine_predictions_for_bb): Preserve zero predicted
5159 (expensive_function_p): Remove useless assert.
5160 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
5162 2017-11-16 Martin Sebor <msebor@redhat.com>
5164 PR tree-optimization/82588
5165 PR tree-optimization/82583
5166 * tree-vrp.c (check_array_ref): Handle flexible array members,
5167 string literals, and inner indices.
5168 (search_for_addr_array): Add detail to diagnostics.
5170 2017-11-16 Nathan Sidwell <nathan@acm.org>
5174 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
5175 (SET_DECL_ASSEMBLER_NAME): Forward to
5176 overwrite_decl_assembler_name.
5177 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
5178 (overwrite_decl_assembler_name): Declare.
5179 * tree.c (overwrite_decl_assembler_name): New.
5180 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
5181 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
5182 (LANG_HOOKS_INITIALIZER): Add it.
5183 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
5184 * langhooks.c (lhd_set_decl_assembler_name): Use
5185 SET_DECL_ASSEMBLER_NAME.
5186 (lhd_overwrite_decl_assembler_name): Default implementation.
5188 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5189 Jackson Woodruff <jackson.woodruff@arm.com>
5191 PR tree-optimization/71026
5192 * match.pd: Canonicalize constant multiplies in division.
5194 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
5196 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
5197 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
5198 Remove OPT_fomit_frame_pointer entry.
5199 * common/config/arc/arc-common.c: Likewise.
5200 * common/config/arm/arm-common.c: Likewise.
5201 * common/config/avr/avr-common.c: Likewise.
5202 * common/config/c6x/c6x-common.c: Likewise.
5203 * common/config/cr16/cr16-common.c: Likewise.
5204 * common/config/cris/cris-common.c: Likewise.
5205 * common/config/epiphany/epiphany-common.c: Likewise.
5206 * common/config/fr30/fr30-common.c: Likewise.
5207 * common/config/frv/frv-common.c: Likewise.
5208 * common/config/ia64/ia64-common.c: Likewise.
5209 * common/config/iq2000/iq2000-common.c: Likewise.
5210 * common/config/lm32/lm32-common.c: Likewise.
5211 * common/config/m32r/m32r-common.c: Likewise.
5212 * common/config/mcore/mcore-common.c: Likewise.
5213 * common/config/microblaze/microblaze-common.c: Likewise.
5214 * common/config/mips/mips-common.c: Likewise.
5215 * common/config/mmix/mmix-common.c: Likewise.
5216 * common/config/mn10300/mn10300-common.c: Likewise.
5217 * common/config/nios2/nios2-common.c: Likewise.
5218 * common/config/pa/pa-common.c: Likewise.
5219 * common/config/pdp11/pdp11-common.c: Likewise.
5220 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
5221 * common/config/riscv/riscv-common.c: Likewise.
5222 * common/config/rs6000/rs6000-common.c: Likewise.
5223 * common/config/rx/rx-common.c: Likewise.
5224 * common/config/s390/s390-common.c: Likewise.
5225 * common/config/sh/sh-common.c: Likewise.
5226 * common/config/sparc/sparc-common.c: Likewise.
5227 * common/config/tilegx/tilegx-common.c: Likewise.
5228 * common/config/tilepro/tilepro-common.c: Likewise.
5229 * common/config/v850/v850-common.c: Likewise.
5230 * common/config/visium/visium-common.c: Likewise.
5231 * common/config/xstormy16/xstormy16-common.c: Likewise.
5232 * common/config/xtensa/xtensa-common.c: Likewise.
5233 * invoke.texi (-fomit-frame-pointer): Update documentation.
5235 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
5237 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
5239 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5241 * cfg.c (scale_bbs_frequencies_int,
5242 cale_bbs_frequencies_gcov_type): Remove.
5243 * cfg.h (scale_bbs_frequencies_int,
5244 cale_bbs_frequencies_gcov_type): Remove.
5246 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5248 * tree-ssa-loop-manip.c
5249 (scale_dominated_blocks_in_loop): Update to profile counts.
5250 (tree_transform_and_unroll_loop): Likewise.
5252 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5254 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
5255 scale_bbs_frequencies_int.
5257 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5259 * final.c (compute_alignments): Use counts rather than frequencies.
5261 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5263 * cfgloopanal.c: Include sreal.h
5264 (average_num_loop_insns): Use counts and sreal for accounting.
5266 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5268 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
5271 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5272 Kito Cheng <kito.cheng@gmail.com>
5274 * config/nds32/constraints.md: Provide more constraints.
5275 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
5276 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
5277 support constraints usage.
5279 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5281 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
5283 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
5284 Kito Cheng <kito.cheng@gmail.com>
5286 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
5287 * config/nds32/nds32.opt: Refine the layout.
5288 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
5289 TARGET_EXT_STRING): Support new options.
5290 * config/nds32/nds32.h: Likewise.
5291 * config/nds32/nds32.md: Likewise.
5292 * config/nds32/nds32-predicates.c: Likewise.
5293 * config/nds32/constraints.md: Likewise.
5294 * common/config/nds32/nds32-common.c: Likewise.
5296 2017-11-16 Julia Koval <julia.koval@intel.com>
5299 * config/i386/gfniintrin.h: Add sse check.
5300 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
5302 2017-11-16 Julia Koval <julia.koval@intel.com>
5304 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
5305 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
5306 (ix86_handle_option): Handle -mavx512vbmi2.
5307 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
5308 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
5309 * config/i386/i386-c.c (__AVX512VBMI2__): New.
5310 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
5311 (ix86_valid_target_attribute_inner_p): Ditto.
5312 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
5313 * config/i386/i386.opt (mavx512vbmi2): New option.
5314 * doc/invoke.texi: Add new option.
5316 2017-11-16 Julia Koval <julia.koval@intel.com>
5318 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
5319 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
5320 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
5321 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
5322 _mm512_gf2p8mul_epi8): New intrinsics.
5323 * config/i386/i386-builtin-types.def
5324 (V64QI_FTYPE_V64QI_V64QI): New type.
5325 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
5326 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
5327 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
5328 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
5329 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
5330 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
5332 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
5334 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
5335 explicitly as a stream of bytes.
5337 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5339 * config/rs6000/altivec.h (vec_xst_be): New #define.
5340 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
5341 and externalize from *altivec_vperm_<mode>_internal.
5342 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
5344 (XL_BE_V8HI): Likewise.
5345 (XL_BE_V4SI): Likewise.
5346 (XL_BE_V4SI): Likewise.
5347 (XL_BE_V2DI): Likewise.
5348 (XL_BE_V4SF): Likewise.
5349 (XL_BE_V2DF): Likewise.
5350 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
5351 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
5352 all array entries with these keys: VSX_BUILTIN_VEC_XL,
5353 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
5354 VSX_BUILTIN_VEC_XST_BE.
5355 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
5356 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
5358 (altivec_init_builtins): Replace conditional calls to def_builtin
5359 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
5360 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
5361 with unconditional calls. Remove calls to def_builtin for
5362 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
5363 __builtin_vec_xst_be.
5364 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
5365 to define_expand, and add alternate RTL generation for P8.
5366 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
5367 vsx_ld_elemrev_v8hi.
5368 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
5369 add alternate RTL generation for P8.
5370 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
5371 vsx_ld_elemrev_v16qi.
5372 (vsx_st_elemrev_v8hi): Convert define_insn
5373 to define_expand, and add alternate RTL generation for P8.
5374 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
5375 vsx_st_elemrev_v8hi.
5376 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
5377 add alternate RTL generation for P8.
5378 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
5379 vsx_st_elemrev_v16qi.
5381 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
5384 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
5385 TARGET_AVX512ER check.
5386 (ix86_option_override_internal): Set MASK_VZEROUPPER if
5387 neither -mzeroupper nor -mno-zeroupper is used and
5388 TARGET_EMIT_VZEROUPPER is set.
5389 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
5390 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
5392 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
5394 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
5395 folding of vector compares.
5396 (fold_build_vec_cmp): New helper function.
5397 (fold_compare_helper): New helper function.
5398 (builtin_function_type): Add compare builtins to the list of functions
5399 having unsigned arguments. Cosmetic updates to comment indentation.
5400 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
5401 the not+eq combination.
5403 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5405 PR tree-optimization/82726
5406 PR tree-optimization/70754
5407 * tree-predcom.c (order_drefs_by_pos): New function.
5408 (combine_chains): Move code setting has_max_use_after to...
5409 (try_combine_chains): ...here. New parameter. Sort combined chains
5410 according to position information.
5411 (tree_predictive_commoning_loop): Update call to above function.
5412 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
5414 2017-11-15 Bin Cheng <bin.cheng@arm.com>
5416 PR tree-optimization/82726
5418 2017-01-23 Bin Cheng <bin.cheng@arm.com>
5420 PR tree-optimization/70754
5421 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
5422 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
5423 combined stmt before it if not NULL.
5424 (combine_chains): Process refs reversely and compute dominance point
5428 2017-02-23 Bin Cheng <bin.cheng@arm.com>
5430 PR tree-optimization/79663
5431 * tree-predcom.c (combine_chains): Process refs in reverse order
5432 only for ZERO length chains, and add explaining comment.
5434 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5436 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
5437 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
5438 * doc/invoke.texi (ARM Options): Add armv8.3-a.
5440 2017-11-15 Tamar Christina <tamar.christina@arm.com>
5442 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
5444 2017-11-15 Martin Liska <mliska@suse.cz>
5446 * tree-cfg.c (pass_warn_function_return::execute):
5447 Compare warn_return_type for greater than zero.
5449 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
5453 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
5454 to return true on Xeon and not on Xeon Phi.
5455 (ix86_check_avx256_register): Changed to ...
5456 (ix86_check_avx_upper_register): ... this. Add extra check for
5457 VALID_AVX512F_REG_OR_XI_MODE.
5458 (ix86_avx_u128_mode_needed): Changed
5459 ix86_check_avx256_register to ix86_check_avx_upper_register.
5460 (ix86_check_avx256_stores): Changed to ...
5461 (ix86_check_avx_upper_stores): ... this. Changed
5462 ix86_check_avx256_register to ix86_check_avx_upper_register.
5463 (ix86_avx_u128_mode_after): Changed
5464 avx_reg256_found to avx_upper_reg_found. Changed
5465 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
5466 (ix86_avx_u128_mode_entry): Changed
5467 ix86_check_avx256_register to ix86_check_avx_upper_register.
5468 (ix86_avx_u128_mode_exit): Ditto.
5469 * config/i386/i386.h: (host_detect_local_cpu): New define.
5471 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
5473 * config/arm/xgene1.md (xgene1): Split into automatons
5474 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
5475 (xgene1_f_load): Adjust reservations and/or types.
5476 (xgene1_f_store): Likewise.
5477 (xgene1_load_pair): Likewise.
5478 (xgene1_store_pair): Likewise.
5479 (xgene1_fp_load1): Likewise.
5480 (xgene1_load1): Likewise.
5481 (xgene1_store1): Likewise.
5482 (xgene1_move): Likewise.
5483 (xgene1_alu): Likewise.
5484 (xgene1_simd): Likewise.
5485 (xgene1_bfm): Likewise.
5486 (xgene1_neon_load1): Likewise.
5487 (xgene1_neon_store1): Likewise.
5488 (xgene1_neon_logic): Likewise.
5489 (xgene1_neon_st1): Likewise.
5490 (xgene1_neon_ld1r): Likewise.
5491 (xgene1_alu_cond): Added.
5492 (xgene1_shift_reg): Likwise.
5493 (xgene1_bfx): Likewise.
5494 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
5496 2017-11-15 Jakub Jelinek <jakub@redhat.com>
5499 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
5501 (can_widen_mult_without_libcall): New function.
5502 (expand_mul_overflow): If only checking unsigned mul overflow,
5503 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
5504 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
5505 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
5506 (expand_DIVMOD): Formatting fix.
5507 * expmed.h (expand_mult): Add NO_LIBCALL argument.
5508 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
5509 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
5511 PR tree-optimization/82977
5512 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
5513 constructed temporary to strlen_to_stridx.put.
5515 2017-11-15 Martin Liska <mliska@suse.cz>
5517 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
5518 * configure: Regenerate.
5520 2017-11-15 Martin Liska <mliska@suse.cz>
5523 * config/sh/sh-mem.cc: Use proper probability for
5526 2017-11-14 Jeff Law <law@redhat.com>
5528 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
5529 the red zone for stack_clash_protection_final_dynamic_probe targets
5530 when the total dynamic stack size is zero bytes.
5532 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
5533 blocks is post order.
5535 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
5537 * dumpfile.h (TDF_COMPARE_DEBUG): New.
5538 * final.c (rest_of_clean_state): Set it for the
5539 -fcompare-debug dump.
5540 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
5541 class when TDF_COMPARE_DEBUG is set.
5543 * dwarf2out.c (gen_producer_string): Discard
5546 2017-11-15 Joseph Myers <joseph@codesourcery.com>
5549 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
5550 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
5551 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
5552 (__TGMATH_REAL_2_3): Remove macros.
5553 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
5554 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
5556 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
5558 (remquo): Define using __TGMATH_REAL_3.
5560 2017-11-14 Jeff Law <law@redhat.com>
5562 * vr-values.c: New file with contents extracted from tree-vrp.c.
5563 * Makefile.in (OBJS): Add vr-values.o
5564 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
5565 (set_value_range, set_and_canonicalize_value_range): Likewise.
5566 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
5567 (value_range_constant_singleton, symbolic_range_p): Likewise.
5568 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
5569 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
5570 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
5571 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
5572 (find_case_label_range, find_case_label_index): Likewise.
5573 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
5574 (range_int_cst_singleton_p, value_inside_range): Likewise.
5575 (get_single_symbol): Likewise.
5576 (switch_update): Move structure definition here.
5577 (to_remove_edges, to_update_switch_stmts): Provide externs.
5578 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
5579 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
5580 (vrp_val_is_min, set_value_range): Likewise.
5581 (set_and_canonicalize_value_range, copy_value_range): Likewise.
5582 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
5583 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
5584 (range_is_nonnull, range_int_cst_p): Likewwise.
5585 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
5586 (get_single_symbol, operand_less_p): Likewise
5587 (compare_values_warnv, compare_values): Likewise.
5588 (value_inside_range, value_range_constant_singleton): Likewise.
5589 (zero_nonzero_bitgs_from_vr): Likewise.
5590 (extract_range_from_binary_expr_1): Likewise.
5591 (overflow_comparison_p): Likewise.
5592 (to_remove_edges, to_update_switch_stmts): Likewise.
5593 (find_case_label-index, find_case_label_range): Likewise.
5594 (switch_update, set_value_range_to_nonnegative): Remove.
5595 (set_value_range_to_truthvalue): Likewise.
5596 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
5597 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
5598 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
5599 (find_case_label_ranges, test_for_singularity): Likewise.
5600 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
5601 (x_vr_values): Move to its remaining use site.
5603 2017-11-10 Jeff Law <law@redhat.com>
5605 * vr-values.h (VR_INITIALIZER): Move #define here.
5606 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
5607 * Makefile.in (OBJS): Add tree-evrp.o
5608 * tree-vrp.h (assert_info): Move structure definition here.
5609 (set_value_range_to_varying): Prototype.
5610 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5611 (infer_value_range, register_edge_assert_for): Likewise.
5612 (stmt_interesting_for_vrp): Likewise.
5613 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
5614 (set_value_range_to_varying): No longer static.
5615 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
5616 (infer_value_range, register_edge_assert_for): Likewise.
5618 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
5620 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
5621 generate the XXBRD instruction.
5623 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
5624 (rs6000_builtin_type_compatible): Treat _Float128 and long double
5625 as being compatible if -mabi=ieeelongdouble.
5626 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
5627 to setup float128 built-ins with hardware support.
5628 (BU_FLOAT128_HW_2): Likewise.
5629 (BU_FLOAT128_HW_3): Likewise.
5630 (BU_FLOAT128_HW_VSX_1): Likewise.
5631 (BU_FLOAT128_HW_VSX_2): Likewise.
5632 (scalar_extract_expq): Change float128 built-in functions to
5633 accommodate having both KFmode and TFmode functions. Use the
5634 KFmode variant as the default.
5635 (scalar_extract_sigq): Likewise.
5636 (scalar_test_neg_qp): Likewise.
5637 (scalar_insert_exp_q): Likewise.
5638 (scalar_insert_exp_qp): Likewise.
5639 (scalar_test_data_class_qp): Likewise.
5640 (sqrtf128_round_to_odd): Delete processing the round to odd
5641 built-in functions as special built-in functions, and define them
5642 as float128 built-ins. Use the KFmode variant as the default.
5643 (truncf128_round_to_odd): Likewise.
5644 (addf128_round_to_odd): Likewise.
5645 (subf128_round_to_odd): Likewise.
5646 (mulf128_round_to_odd): Likewise.
5647 (divf128_round_to_odd): Likewise.
5648 (fmaf128_round_to_odd): Likewise.
5649 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
5650 support for KFmode and TFmode xststdcqp calls.
5651 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
5652 point, switch the built-in handlers for the get/set float128
5653 exponent, get float128 mantissa, float128 test built-ins, and the
5654 float128 round to odd built-in functions. Eliminate creating the
5655 float128 round to odd built-in functions as special built-ins.
5656 (rs6000_init_builtins): Eliminate special creation of the float128
5657 round to odd built-in functions.
5658 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
5659 function insns to support both TFmode and KFmode variants.
5660 (xsxsigqp_<mode>): Likewise.
5661 (xsiexpqpf_<mode>): Likewise.
5662 (xsiexpqp_<mode>): Likewise.
5663 (xststdcqp_<mode>): Likewise.
5664 (xststdcnegqp_<mode>): Likewise.
5665 (xststdcqp_<mode>): Likewise.
5667 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5669 * tree-ssa-threadupdate.c (compute_path_counts): Remove
5670 unused path_in_freq_ptr parameter.
5671 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
5673 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
5675 * ipa-inline.c (edge_badness): Dump sreal frequency.
5676 (compute_inlined_call_time): Match natural implementaiton ...
5677 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
5678 forgotten division by CGRAPH_FREQ_BASE.
5680 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5682 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
5683 Solaris 11. Update comment.
5684 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
5686 * config/sol2.h (STARTFILE_SPEC): Likewise.
5687 * configure: Regenerate.
5689 2017-11-14 Carl Love <cel@us.ibm.com>
5691 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
5692 le_ and be_ prefixes to swap* variables. Remove
5693 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
5696 2017-11-14 Jason Merrill <jason@redhat.com>
5698 Support GTY((cache)) on hash_map.
5699 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
5700 (ggc_cache_remove): Override it instead of ggc_mx.
5701 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
5702 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
5703 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
5704 (simple_cache_map_traits): Override maybe_mx.
5705 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
5706 (hash_map): Friend gt_cleare_cache.
5707 (gt_cleare_cache): New.
5708 * tree.h (tree_cache_traits): New hash_map traits class.
5709 (tree_cache_map): New typedef.
5711 2017-11-14 Richard Biener <rguenther@suse.de>
5713 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
5714 paramter and handling.
5715 (cleanup_control_flow_bb): Likewise.
5716 (cleanup_control_flow_pre): New helper performing a DFS walk
5717 to call cleanup_control_flow_bb in PRE order.
5718 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
5719 via cleanup_control_flow_pre.
5721 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
5723 * config/aarch64/aarch64-simd.md
5724 (aarch64_simd_bsl<mode>_internal): Remove DImode.
5725 (*aarch64_simd_bsl<mode>_alt): Likewise.
5726 (aarch64_simd_bsldi_internal): New.
5727 (aarch64_simd_bsldi_alt): Likewise.
5729 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5731 * tracer.c (better_p): Do not compare frequencies.
5732 * reg-stack.c (better_edge): Likewise.
5733 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
5736 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5738 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
5739 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
5740 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
5741 * lto-streamer-in.c (input_function): Use update_max_bb_count.
5742 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
5743 * predict.c (maybe_hot_frequency_p): Inline to ...
5744 (maybe_hot_count_p): ... here; rewrite to counts.
5745 (counts_to_freqs): Rename to ...
5746 (update_max_bb_count): ... this one.
5747 (expensive_function_p): Use counts.
5748 (estimate_bb_frequencies): Update.
5749 (rebuild_frequencies): Update.
5750 * predict.h (counts_to_freqs): Rename to ...
5751 (update_max_bb_count): ... this one.
5752 * profile.c (compute_branch_probabilities): Add debug info
5753 * tree-inline.c (expand_call_inline): Update debug info.
5754 (optimize_inline_calls): Use update_max_bb_count..
5755 (tree_function_versioning): Use update_max_bb_count..
5756 * value-prof.c (gimple_value_profile_transformations):
5757 Do not use update_max_bb_count.
5759 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5761 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
5762 always use frequencies.
5764 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5766 * bb-reorder.c: Remove frequencies from comments.
5767 (better_edge_p): Use profile counts.
5768 (find_traces): Dump profile counts.
5769 (rotate_loop): Use profile counts.
5770 (find_traces_1_round): Likewise.
5771 (connect_better_edge_p): Use counts instead of probabilities for
5773 (copy_bb_p): Drop early check for non-0 frequency.
5774 (sanitize_hot_paths): Update comments.
5776 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5778 * ipa-split.c (struct split_point): Add count.
5779 (consider_split): Do not compute incoming frequency; compute incoming
5780 count and store it to split_point.
5781 (split_function): Set count of the call to split part correctly.
5783 2017-11-13 Carl Love <cel@us.ibm.com>
5785 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
5787 2017-11-13 Tom Tromey <tom@tromey.com>
5789 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
5791 2017-11-13 Carl Love <cel@us.ibm.com>
5793 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
5794 Add support for builtins:
5795 unsigned int vec_first_{,miss}_match_{,or_eos}index,
5796 vector {un,}signed {char,int,short},
5797 vector {un,}signed {char,int,short}) arguments.
5798 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
5799 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
5800 Add BU_P9V_AV_2 expansions for the builtins.
5801 * config/rs6000/altivec.h (vec_first_match_index,
5802 vec_first_mismatch_index, vec_first_match_or_eos_index,
5803 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
5804 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
5805 new extern declaration.
5806 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
5807 * config/rs6000/vsx.md (first_match_index_<mode>,
5808 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
5809 first_mismatch_or_eos_index_<mode>): Add define expand.
5810 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
5811 * doc/extend.texi: Update the built-in documenation file for the new
5814 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
5816 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
5817 into the min/max operations for _Float<N> and _Float<N>X types.
5819 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5822 * dwarf2out.c (do_eh_frame): New static variable.
5823 (dwarf2out_begin_prologue): Set it.
5824 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
5826 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5828 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
5830 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
5832 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5835 * debug.h (dwarf2out_do_eh_frame): Declare.
5836 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
5837 (dwarf2out_do_frame): Use it.
5838 (dwarf2out_do_cfi_asm): Likewise.
5839 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
5840 (dwarf2out_assembly_start): Likewise.
5841 (dwarf2out_begin_prologue): Fix comment.
5842 * toplev.c (compile_file): Always call dwarf2out_frame_finish
5843 if the target needs either debug or unwind DWARF2 info.
5844 * lto-opts.c (lto_write_options): Do not save -fexceptions,
5845 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
5846 -ftrapping-math, -ftrapv and -fwrapv.
5848 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5850 * cgraph.c (cgraph_edge::sreal_frequency): New function.
5851 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
5852 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
5853 (estimate_edge_size_and_time): Likewise.
5854 (ipa_merge_fn_summary_after_inlining): Likewise.
5855 * ipa-inline.c (cgraph_freq_base_rec): Remove.
5856 (compute_uninlined_call_time): Use sreal_frequency.
5857 (compute_inlined_call_time): Likewise.
5858 (ipa_inline): Do not initialize cgraph_freq_base_rec.
5859 * profile-count.c: Include sreal.h.
5860 (profile_count::to_sreal_scale): New.
5861 * profile-count.h: Forward declare sreal.
5862 (profile_count::to_sreal_scale): Declare.
5864 2017-11-13 Nathan Sidwell <nathan@acm.org>
5866 * diagnostic.c (maybe_line_and_column): New.
5867 (diagnostic_get_location_text): Use it.
5868 (diagnostic_report_current_module): Likewise.
5869 (test_diagnostic_get_location_text): Add tests.
5871 2017-11-13 Luis Machado <luis.machado@linaro.org>
5873 * doc/md.texi (Specifying processor pipeline description): Fix
5874 incorrect latency for the div instruction example.
5876 2017-11-13 Jakub Jelinek <jakub@redhat.com>
5878 PR tree-optimization/78821
5879 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
5880 that bit_not_p is the same.
5881 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
5882 (split_group): Count precisely bit_not_p bits in each statement.
5883 (invert_op): New function.
5884 (imm_store_chain_info::output_merged_store): Use invert_op to
5885 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
5886 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
5888 2017-11-13 Martin Liska <mliska@suse.cz>
5890 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
5891 (struct source_info): Likewise.
5892 (add_branch_counts): Likewise.
5893 (add_line_counts): Likewise.
5894 (function_summary): Likewise.
5895 (output_intermediate_line): Likewise.
5896 (generate_results): Likewise.
5898 2017-11-13 Martin Liska <mliska@suse.cz>
5900 * gcov.c (struct block_info): Remove typedef for block_t.
5901 (struct line_info): Likewise.
5902 (line_info::has_block): Likewise.
5903 (EXIT_BLOCK): Likewise.
5904 (unblock): Likewise.
5905 (circuit): Likewise.
5906 (get_cycles_count): Likewise.
5907 (process_file): Likewise.
5908 (read_graph_file): Likewise.
5909 (solve_flow_graph): Likewise.
5910 (find_exception_blocks): Likewise.
5911 (add_line_counts): Likewise.
5912 (accumulate_line_info): Likewise.
5913 (output_line_details): Likewise.
5915 2017-11-13 Martin Liska <mliska@suse.cz>
5917 * gcov.c (struct arc_info): Remove typedef for arc_t.
5918 (struct line_info): Likewise.
5919 (add_branch_counts): Likewise.
5920 (output_branch_count): Likewise.
5921 (function_info::~function_info): Likewise.
5922 (circuit): Likewise.
5923 (output_intermediate_line): Likewise.
5924 (read_graph_file): Likewise.
5925 (solve_flow_graph): Likewise.
5926 (find_exception_blocks): Likewise.
5927 (add_line_counts): Likewise.
5928 (accumulate_line_info): Likewise.
5929 (output_line_details): Likewise.
5930 (output_function_details): Likewise.
5932 2017-11-13 Martin Liska <mliska@suse.cz>
5934 * gcov.c (struct function_info): Remove typedef for function_t.
5935 (struct source_info): Likewise.
5936 (source_info::get_functions_at_location): Likewise.
5937 (solve_flow_graph): Likewise.
5938 (find_exception_blocks): Likewise.
5939 (add_line_counts): Likewise.
5940 (output_intermediate_file): Likewise.
5941 (process_file): Likewise.
5942 (generate_results): Likewise.
5943 (release_structures): Likewise.
5944 (read_graph_file): Likewise.
5945 (read_count_file): Likewise.
5946 (accumulate_line_counts): Likewise.
5947 (output_lines): Likewise.
5949 2017-11-13 Martin Liska <mliska@suse.cz>
5951 * gcov.c (function_info::function_info): Remove num_counts
5952 and add vector<gcov_type>.
5953 (function_info::~function_info): Use the vector.
5954 (process_file): Likewise.
5955 (read_graph_file): Likewise.
5956 (read_count_file): Likewise.
5957 (solve_flow_graph): Likewise.
5959 2017-11-13 Martin Liska <mliska@suse.cz>
5961 * gcov.c (function_info::is_artificial): New function.
5962 (process_file): Erase all artificial early.
5963 (generate_results): Skip as all artificial are already
5966 2017-11-13 Martin Liska <mliska@suse.cz>
5968 * gcov.c (read_graph_file): Store to global vector of functions.
5969 (read_count_file): Iterate the vector.
5970 (process_file): Likewise.
5971 (generate_results): Likewise.
5972 (release_structures): Likewise.
5974 2017-11-13 Jakub Jelinek <jakub@redhat.com>
5976 PR tree-optimization/82954
5977 * gimple-ssa-store-merging.c
5978 (imm_store_chain_info::coalesce_immediate_stores): If
5979 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
5981 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
5983 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
5984 Upddate call to ENDIAN_LANE_N.
5985 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
5986 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
5987 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
5988 and use aarch64_endian_lane_rtx.
5989 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
5991 2017-11-12 Tom de Vries <tom@codesourcery.com>
5993 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
5995 2017-11-12 Tom de Vries <tom@codesourcery.com>
5997 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
5999 2017-11-12 Tom de Vries <tom@codesourcery.com>
6001 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
6003 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6004 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
6005 * defaults.h (ASM_OUTPUT_LABELREF): Same.
6007 2017-11-11 Martin Sebor <msebor@redhat.com>
6010 * doc/extend.texi (attribute nonstring): Remove spurious argument.
6013 * prefic.c (translate_name): Replace strncpy with memcpy to
6014 avoid -Wstringop-truncation.
6016 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6018 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
6020 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6022 * predict.c (maybe_hot_frequency_p): Do not use cfun.
6024 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6026 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
6029 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
6031 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
6032 enable generating XXBRH if the value is in a vector register.
6033 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
6034 value is in a vector register.
6035 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
6036 register to register bswap64's instead of doing the GPR sequence
6037 used on previous machines.
6038 (bswapdi2_xxbrd): New insn.
6039 (bswapdi2_reg): Disallow on ISA 3.0.
6040 (register to register bswap64 splitter): Do not split the insn on
6041 ISA 3.0 systems that use XXBRD.
6043 2017-11-10 Martin Sebor <msebor@redhat.com>
6046 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
6048 (find_subframework_file): Same.
6050 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
6052 * auto-profile.c (afdo_indirect_call): Drop frequency.
6053 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
6054 (cgraph_node::create_edge): Drop frequency argument.
6055 (cgraph_node::create_indirect_edge): Drop frequency argument.
6056 (cgraph_edge::make_speculative): Drop frequency arguments.
6057 (cgraph_edge::resolve_speculation): Do not update frequencies
6058 (cgraph_edge::dump_edge_flags): Do not dump frequency.
6059 (cgraph_node::dump): Check consistency in IPA mode.
6060 (cgraph_edge::maybe_hot_p): Use IPA counter.
6061 (cgraph_edge::verify_count_and_frequency): Rename to ...
6062 (cgraph_edge::verify_count): ... this one; drop frequency checking.
6063 (cgraph_node::verify_node): Update.
6064 * cgraph.h (struct cgraph_edge): Drop frequency.
6065 (cgraph_edge::frequency): New function.
6066 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
6068 (cgraph_edge::rebuild_edges): Likewise.
6069 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
6070 (duplicate_thunk_for_node): Do not pass frequency.
6071 (cgraph_node::create_clone): Scale only counts.
6072 (cgraph_node::create_virtual_clone): Do not pass frequency.
6073 (cgraph_node::create_edge_including_clones): Do not pass frequency.
6074 (cgraph_node::create_version_clone): Do not pass frequency.
6075 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
6076 (cgraph_node::expand_thunk): Do not pass frequency.
6077 (cgraph_node::create_wrapper): Do not pass frequency.
6078 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
6080 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
6081 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
6082 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
6083 (gather_caller_stats): Use frequency function.
6084 (ipcp_cloning_candidate_p): Use frequency function.
6085 (ipcp_propagate_stage): Use frequency function.
6086 (get_info_about_necessary_edges): Use frequency function.
6087 (update_profiling_info): Update only IPA profile.
6088 (update_specialized_profile): Use frequency functoin.
6089 (perhaps_add_new_callers): Update only IPA profile.
6090 * ipa-devirt.c (ipa_devirt): Use IPA profile.
6091 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
6092 (dump_ipa_call_summary): Use frequency function.
6093 (estimate_edge_size_and_time): Use frequency function.
6094 (ipa_merge_fn_summary_after_inlining): Use frequency function.
6095 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
6096 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
6097 (update_noncloned_counts): ... ths one; scale counts only.
6098 (clone_inlined_nodes): Do not scale frequency.
6099 (inline_call): Do not pass frequency.
6100 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
6101 (compute_inlined_call_time): Use IPA profile.
6102 (want_inline_small_function_p): Use IPA profile.
6103 (want_inline_self_recursive_call_p): Use IPA profile.
6104 (edge_badness): Use IPA profile.
6105 (lookup_recursive_calls): Use IPA profile.
6106 (recursive_inlining): Do not pass frequency.
6107 (resolve_noninline_speculation): Do not update frequency.
6108 (inline_small_functions): Collect max of IPA profile.
6109 (dump_overall_stats): Dump IPA porfile.
6110 (dump_inline_stats): Dump IPA porfile.
6111 (ipa_inline): Collect IPA stats.
6112 * ipa-inline.h (clone_inlined_nodes): Update prototype.
6113 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
6114 (ipa_propagate_frequency): Use frequency function.
6115 (ipa_profile): Cleanup.
6116 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
6117 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
6118 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
6119 (input_node): Do not stream frequency.
6120 (input_edge): Do not stream frequency.
6121 (merge_profile_summaries): Scale only IPA profiles.
6122 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
6123 * predict.c (drop_profile): Do not recompute frequency.
6124 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
6125 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
6126 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
6127 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
6128 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
6129 * tree-inline.c (copy_bb): Do not scale frequency.
6130 (expand_call_inline): Do not scale frequency.
6131 (tree_function_versioning): Do not scale frequency.
6132 * ubsan.c (ubsan_create_edge): Do not pass frequency.
6134 2017-11-10 Julia Koval <julia.koval@intel.com>
6136 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
6137 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
6138 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
6139 (_mm256_maskz_gf2p8affine_epi64_epi8)
6140 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
6141 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
6142 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
6143 (__builtin_ia32_vgf2p8affineqb_v32qi)
6144 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
6145 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
6147 2017-11-10 Tamar Christina <tamar.christina@arm.com>
6151 (arm_option_override): Refactor.
6152 (arm_option_reconfigure_globals): New.
6153 (arm_options_perform_arch_sanity_checks): New.
6154 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
6156 (arm_options_perform_arch_sanity_checks): Likewise
6158 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
6160 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
6161 (power9-qpmul): New.
6162 * rs6000/rs6000.md ("type" attr): Add qmul.
6163 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
6164 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
6165 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
6167 2017-11-10 Martin Sebor <msebor@redhat.com>
6170 * builtins.c (compute_objsize): Handle arrays that
6171 compute_builtin_object_size likes to fail for. Make extern.
6172 * builtins.h (compute_objsize): Declare.
6173 (check_strncpy_sizes): New function.
6174 (expand_builtin_strncpy): Call check_strncpy_sizes.
6175 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
6176 -Wstringop-truncation.
6177 (gimple_fold_builtin_strncat): Same.
6178 * gimple.c (gimple_build_call_from_tree): Set call location.
6179 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
6180 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
6181 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
6182 (handle_builtin_strlen): Use strlen_to_stridx.
6183 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
6185 Use strlen_to_stridx.
6186 (pass_strlen::execute): Release strlen_to_stridx.
6187 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
6188 (-Wstringop-truncation): Document new option.
6190 2017-11-10 Martin Liska <mliska@suse.cz>
6192 PR gcov-profile/82702
6193 * gcov.c (main): Handle intermediate files in a different
6195 (get_gcov_intermediate_filename): New function.
6196 (output_gcov_file): Remove support of intermediate files.
6197 (generate_results): Allocate intermediate file.
6198 (release_structures): Clean-up properly fn_end.
6199 (output_intermediate_file): Start iterating with line 1.
6201 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6203 PR tree-optimization/82929
6204 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6205 ops_swapped_p non-static data member.
6206 (store_immediate_info::store_immediate_info): Clear it.
6207 (imm_store_chain_info::coalesce_immediate_stores): If swapping
6208 ops set ops_swapped_p.
6209 (count_multiple_uses): Handle ops_swapped_p.
6211 2017-11-10 Martin Liska <mliska@suse.cz>
6213 * coverage.c (coverage_init): Stream information about
6214 support of has_unexecuted_blocks.
6215 * doc/gcov.texi: Document that.
6216 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
6217 * gcov.c (read_graph_file): Likewise.
6218 (output_line_beginning): Fix a small issue with
6221 2017-11-10 Bin Cheng <bin.cheng@arm.com>
6223 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
6224 reference of trivial component.
6226 2017-11-10 Jakub Jelinek <jakub@redhat.com>
6229 * gimple-ssa-store-merging.c
6230 (pass_store_merging::terminate_all_aliasing_chains): For
6231 gimple_store_p stmts also call refs_output_dependent_p.
6233 PR rtl-optimization/82913
6234 * compare-elim.c (try_merge_compare): Punt if def_insn is not
6237 2017-11-09 Jeff Law <law@redhat.com>
6239 * vr-values.h: New file with vr_values class.
6240 * tree-vrp.c: Include vr-values.h
6241 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
6242 data objects into the vr_values class.
6243 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
6244 (get_value_range): Make it a member function within vr_values class.
6245 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
6246 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
6247 (op_with_constant_singleton_value_range): Likewise.
6248 (extract_range_for_var_from_comparison_expr): Likewise.
6249 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
6250 (extract_range_from_binary_expr): Likewise.
6251 (extract_range_from_unary_expr): Likewise.
6252 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
6253 (check_for_binary_op_overflow, extract_range_basic): Likewise.
6254 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
6255 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
6256 (compare_name_with_value, compare_names): Likewise.
6257 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
6258 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
6259 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
6260 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
6261 (extract_range_from_phi_node): Likewise.
6262 (simplify_truth_ops_using_ranges): Likewise.
6263 (simplify_div_or_mod_using_ranges): Likewise.
6264 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
6265 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
6266 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
6267 (simplify_float_conversion_using_ranges): Likewise.
6268 (simplify_internal_call_using_ranges): Likewise.
6269 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
6270 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
6271 poitner via x_vr_values for calls into gimple folder.
6272 (vrp_initialize_lattice): Make this the vr_values ctor.
6273 (vrp_free_lattice): Make this the vr_values dtor.
6274 (set_vr_value): New function.
6275 (class vrp_prop): Add vr_values data member. Add various member
6276 functions as well as member functions that delegate to vr_values.
6277 (check_array_ref): Make a member function within vrp_prop class.
6278 (search_for_addr_array, vrp_initialize): Likewise.
6279 (vrp_finalize): Likewise. Revamp to avoid direct access to
6280 vr_value, values_propagated, etc.
6281 (check_array_bounds): Extract vrp_prop class instance pointer from
6282 walk info structure. Use it to call member functions.
6283 (check_all_array_refs): Make a member function within vrp_prop class.
6284 Smuggle class instance pointer via walk info structure.
6285 (x_vr_values): New local static.
6286 (vrp_valueize): Use x_vr_values to get class instance.
6287 (vr_valueize_1): Likewise.
6288 (class vrp_folder): Add vr_values data member. Add various member
6289 functions as well as member functions that delegate to vr_values.
6290 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
6291 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
6292 class instance from vr_values. Use it to call member functions.
6293 (vrp_dom_walker): Add vr_values data member.
6294 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
6295 instance via x_vr_values.
6296 (identify_jump_threads): Accept vr_values as argument. Store
6297 it into the walker structure.
6298 (evrp_dom_walker): Add vr_values class data member. Add various
6300 (evrp_dom_walker::try_find_new_range): Use vr_values data
6301 member to access the memory allocator.
6302 (evrp_dom_walker::before_dom_children): Store vr_values class
6303 instance into the vrp_folder class.
6304 (evrp_dom_walker::push_value_range): Rework to avoid direct
6305 access to num_vr_values and vr_value.
6306 (evrp_dom_walker::pop_value_range): Likewise.
6307 (execute_early_vrp): Remove call to vrp_initialize_lattice.
6308 Use vr_values to get to dump_all_value_ranges member function.
6309 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
6310 and simplify_cond_using_ranges_2 via vrp_prop class instance.
6311 Pass vr_values class instance down to identify_jump_threads.
6312 Remove call to vrp_free_lattice.
6313 (debug_all_value_ranges): Remove.
6315 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
6316 (vrp_folder): Likewise.
6318 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
6319 Get it from the existing bitmap instead.
6320 (vrp_intersect_ranges_1): Likewise.
6322 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6324 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
6326 (store_immediate_info::store_immediate_info): Add bitnotp argument,
6327 set bit_not_p to it.
6328 (imm_store_chain_info::coalesce_immediate_stores): Break group
6329 if bit_not_p is different.
6330 (count_multiple_uses, split_group,
6331 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
6332 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
6333 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
6334 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
6335 to store_immediate_info ctor.
6337 2017-11-09 Jim Wilson <jimw@sifive.com>
6339 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
6340 (scan_prog_file): Likewise.
6342 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6344 * bb-reorder.c (max_entry_frequency): Remove.
6345 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
6346 connect_traces, push_to_next_round_p): Remove prototypes.
6347 (find_traces_1_round): Use counts only.
6348 (push_to_next_round_p): Likewise.
6349 (find_traces): Likewise.
6350 (rotate_loop): Likewise.
6351 (find_traces_1_round): Likewise.
6352 (connect_traces): Likewise.
6353 (edge_order): Likewise.
6355 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
6357 * config/arm/arm.c (output_return_instruction): Add comments to
6358 indicate requirement for cmse_nonsecure_entry return to account
6359 for the size of clearing instruction output here.
6360 (thumb_exit): Likewise.
6361 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
6362 return in hardfloat mode.
6364 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
6366 * config/rs6000/rs6000.c (machine_function): Add a bool,
6367 "toc_is_wrapped_separately".
6368 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
6369 if it wasn't explicitly set or unset, we are optimizing for speed, and
6370 doing separate shrink-wrapping.
6371 (rs6000_get_separate_components): Enable the TOC component if
6372 saving the TOC register in the prologue.
6373 (rs6000_components_for_bb): Handle the TOC component.
6374 (rs6000_emit_prologue_components): Store the TOC register where needed.
6375 (rs6000_set_handled_components): Mark TOC as handled, if handled.
6376 (rs6000_emit_prologue): Don't save the TOC if that is already done.
6378 2017-11-09 Martin Jambor <mjambor@suse.cz>
6380 * ipa-param-manipulation.c: New file.
6381 * ipa-param-manipulation.h: Likewise.
6382 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
6383 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
6384 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
6385 (ipa_parm_adjustment): Likewise.
6386 (ipa_parm_adjustment_vec): Likewise.
6387 (ipa_get_vector_of_formal_parms): Moved declaration to
6388 ipa-param-manipulation.h.
6389 (ipa_get_vector_of_formal_parm_types): Likewise.
6390 (ipa_modify_formal_parameters): Likewise.
6391 (ipa_modify_call_arguments): Likewise.
6392 (ipa_combine_adjustments): Likewise.
6393 (ipa_dump_param_adjustments): Likewise.
6394 (ipa_modify_expr): Likewise.
6395 (ipa_get_adjustment_candidate): Likewise.
6396 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
6397 ipa-param-manipulation.c.
6398 (ipa_get_vector_of_formal_parm_types): Likewise.
6399 (ipa_modify_formal_parameters): Likewise.
6400 (ipa_modify_call_arguments): Likewise.
6401 (ipa_modify_expr): Likewise.
6402 (get_ssa_base_param): Likewise.
6403 (ipa_get_adjustment_candidate): Likewise.
6404 (index_in_adjustments_multiple_times_p): Likewise.
6405 (ipa_combine_adjustments): Likewise.
6406 (ipa_dump_param_adjustments): Likewise.
6407 * tree-sra.c: Also include ipa-param-manipulation.h
6408 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
6411 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6412 Alan Hayward <alan.hayward@arm.com>
6413 David Sherwood <david.sherwood@arm.com>
6415 * doc/sourcebuild.texi (vect_masked_store): Document.
6417 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6418 Alan Hayward <alan.hayward@arm.com>
6419 David Sherwood <david.sherwood@arm.com>
6421 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
6423 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6424 Alan Hayward <alan.hayward@arm.com>
6425 David Sherwood <david.sherwood@arm.com>
6427 * doc/sourcebuild.texi (vect_variable_length): Document.
6429 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6430 Alan Hayward <alan.hayward@arm.com>
6431 David Sherwood <david.sherwood@arm.com>
6433 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
6435 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6436 Alan Hayward <alan.hayward@arm.com>
6437 David Sherwood <david.sherwood@arm.com>
6439 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
6441 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6442 Alan Hayward <alan.hayward@arm.com>
6443 David Sherwood <david.sherwood@arm.com>
6445 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
6446 previously undocumented selectors.
6447 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
6449 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6451 * doc/rtl.texi (const_vector): Say that elements can be
6452 const_wide_ints too.
6453 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
6454 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
6455 (gen_vec_duplicate): Use it instead of CONSTANT_P.
6456 * optabs.c (expand_vector_broadcast): Likewise.
6458 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6459 Alan Hayward <alan.hayward@arm.com>
6460 David Sherwood <david.sherwood@arm.com>
6462 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
6463 scaled index even if the unscaled address was invalid.
6464 Don't increase the complexity of using a scale in that case.
6466 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
6467 Alan Hayward <alan.hayward@arm.com>
6468 David Sherwood <david.sherwood@arm.com>
6470 * doc/rtl.texi: Rewrite the subreg rules so that they partition
6471 the inner register into REGMODE_NATURAL_SIZE bytes rather than
6472 UNITS_PER_WORD bytes.
6473 * emit-rtl.c (validate_subreg): Divide subregs into blocks
6474 based on REGMODE_NATURAL_SIZE of the inner mode.
6475 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
6476 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
6477 * expmed.c (lowpart_bit_field_p): Divide the value up into
6478 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
6479 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
6480 whether something is likely to occupy more than one register.
6482 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
6485 * ipa-inline-transform.c (update_noncloned_frequencies): Use
6486 profile_count::adjust_for_ipa_scaling.
6487 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
6488 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
6490 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
6492 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6494 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
6495 (split_group): Add total_orig and total_new arguments, estimate the
6496 number of statements related to the store group without store merging
6497 and with store merging.
6498 (imm_store_chain_info::output_merged_store): Adjust split_group
6499 callers, punt if estimated number of statements with store merging
6500 is not smaller than estimated number of statements without it.
6502 (handled_load): Remove has_single_use checks.
6503 (pass_store_merging::process_store): Likewise.
6505 2017-11-09 Richard Biener <rguenther@suse.de>
6507 PR tree-optimization/82902
6508 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
6510 2017-11-09 Martin Liska <mliska@suse.cz>
6513 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
6516 2017-11-09 Martin Liska <mliska@suse.cz>
6518 PR tree-optimization/82669
6519 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
6521 2017-11-09 Martin Liska <mliska@suse.cz>
6523 PR gcov-profile/48463
6524 * coverage.c (coverage_begin_function): Output also end locus
6525 of a function and information whether the function is
6527 * gcov-dump.c (tag_function): Parse and print the information.
6528 * gcov.c (INCLUDE_MAP): Add include.
6529 (INCLUDE_SET): Likewise.
6530 (struct line_info): Move earlier in the source file because
6531 of vector<line_info> in function_info structure.
6532 (line_info::line_info): Likewise.
6533 (line_info::has_block): Likewise.
6534 (struct source_info): Add new member index.
6535 (source_info::get_functions_at_location): New function.
6536 (function_info::group_line_p): New function.
6537 (output_intermediate_line): New function.
6538 (output_intermediate_file): Use the mentioned function.
6539 (struct function_start): New.
6540 (struct function_start_pair_hash): Likewise.
6541 (process_file): Add code that identifies group functions.
6542 Assign lines either to global or function scope.
6543 (generate_results): Skip artificial functions.
6544 (find_source): Assign index for each source file.
6545 (read_graph_file): Read new flag artificial and end_line.
6546 (add_line_counts): Assign it either to global of function scope.
6547 (accumulate_line_counts): Isolate core of the function to
6548 accumulate_line_info and call it for both function and global
6550 (accumulate_line_info): New function.
6551 (output_line_beginning): Fix GNU coding style.
6552 (print_source_line): New function.
6553 (output_line_details): Likewise.
6554 (output_function_details): Likewise.
6555 (output_lines): Iterate both source (global) scope and function
6557 (struct function_line_start_cmp): New class.
6558 * doc/gcov.texi: Reflect changes in documentation.
6560 2017-11-09 Jakub Jelinek <jakub@redhat.com>
6563 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
6564 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
6565 and similarly for not instead of neg.
6567 2017-11-08 Andi Kleen <ak@linux.intel.com>
6569 * config/i386/i386.opt: Add -mforce-indirect-call.
6570 * config/i386/predicates.md: Check for flag_force_indirect_call.
6571 * doc/invoke.texi: Document -mforce-indirect-call
6573 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
6575 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
6577 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
6578 config/riscv/riscv.c (predict.h): New include.
6579 (riscv_slow_unaligned_access_p): No longer static.
6580 (riscv_block_move_straight): Add require.
6581 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
6583 2017-11-08 Jakub Jelinek <jakub@redhat.com>
6586 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
6587 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
6588 nonimmediate_operand predicate for operand 1 instead of
6591 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6593 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
6595 * config/aarch64/constraints.md (Uml): New constraint.
6596 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
6599 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6601 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
6602 of two vec_duplicates into a vec_concat.
6604 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6606 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6607 Simplify vec_merge of vec_duplicate and vec_concat.
6608 * config/aarch64/constraints.md (Utq): New constraint.
6609 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
6612 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6614 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
6615 Simplify vec_merge of vec_duplicate and const_vector.
6616 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
6618 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
6619 mode iterator. Update predicate on operand 1 to
6620 handle non-const_vec constants. Delete constraints.
6621 (*aarch64_combinez_be<mode>): Likewise for operand 2.
6623 2017-11-08 Jakub Jelinek <jakub@redhat.com>
6625 PR tree-optimization/78821
6626 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
6628 (store_operand_info::store_operand_info): Initialize it to false.
6629 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
6630 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
6631 store in the group, and if chain_info is non-NULL, to ignore altogether
6633 (compatible_load_p): Fail if bit_not_p does not match.
6634 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
6635 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
6636 (pass_store_merging::process_store): Adjust
6637 terminate_all_aliasing_chains calls to pass NULL in all current spots,
6638 call terminate_all_aliasing_chains newly when adding a store into
6639 a chain with non-NULL chain_info.
6641 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
6643 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
6645 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
6647 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
6649 (aarch64_layout_frame): Initialise emit_frame_chain.
6650 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
6651 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
6653 2017-11-08 Martin Liska <mliska@suse.cz>
6655 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
6656 of gimple_call_internal_p.
6658 2017-11-07 Tom de Vries <tom@codesourcery.com>
6660 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
6662 2017-11-07 Tom de Vries <tom@codesourcery.com>
6664 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6667 2017-11-08 Martin Liska <mliska@suse.cz>
6670 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
6672 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
6674 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
6676 2017-11-07 Jakub Jelinek <jakub@redhat.com>
6679 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
6680 (*cmp<mode>_ccz_1): New insn with $k alternative.
6683 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
6684 RTX_COMM_COMPARE as commutative as well.
6685 (ix86_binary_operator_ok): Formatting fix.
6686 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
6687 *<code><mode>3<mask_name><round_saeonly_name>,
6688 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
6689 *<s>mul<mode>3_highpart<mask_name>,
6690 *vec_widen_umult_even_v16si<mask_name>,
6691 *vec_widen_umult_even_v8si<mask_name>,
6692 *vec_widen_umult_even_v4si<mask_name>,
6693 *vec_widen_smult_even_v16si<mask_name>,
6694 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
6695 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
6696 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
6697 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
6698 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
6699 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
6700 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
6701 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
6702 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
6703 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
6704 ix86_binary_operator_ok. Formatting fixes.
6705 (*<plusminus_insn><mode>3<mask_name><round_name>,
6706 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
6709 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
6711 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
6712 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
6713 eq and ne if TARGET_ISEL.
6714 (cmp): New code_iterator.
6715 (UNS, UNSU_, UNSIK): New code_attrs.
6716 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
6717 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
6719 ("eq<mode>3"): ... this.
6720 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
6722 ("ne<mode>3"): ... this.
6724 2017-11-07 Julia Koval <julia.koval@intel.com>
6727 * common/config/i386/i386-common.c
6728 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
6729 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
6730 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
6731 * config/i386/i386.opt: Ditto.
6732 * config/i386/i386.c (ix86_target_string): Ditto.
6733 (ix86_option_override_internal): Ditto.
6734 (ix86_init_mpx_builtins): Move MPX to args2.
6735 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
6736 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
6737 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
6738 __builtin_ia32_vgf2p8affineinvqb_v32qi,
6739 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
6740 __builtin_ia32_vgf2p8affineinvqb_v16qi,
6741 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
6743 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
6746 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
6747 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
6748 (zero-extendsidi peephole2): Remove peephole.
6750 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
6753 * stor-layout.c: Include attribs.h.
6754 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
6755 explicit lookup of "aligned" attribute.
6757 2017-11-07 Andrew Waterman <andrew@sifive.com>
6759 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
6760 (riscv_expand_block_move): Likewise.
6761 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
6763 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
6764 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
6765 * config/riscv/riscv.c (riscv_block_move_straight): New function.
6766 (riscv_adjust_block_mem): Likewise.
6767 (riscv_block_move_loop): Likewise.
6768 (riscv_expand_block_move): Likewise.
6769 * config/riscv/riscv.md (movmemsi): New pattern.
6771 2017-11-07 Michael Clark <michaeljclark@mac.com>
6773 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
6774 (MUSL_DYNAMIC_LINKER): Likewise.
6776 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
6778 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
6779 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
6782 2017-11-07 Richard Biener <rguenther@suse.de>
6784 * match.pd: Fix build.
6786 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
6787 Jackson Woodruff <jackson.woodruff@arm.com>
6789 PR tree-optimization/71026
6790 * match.pd: Canonicalize negate in division.
6792 2017-11-07 Sudakshina Das <sudi.das@arm.com>
6795 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
6797 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6799 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
6800 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
6802 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6804 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
6805 non-scalar integral types.
6806 * match.pd (negate_expr_p): Handle MINUS_EXPR.
6807 (-(A-B), -(~A)): New transformations.
6809 2017-11-07 Tom de Vries <tom@codesourcery.com>
6811 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
6812 semicolon after "do {} while (0)".
6813 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6814 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6815 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6816 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6817 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6819 2017-11-07 Tom de Vries <tom@codesourcery.com>
6821 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
6822 after "do {} while (0)".
6823 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6824 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6825 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6826 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6827 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6829 2017-11-07 Tom de Vries <tom@codesourcery.com>
6832 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
6834 (arm_rtx_costs_internal): Add missing semicolon after
6835 HANDLE_NARROW_SHIFT_ARITH call.
6837 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
6839 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6840 disable isel if it was not set explicitly.
6842 2017-11-06 James Bowman <james.bowman@ftdichip.com>
6844 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
6845 (add_type_attribute) likewise.
6847 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
6849 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
6850 of struct ix86_frame.
6851 (ix86_initial_elimination_offset): Likewise.
6852 (ix86_expand_split_stack_prologue): Likewise.
6854 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
6856 * tree-vrp.h (enum value_range_type): Update stale comment.
6858 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6859 Alan Hayward <alan.hayward@arm.com>
6860 David Sherwood <david.sherwood@arm.com>
6862 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
6863 (aarch64_expand_vec_perm_const): Take the number of units too.
6864 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
6865 (aarch64_expand_vec_perm_const): Likewise.
6866 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
6867 (vec_perm<mode>): Update accordingly.
6869 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6870 Alan Hayward <alan.hayward@arm.com>
6871 David Sherwood <david.sherwood@arm.com>
6873 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
6874 Take the number of units too.
6875 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
6876 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
6877 but check for a vector mode before rather than after the call.
6878 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
6879 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
6880 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
6881 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
6882 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
6883 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
6884 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
6885 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
6886 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
6887 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
6888 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
6889 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
6890 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
6891 (aarch64_sqdmull2_n<mode>): Update accordingly.
6893 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6894 Alan Hayward <alan.hayward@arm.com>
6895 David Sherwood <david.sherwood@arm.com>
6897 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
6898 the number of units too.
6899 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
6900 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
6901 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
6902 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
6903 (vec_store_lanesxi<mode>): Update accordingly.
6905 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6906 Alan Hayward <alan.hayward@arm.com>
6907 David Sherwood <david.sherwood@arm.com>
6909 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
6910 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
6911 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
6912 of units rather than the mode.
6913 * config/aarch64/iterators.md (nunits): New mode attribute.
6914 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
6915 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
6916 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
6917 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
6918 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
6919 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
6920 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
6921 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
6922 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
6923 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
6924 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
6925 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
6926 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
6927 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
6928 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
6929 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
6930 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
6931 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
6932 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
6933 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
6934 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
6935 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
6936 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
6937 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
6938 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
6939 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
6940 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
6941 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
6942 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
6943 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
6944 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
6945 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
6946 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
6947 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
6948 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
6949 (aarch64_simd_vec_setv2di): Likewise.
6951 2017-11-06 Carl Love <cel@us.ibm.com>
6953 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
6955 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
6956 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
6958 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
6960 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
6961 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
6962 Add power 8 macro expansions.
6963 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
6964 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
6965 power 8 instructions. (VSX_XXBR): Add iterator.
6967 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
6969 * config/arm/arm.md (predicable_short_it): Change default to "no",
6970 improve documentation, remove uses that are identical to the default.
6971 (enabled_for_depr_it): Rename to enabled_for_short_it.
6972 * gcc/config/arm/arm-fixed.md (predicable_short_it):
6973 Remove default uses.
6974 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
6975 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
6976 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
6977 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
6979 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
6982 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
6983 float128 helper macros, which are no longer used after deleting
6984 the old 'q' built-in functions, and moving the round to odd
6985 built-in functions to being special built-in functions.
6986 (BU_FLOAT128_2): Likewise.
6987 (BU_FLOAT128_1_HW): Likewise.
6988 (BU_FLOAT128_2_HW): Likewise.
6989 (BU_FLOAT128_3_HW): Likewise.
6990 (FABSQ): Delete old 'q' built-in functions.
6991 (COPYSIGNQ): Likewise.
6992 (SQRTF128_ODD): Move round to odd built-in functions to be
6993 special built-in functions, so that we can handle
6994 -mabi=ieeelongdouble.
6995 (TRUNCF128_ODD): Likewise.
6996 (ADDF128_ODD): Likewise.
6997 (SUBF128_ODD): Likewise.
6998 (MULF128_ODD): Likewise.
6999 (DIVF128_ODD): Likewise.
7000 (FMAF128_ODD): Likewise.
7001 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
7002 built-in names to 'f128'.
7003 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
7004 old 'q' built-in functions, as the machine independent code for
7005 'f128' built-in functions handles this.
7006 (rs6000_expand_builtin): Add expansion for float128 round to odd
7007 functions, keying off on -mabi=ieeelongdouble of whether to use
7008 the KFmode or TFmode variant.
7009 (rs6000_init_builtins): Initialize the _Float128 round to odd
7011 * doc/extend.texi (PowerPC Built-in Functions): Document the old
7012 _Float128 'q' built-in functions are now mapped into the new
7013 'f128' built-in functions.
7015 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
7017 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
7018 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
7020 2017-11-06 David Malcolm <dmalcolm@redhat.com>
7023 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
7024 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
7025 * toplev.c: Include "ipa-fnsummary.h".
7026 (toplev::finalize): Call ipa_fnsummary_c_finalize.
7028 2017-11-06 Jakub Jelinek <jakub@redhat.com>
7030 PR tree-optimization/82838
7031 * gimple-ssa-store-merging.c
7032 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
7033 on a separate gimple_seq which is then appended to seq.
7035 2017-11-06 Jeff Law <law@redhat.com>
7038 * config/i386/i386.c (PROBE_INTERVAL): Remove.
7039 (get_probe_interval): New functions.
7040 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
7041 (ix86_adjust_stack_and_probe): Likewise.
7042 (output_adjust_stack_and_probe): Likewise.
7043 (ix86_emit_probe_stack_range): Likewise.
7044 (ix86_expand_prologue): Likewise.
7046 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7048 PR tree-optimization/82816
7049 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
7050 if the modes of the two types are the same.
7051 (convert_plusminus_to_widen): Likewise.
7053 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7055 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
7057 (usadv16qi): New define_expand.
7058 (usadv8hi): New define_expand.
7060 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
7063 * ipa-inline-transform.c (update_noncloned_frequencies): Always
7065 (inline_transform): Likewise.
7066 * predict.c (counts_to_freqs): Remove useless conditional.
7067 * profile-count.h (profile_count::apply_scale): Move sanity check.
7068 * tree-inline.c (copy_bb): Always scale.
7069 (copy_cfg_body): Likewise.
7071 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
7074 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
7076 (*cmp_ite0): Add enabled_for_depr_it attribute.
7077 (*cmp_ite1): Likewise.
7079 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
7081 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
7084 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
7086 * tree-vrp.c (vrp_int_const_binop): Return true on success and
7087 return the value by pointer.
7088 (extract_range_from_multiplicative_op_1): Update accordingly.
7089 Return as soon as an operation fails.
7091 2017-11-05 Tom de Vries <tom@codesourcery.com>
7094 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
7095 (DEF_SANITIZER_BUILTIN): ... here.
7096 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
7097 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
7099 2017-11-05 Tom de Vries <tom@codesourcery.com>
7102 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
7104 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
7105 ASM_OUTPUT_BEFORE_CASE_LABEL call.
7106 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
7108 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7109 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7110 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
7112 2017-11-05 Tom de Vries <tom@codesourcery.com>
7115 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
7118 2017-11-04 Michael Clark <michaeljclark@mac.com>
7120 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
7121 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
7123 (*addsi3_extended): Likewise.
7124 (*addsi3_extended2): Likewise.
7125 (<optab>si3): Likewise.
7126 (<optab>di3): Likewise.
7127 (<optab><mode>3): Likewise.
7128 (<*optabe>si3_internal): Likewise.
7129 (zero_extendqi<SUPERQI:mode>2): Likewise.
7130 (*add<mode>hi3): Likewise.
7131 (*xor<mode>hi3): Likewise.
7132 (<optab>di3): Likewise.
7133 (*<optab>si3_extend): Likewise.
7134 (*sge<u>_<X:mode><GPR:mode>): Likewise.
7135 (*slt<u>_<X:mode><GPR:mode>): Likewise.
7136 (*sle<u>_<X:mode><GPR:mode>): Likewise.
7138 2017-11-04 Andrew Waterman <andrew@sifive.com>
7140 * config/riscv/riscv.c (riscv_option_override): Conditionally set
7141 TARGET_STRICT_ALIGN based upon -mtune argument.
7143 2017-11-04 Andrew Waterman <andrew@sifive.com>
7145 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
7147 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
7149 * config/i386/i386.c (choose_basereg): Use optional scratch
7150 register and add assertion.
7151 (x86_emit_outlined_ms2sysv_save): Use scratch register when
7152 needed, and don't allocate stack.
7153 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
7154 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
7155 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
7157 2017-11-03 Jeff Law <law@redhat.com>
7159 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
7160 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
7161 to probe at the start of a noreturn function.
7163 2017-11-03 Jakub Jelinek <jakub@redhat.com>
7165 PR tree-optimization/78821
7166 * gimple-ssa-store-merging.c: Update the file comment.
7167 (MAX_STORE_ALIAS_CHECKS): Define.
7168 (struct store_operand_info): New type.
7169 (store_operand_info::store_operand_info): New constructor.
7170 (struct store_immediate_info): Add rhs_code and ops data members.
7171 (store_immediate_info::store_immediate_info): Add rhscode, op0r
7172 and op1r arguments to the ctor, initialize corresponding data members.
7173 (struct merged_store_group): Add load_align_base and load_align
7175 (merged_store_group::merged_store_group): Initialize them.
7176 (merged_store_group::do_merge): Update them.
7177 (merged_store_group::apply_stores): Pick the constant for
7178 encode_tree_to_bitpos from one of the two operands, or skip
7179 encode_tree_to_bitpos if neither operand is a constant.
7180 (class pass_store_merging): Add process_store method decl. Remove
7181 bool argument from terminate_all_aliasing_chains method decl.
7182 (pass_store_merging::terminate_all_aliasing_chains): Remove
7183 var_offset_p argument and corresponding handling.
7184 (stmts_may_clobber_ref_p): New function.
7185 (compatible_load_p): New function.
7186 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
7187 if there is overlap and rhs_code is not INTEGER_CST. For
7188 non-overlapping stores terminate group if rhs is not mergeable.
7189 (get_alias_type_for_stmts): Change first argument from
7190 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
7191 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
7192 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
7194 (get_location_for_stmts): Change first argument from
7195 auto_vec<gimple *> & to vec<gimple *> &.
7196 (struct split_store): Remove orig_stmts data member, add orig_stores.
7197 (split_store::split_store): Create orig_stores rather than orig_stmts.
7198 (find_constituent_stmts): Renamed to ...
7199 (find_constituent_stores): ... this. Change second argument from
7200 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
7201 to info structures rather than the statements.
7202 (split_group): Rename ALLOW_UNALIGNED argument to
7203 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
7204 it. Adjust find_constituent_stores caller.
7205 (imm_store_chain_info::output_merged_store): Handle rhs_code other
7206 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
7207 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
7208 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
7209 (mem_valid_for_store_merging): New function.
7210 (handled_load): New function.
7211 (pass_store_merging::process_store): New method.
7212 (pass_store_merging::execute): Use process_store method. Adjust
7213 terminate_all_aliasing_chains caller.
7215 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7217 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
7218 Return true for more constants, symbols and label references.
7219 (aarch64_valid_floating_const): Remove unused function.
7221 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7224 * config/aarch64/aarch64.c (aarch64_layout_frame):
7225 Undo forcing of LR at bottom of frame.
7227 2017-11-03 Jeff Law <law@redhat.com>
7230 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
7231 for int_registers_saved.
7233 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
7234 extracted from tree-ssa-dom.c.
7235 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
7236 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
7237 (record_equivalences_from_incoming_edge): Add additional argument
7238 to single_pred_edge_ignoring_loop_edges call.
7239 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
7240 (uncprop_dom_walker::before_dom_children): Add additional argument
7241 to single_pred_edge_ignoring_loop_edges call.
7242 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
7243 single_pred_edge_ignoring_loop_edges rather than open coding.
7244 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
7246 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
7248 * match.pd (-(-A)): Rewrite.
7250 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7252 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
7253 (rs6000_emit_int_cmove): New declaration.
7254 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
7255 (rs6000_emit_sISEL): Delete.
7256 (rs6000_emit_int_cmove): Make non-static.
7257 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
7258 instead of rs6000_emit_sISEL.
7260 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
7262 * asan.c (create_cond_insert_point): Maintain profile.
7263 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
7265 * basic-block.h (struct basic_block_def): Remove frequency.
7266 (EDGE_FREQUENCY): Use to_frequency
7267 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
7269 (find_traces): Update to use to_frequency.
7270 (find_traces_1_round): Likewise; use only IPA counts.
7271 (bb_to_key): Likewise.
7272 (connect_traces): Use IPA counts only.
7273 (copy_bb_p): Update to use to_frequency.
7274 (fix_up_crossing_landing_pad): Likewise.
7275 (sanitize_hot_paths): Likewise.
7276 * bt-load.c (basic_block_freq): Likewise.
7277 * cfg.c (init_flow): Set count_max to uninitialized.
7278 (check_bb_profile): Remove frequencies; check counts.
7279 (dump_bb_info): Do not dump frequencies.
7280 (update_bb_profile_for_threading): Update counts only.
7281 (scale_bbs_frequencies_int): Likewise.
7282 (MAX_SAFE_MULTIPLIER): Remove.
7283 (scale_bbs_frequencies_gcov_type): Update counts only.
7284 (scale_bbs_frequencies_profile_count): Update counts only.
7285 (scale_bbs_frequencies): Update counts only.
7286 * cfg.h (struct control_flow_graph): Add count-max.
7287 (update_bb_profile_for_threading): Update prototype.
7288 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
7289 (find_many_sub_basic_blocks): Likewise.
7290 * cfgcleanup.c (try_forward_edges): Likewise.
7291 (try_crossjump_to_edge): Likewise.
7292 * cfgexpand.c (expand_gimple_cond): Likewise.
7293 (expand_gimple_tailcall): Likewise.
7294 (construct_init_block): Likewise.
7295 (construct_exit_block): Likewise.
7296 * cfghooks.c (verify_flow_info): Check consistency of counts.
7297 (dump_bb_for_graph): Do not dump frequencies.
7298 (split_block_1): Do not update frequencies.
7299 (split_edge): Do not update frequencies.
7300 (make_forwarder_block): Do not update frequencies.
7301 (duplicate_block): Do not update frequencies.
7302 (account_profile_record): Do not update frequencies.
7303 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
7304 for global heuristics.
7305 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
7306 (expected_loop_iterations_unbounded): Use counts only.
7307 * cfgloopmanip.c (scale_loop_profile): Simplify.
7308 (create_empty_loop_on_edge): Simplify
7310 (duplicate_loop_to_header_edge): Simplify
7311 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
7312 (update_br_prob_note): Take care of removing note when profile
7314 (relink_block_chain): Do not dump frequency.
7315 (rtl_account_profile_record): Use to_frequency.
7316 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
7317 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
7318 (cgraph_update_edges_for_call_stmt_node): Likewise.
7319 (cgraph_edge::verify_count_and_frequency): Update.
7320 (cgraph_node::verify_node): Temporarily disable frequency verification.
7321 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
7322 to_cgraph_frequency.
7323 (cgraph_edge::rebuild_edges): Convert to ipa counts.
7324 * cgraphunit.c (init_lowered_empty_function): Do not initialize
7326 (cgraph_node::expand_thunk): Update profile.
7327 * except.c (dw2_build_landing_pads): Do not update frequency.
7328 * final.c (compute_alignments): Use to_frequency.
7329 (dump_basic_block_info): Do not dump frequency.
7330 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
7331 (dump_gimple_bb_header): Do not dump frequency.
7332 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
7334 * gimple-streamer-in.c (input_bb): Do not stream frequency.
7335 * gimple-streamer-out.c (output_bb): Do not stream frequency.
7336 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
7337 (init_before_recovery): Do not update frequency.
7338 (sched_create_recovery_edges): Do not update frequency.
7339 * hsa-gen.c (convert_switch_statements): Do not update frequency.
7340 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
7341 (ipa_cp_c_finalize): Set max_count to uninitialized.
7342 * ipa-fnsummary.c (get_minimal_bb): Use counts.
7343 (param_change_prob): Use counts.
7344 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
7346 * ipa-split.c (consider_split): Use to_frequency.
7347 (split_function): Use to_frequency.
7348 * ira-build.c (loop_compare_func): Likewise.
7349 (mark_loops_for_removal): Likewise.
7350 (mark_all_loops_for_removal): Likewise.
7351 * loop-doloop.c (doloop_modify): Do not update frequency.
7352 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
7354 * lto-streamer-in.c (input_function): Update count_max.
7355 * omp-expand.c (expand_omp_taskreg): Update count_max.
7356 * omp-simd-clone.c (simd_clone_adjust): Update profile.
7357 * predict.c (maybe_hot_frequency_p): Use to_frequency.
7358 (maybe_hot_count_p): Use ipa counts only.
7359 (maybe_hot_bb_p): Simplify.
7360 (maybe_hot_edge_p): Simplify.
7361 (probably_never_executed): Do not take frequency argument.
7362 (probably_never_executed_bb_p): Do not pass frequency.
7363 (probably_never_executed_edge_p): Likewise.
7364 (combine_predictions_for_bb): Check that profile is nonzero.
7365 (propagate_freq): Do not set frequency.
7366 (drop_profile): Simplify.
7367 (counts_to_freqs): Simplify.
7368 (expensive_function_p): Use to_frequency.
7369 (propagate_unlikely_bbs_forward): Simplify.
7370 (determine_unlikely_bbs): Simplify.
7371 (estimate_bb_frequencies): Add hack to silence graphite issues.
7372 (compute_function_frequency): Use ipa counts.
7373 (pass_profile::execute): Update.
7374 (rebuild_frequencies): Use counts only.
7375 (force_edge_cold): Use counts only.
7376 * profile-count.c (profile_count::dump): Dump new count types.
7377 (profile_count::differs_from_p): Check compatiblity.
7378 (profile_count::to_frequency): New function.
7379 (profile_count::to_cgraph_frequency): New function.
7380 * profile-count.h (struct function): Declare.
7381 (enum profile_quality): Add profile_guessed_local and
7382 profile_guessed_global0.
7383 (class profile_proability): Decrease number of bits to 29;
7384 update from_reg_br_prob_note and to_reg_br_prob_note.
7385 (class profile_count: Update comment; decrease number of bits
7386 to 61. Check compatibility.
7387 (profile_count::compatible_p): New private member function.
7388 (profile_count::ipa_p): New member function.
7389 (profile_count::operator<): Handle global zero correctly.
7390 (profile_count::operator>): Handle global zero correctly.
7391 (profile_count::operator<=): Handle global zero correctly.
7392 (profile_count::operator>=): Handle global zero correctly.
7393 (profile_count::nonzero_p): New member function.
7394 (profile_count::force_nonzero): New member function.
7395 (profile_count::max): New member function.
7396 (profile_count::apply_scale): Handle IPA scalling.
7397 (profile_count::guessed_local): New member function.
7398 (profile_count::global0): New member function.
7399 (profile_count::ipa): New member function.
7400 (profile_count::to_frequency): Declare.
7401 (profile_count::to_cgraph_frequency): Declare.
7402 * profile.c (OVERLAP_BASE): Delete.
7403 (compute_frequency_overlap): Delete.
7404 (compute_branch_probabilities): Do not use compute_frequency_overlap.
7405 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
7406 * sched-ebb.c (rank): Use counts only.
7407 * shrink-wrap.c (handle_simple_exit): Use counts only.
7408 (try_shrink_wrapping): Use counts only.
7409 (place_prologue_for_one_component): Use counts only.
7410 * tracer.c (find_best_predecessor): Use to_frequency.
7411 (find_trace): Use to_frequency.
7412 (tail_duplicate): Use to_frequency.
7413 * trans-mem.c (expand_transaction): Do not update frequency.
7414 * tree-call-cdce.c: Do not update frequency.
7415 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
7416 (gimple_merge_blocks): Likewise.
7417 (gimple_split_edge): Likewise.
7418 (gimple_duplicate_sese_region): Likewise.
7419 (gimple_duplicate_sese_tail): Likewise.
7420 (move_sese_region_to_fn): Likewise.
7421 (gimple_account_profile_record): Likewise.
7422 (insert_cond_bb): Likewise.
7423 * tree-complex.c (expand_complex_div_wide): Likewise.
7424 * tree-eh.c (lower_resx): Update profile.
7425 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
7427 (initialize_cfun): Do not initialize frequencies
7428 (freqs_to_counts): Delete.
7429 (copy_cfg_body): Ignore count parameter.
7430 (copy_body): Update.
7431 (expand_call_inline): Update count_max.
7432 (optimize_inline_calls): Update count_max.
7433 (tree_function_versioning): Update count_max.
7434 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
7435 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
7437 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
7438 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
7439 (try_peel_loop): Likewise.
7440 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
7442 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
7443 (tree_transform_and_unroll_loop): Do not use frequencies
7444 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
7445 Use reliable prediction only.
7446 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
7447 * tree-ssa-sink.c (select_best_block): Use to_frequency.
7448 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
7449 probability scaling.
7450 * tree-ssa-threadupdate.c (create_block_for_threading): Do
7451 not update frequency
7452 (any_remaining_duplicated_blocks): Likewise.
7453 (update_profile): Likewise.
7454 (estimated_freqs_path): Delete.
7455 (freqs_to_counts_path): Delete.
7456 (clear_counts_path): Delete.
7457 (ssa_fix_duplicate_block_edges): Likewise.
7458 (duplicate_thread_path): Likewise.
7459 * tree-switch-conversion.c (gen_inbound_check): Use counts.
7460 * tree-tailcall.c (decrease_profile): Do not update frequency.
7461 (eliminate_tail_call): Likewise.
7462 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
7463 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
7464 (optimize_mask_stores): Likewise.
7465 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
7466 * ubsan.c (ubsan_expand_null_ifn): Update profile.
7467 (ubsan_expand_ptr_ifn): Update profile.
7468 * value-prof.c (gimple_ic): Simplify.
7469 * value-prof.h (gimple_ic): Update prototype.
7470 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
7471 * ipa-inline.c (compute_uninlined_call_time): Be sure that
7473 (want_inline_self_recursive_call_p): Likewise.
7474 (resolve_noninline_speculation): Only cummulate defined counts.
7475 (inline_small_functions): Use nonzero_p.
7476 (ipa_inline): Do not access freed node.
7478 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
7480 * config/aarch64/aarch64.c (aarch64_override_options_internal):
7481 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
7483 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
7485 * config/riscv/riscv.c (riscv_legitimize_move): Handle
7486 non-legitimate address.
7488 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7490 * config/rs6000/rs6000.md (*lt0_disi): Delete.
7491 (*lt0_<mode>di, *lt0_<mode>si): New.
7493 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
7495 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
7496 TARGET_PAIRED_FLOAT.
7498 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
7499 Jim Wilson <jim.wilson@linaro.org>
7501 * config/aarch64/aarch64-cores.def (saphira): New CPU.
7502 * config/aarch64/aarch64-tune.md: Regenerated.
7503 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
7504 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
7506 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
7508 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
7509 (arc_expand_prologue): Restore blink for millicode.
7510 * config/arc/linux.h (LINK_EH_SPEC): Defined.
7512 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
7515 * config/i386/i386.c (ix86_vector_duplicate_value): Use
7516 gen_vec_duplicate after forcing the scalar into a register.
7518 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
7520 * combine (try_combine): Print the insns input to try_combine to the
7523 2017-11-02 Steve Ellcey <sellcey@cavium.com>
7526 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
7527 Remove second argument from aarch64_process_target_attr call.
7528 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
7530 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
7532 (aarch64_handle_attr_arch): Remove second argument.
7533 (aarch64_handle_attr_cpu): Ditto.
7534 (aarch64_handle_attr_tune): Ditto.
7535 (aarch64_handle_attr_isa_flags): Ditto.
7536 (aarch64_process_one_target_attr): Ditto.
7537 (aarch64_process_target_attr): Ditto.
7538 (aarch64_option_valid_attribute_p): Remove second argument.
7539 on aarch64_process_target_attr call.
7541 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7543 * diagnostic.c: Include "selftest-diagnostic.h".
7544 (selftest::assert_location_text): New function.
7545 (selftest::test_diagnostic_get_location_text): New function.
7546 (selftest::diagnostic_c_tests): Call it.
7548 2017-11-02 David Malcolm <dmalcolm@redhat.com>
7550 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
7551 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
7552 (class selftest::test_diagnostic_context): Move to...
7553 * selftest-diagnostic.c: New file.
7554 * selftest-diagnostic.h: New file.
7556 2017-11-02 James Bowman <james.bowman@ftdichip.com>
7558 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
7559 offset range for FT32B.
7560 * config/ft32/ft32.h: option "mcompress" enables relaxation.
7561 * config/ft32/ft32.md: Add TARGET_NOPM.
7562 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
7563 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
7565 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
7567 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
7569 2017-11-02 Jeff Law <law@redhat.com>
7571 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
7572 virtual keyword on FINAL OVERRIDE members.
7574 * tree-ssa-propagate.h (ssa_propagation_engine): Group
7575 virtuals together. Add virtual destructor.
7576 (substitute_and_fold_engine): Similarly.
7578 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
7580 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
7582 2017-11-02 Richard Biener <rguenther@suse.de>
7584 PR tree-optimization/82795
7585 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
7587 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7589 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
7591 (gcc_GAS_CHECK_FEATURE): Remove.
7592 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
7593 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
7594 * configure: Regenerate.
7596 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
7598 * config/arc/arc.c (hwloop_optimize): Account for empty
7601 2017-11-02 Richard Biener <rguenther@suse.de>
7604 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
7605 Truncate ARRAY_REF index and element size.
7607 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
7609 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
7611 2017-11-01 Jeff Law <law@redhat.com>
7613 * tree-ssa-ccp.c (ccp_folder): New class derived from
7614 substitute_and_fold_engine.
7615 (ccp_folder::get_value): New member function.
7616 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
7617 (ccp_fold_stmt): Remove prototype.
7618 (ccp_finalize): Call substitute_and_fold from the ccp_class.
7619 * tree-ssa-copy.c (copy_folder): New class derived from
7620 substitute_and_fold_engine.
7621 (copy_folder::get_value): Renamed from get_value.
7622 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
7623 * tree-vrp.c (vrp_folder): New class derived from
7624 substitute_and_fold_engine.
7625 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
7626 (vrp_folder::get_value): New member function.
7627 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
7628 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
7629 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
7630 provide a class interface to folder/substitute routines.
7631 (ssa_prop_fold_stmt_fn): Remove typedef.
7632 (ssa_prop_get_value_fn): Likewise.
7633 (subsitute_and_fold): Remove prototype.
7634 (replace_uses_in): Likewise.
7635 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
7636 Renamed from replace_uses_in. Call the virtual member function
7637 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
7638 (substitute_and_fold_dom_walker): Remove initialization of
7639 data member entries for calbacks. Add substitute_and_fold_engine
7640 member and initialize it.
7641 (substitute_and_fold_dom_walker::before_dom_children0: Use the
7642 member functions for get_value, replace_phi_args_in c
7643 replace_uses_in, and fold_stmt calls.
7644 (substitute_and_fold_engine::substitute_and_fold): Renamed from
7645 substitute_and_fold. Remove assert. Update ctor call.
7647 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
7648 (ssa_prop_visit_phi_fn): Likewise.
7649 (class ssa_propagation_engine): New class to provide an interface
7651 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
7653 (ssa_prop_visit_phi): Likewise.
7654 (ssa_propagation_engine::simulate_stmt): Moved into class.
7655 Call visit_phi/visit_stmt from the class rather than via
7656 file scoped static variables.
7657 (ssa_propagation_engine::simulate_block): Moved into class.
7658 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
7659 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
7660 set file scoped statics for the visit_stmt/visit_phi callbacks.
7661 * tree-complex.c (complex_propagate): New class derived from
7662 ssa_propagation_engine.
7663 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
7664 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
7665 (tree_lower_complex): Call ssa_propagate via the complex_propagate
7667 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
7668 ssa_propagation_engine.
7669 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
7670 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
7671 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
7672 * tree-ssa-copy.c (copy_prop): New class derived from
7673 ssa_propagation_engine.
7674 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
7675 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
7676 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
7677 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
7678 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
7679 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
7680 (execute_vrp): Call ssa_propagate from the vrp_prop class.
7682 2017-11-01 Jakub Jelinek <jakub@redhat.com>
7684 PR rtl-optimization/82778
7685 PR rtl-optimization/82597
7686 * compare-elim.c (struct comparison): Add in_a_setter field.
7687 (find_comparison_dom_walker::before_dom_children): Remove killed
7688 bitmap and df_simulate_find_defs call, instead walk the defs.
7689 Compute last_setter and initialize in_a_setter. Merge definitions
7690 with first initialization for a few variables.
7691 (try_validate_parallel): Use insn_invalid_p instead of
7692 recog_memoized. Return insn rather than just the pattern.
7693 (try_merge_compare): Fix up comment. Don't uselessly test if
7694 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
7696 (execute_compare_elim_after_reload): Remove df_chain_add_problem
7699 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7700 Alan Hayward <alan.hayward@arm.com>
7701 David Sherwood <david.sherwood@arm.com>
7703 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
7704 aarch64_hard_regno_nregs to get the number of registers
7707 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7708 Alan Hayward <alan.hayward@arm.com>
7709 David Sherwood <david.sherwood@arm.com>
7711 * config/aarch64/constraints.md (Upl): Rename to...
7713 * config/aarch64/aarch64.md
7714 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
7717 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7718 Alan Hayward <alan.hayward@arm.com>
7719 David Sherwood <david.sherwood@arm.com>
7721 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
7722 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
7725 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7726 Alan Hayward <alan.hayward@arm.com>
7727 David Sherwood <david.sherwood@arm.com>
7729 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
7730 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
7731 (aarch64_evpc_dup): Generate rtl direcly, rather than using
7733 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
7734 of a single element.
7735 * config/aarch64/iterators.md: Add a comment above the permute
7736 unspecs to say that they are generated directly by
7737 aarch64_expand_vec_perm_const.
7738 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
7740 2017-11-01 Nathan Sidwell <nathan@acm.org>
7742 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
7744 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
7746 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
7747 and medany code models, and describe what they do.
7749 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7751 Revert accidental duplicate:
7753 * combine.c (can_change_dest_mode): Reject changes in
7754 REGMODE_NATURAL_SIZE.
7756 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
7758 PR rtl-optimization/64682
7759 PR rtl-optimization/69567
7760 PR rtl-optimization/69737
7761 PR rtl-optimization/82683
7762 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
7763 register mentioned in the note, drop the note, unless it came from I3,
7764 in which case it should go to I3 again.
7766 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7768 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
7769 and return false if not.
7770 (clear_bytes_written_by, live_bytes_read): Update accordingly.
7772 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7774 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
7775 range is known to be empty.
7777 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7778 Alan Hayward <alan.hayward@arm.com>
7779 David Sherwood <david.sherwood@arm.com>
7781 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
7782 and CONST_VECTOR_NUNITS instead of computing the number of units from
7783 the byte sizes of the vector and element.
7784 (simplify_binary_operation_1): Likewise.
7785 (simplify_const_binary_operation): Likewise.
7786 (simplify_ternary_operation): Likewise.
7788 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7789 Alan Hayward <alan.hayward@arm.com>
7790 David Sherwood <david.sherwood@arm.com>
7792 * var-tracking.c (INT_MEM_OFFSET): Replace with...
7793 (int_mem_offset): ...this new function.
7794 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
7795 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
7796 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
7799 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7800 Alan Hayward <alan.hayward@arm.com>
7801 David Sherwood <david.sherwood@arm.com>
7803 * lower-subreg.c (interesting_mode_p): New function.
7804 (compute_costs, find_decomposable_subregs, decompose_register)
7805 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
7806 (resolve_clobber, dump_choices): Use it.
7808 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7809 Alan Hayward <alan.hayward@arm.com>
7810 David Sherwood <david.sherwood@arm.com>
7812 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
7814 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7815 Alan Hayward <alan.hayward@arm.com>
7816 David Sherwood <david.sherwood@arm.com>
7818 * alias.c (find_base_value, find_base_term): Only process integer
7819 truncations. Check the precision rather than the size.
7821 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7822 Alan Hayward <alan.hayward@arm.com>
7823 David Sherwood <david.sherwood@arm.com>
7825 * machmode.h (is_narrower_int_mode): New function
7826 * optabs.c (expand_float, expand_fix): Use it.
7827 * dwarf2out.c (rotate_loc_descriptor): Likewise.
7829 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7830 Alan Hayward <alan.hayward@arm.com>
7831 David Sherwood <david.sherwood@arm.com>
7833 * rtl.h (narrower_subreg_mode): New function.
7834 * ira-color.c (update_costs_from_allocno): Use it.
7836 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7837 Alan Hayward <alan.hayward@arm.com>
7838 David Sherwood <david.sherwood@arm.com>
7840 * optabs-query.h (convert_optab_p): New function, split out from...
7841 (convert_optab_handler): ...here.
7842 (widening_optab_handler): Delete.
7843 (find_widening_optab_handler): Remove permit_non_widening parameter.
7844 (find_widening_optab_handler_and_mode): Likewise. Provide an
7845 override that operates on mode class wrappers.
7846 * optabs-query.c (widening_optab_handler): Delete.
7847 (find_widening_optab_handler_and_mode): Remove permit_non_widening
7848 parameter. Assert that the two modes are the same class and that
7849 the "from" mode is narrower than the "to" mode. Use
7850 convert_optab_handler instead of widening_optab_handler.
7851 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
7852 instead of widening_optab_handler.
7853 * expr.c (expand_expr_real_2): Update calls to
7854 find_widening_optab_handler.
7855 * optabs.c (expand_widen_pattern_expr): Likewise.
7856 (expand_binop_directly): Take the insn_code as a parameter.
7857 (expand_binop): Only call find_widening_optab_handler for
7858 conversion optabs; use optab_handler otherwise. Update calls
7859 to find_widening_optab_handler and expand_binop_directly.
7860 Use convert_optab_handler instead of widening_optab_handler.
7861 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
7862 find_widening_optab_handler and use scalar_mode rather than
7864 (convert_plusminus_to_widen): Likewise.
7866 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7867 Alan Hayward <alan.hayward@arm.com>
7868 David Sherwood <david.sherwood@arm.com>
7870 * machmode.h (fixed_size_mode): New class.
7871 * rtl.h (get_pool_mode): Return fixed_size_mode.
7872 * gengtype.c (main): Add fixed_size_mode.
7873 * target.def (get_raw_result_mode): Return a fixed_size_mode.
7874 (get_raw_arg_mode): Likewise.
7875 * doc/tm.texi: Regenerate.
7876 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
7877 * targhooks.c (default_get_reg_raw_mode): Likewise.
7878 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
7879 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
7880 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
7881 (msp430_get_raw_result_mode): Likewise.
7882 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
7883 * dbxout.c (dbxout_parms): Require fixed-size modes.
7884 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
7885 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
7886 * omp-low.c (lower_oacc_reductions): Likewise.
7887 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
7888 (simplify_subreg): Update accordingly.
7889 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
7890 (force_const_mem): Update accordingly. Return NULL_RTX for modes
7891 that aren't fixed-size.
7892 (get_pool_mode): Return a fixed_size_mode.
7893 (output_constant_pool_2): Take a fixed_size_mode.
7895 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7896 Alan Hayward <alan.hayward@arm.com>
7897 David Sherwood <david.sherwood@arm.com>
7899 * doc/rtl.texi (vec_series): Document.
7900 (const): Say that the operand can be a vec_series.
7901 * rtl.def (VEC_SERIES): New rtx code.
7902 * rtl.h (const_vec_series_p_1): Declare.
7903 (const_vec_series_p): New function.
7904 * emit-rtl.h (gen_const_vec_series): Declare.
7905 (gen_vec_series): Likewise.
7906 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
7907 (gen_vec_series): Likewise.
7908 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
7909 * simplify-rtx.c (simplify_unary_operation): Handle negations
7911 (simplify_binary_operation_series): New function.
7912 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
7913 (test_vector_ops_series): New function.
7914 (test_vector_ops): Call it.
7915 * config/powerpcspe/altivec.md (altivec_lvsl): Use
7916 gen_const_vec_series.
7917 (altivec_lvsr): Likewise.
7918 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
7920 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7921 Alan Hayward <alan.hayward@arm.com>
7922 David Sherwood <david.sherwood@arm.com>
7924 * doc/rtl.texi (const): Update description of address constants.
7925 Say that vector constants are allowed too.
7926 * common.md (E, F): Use CONSTANT_P instead of checking for
7928 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
7929 checking for CONST_VECTOR.
7930 * expmed.c (make_tree): Use build_vector_from_val for a CONST
7932 * expr.c (expand_expr_real_2): Check for vector modes instead
7933 of checking for CONST_VECTOR.
7934 * rtl.h (const_vec_p): New function.
7935 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
7936 (unwrap_const_vec_duplicate): Handle them here too.
7938 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7939 David Malcolm <dmalcolm@redhat.com>
7940 Alan Hayward <alan.hayward@arm.com>
7941 David Sherwood <david.sherwood@arm.com>
7943 * rtl.h (vec_duplicate_p): New function.
7944 * selftest-rtl.c (assert_rtx_eq_at): New function.
7945 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
7946 (assert_rtx_eq_at): Declare.
7947 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
7948 * selftest-run-tests.c (selftest::run_tests): Call it.
7949 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
7950 (simplify_unary_operation_1): Recursively handle vector duplicates.
7951 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
7953 (simplify_subreg): Handle subregs of vector duplicates.
7954 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
7955 (selftest::simplify_rtx_c_tests): New functions.
7957 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7958 Alan Hayward <alan.hayward@arm.com>
7959 David Sherwood <david.sherwood@arm.com>
7961 * emit-rtl.h (gen_const_vec_duplicate): Declare.
7962 (gen_vec_duplicate): Likewise.
7963 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
7965 (gen_const_vector): ...here.
7966 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
7967 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
7968 whose elements are all equal.
7969 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
7970 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
7971 (simplify_relational_operation): Likewise.
7972 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
7974 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
7975 (aarch64_expand_vector_init): Likewise.
7976 * config/arm/arm.c (neon_vdup_constant): Likewise.
7977 (neon_expand_vector_init): Likewise.
7978 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
7979 (arm_block_set_unaligned_vect): Likewise.
7980 (arm_block_set_aligned_vect): Likewise.
7981 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
7982 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
7983 (expand_vec_perm_even_odd_pack): Likewise.
7984 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
7985 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
7986 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
7987 gen_const_vec_duplicate.
7988 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
7989 * config/mips/mips.c (mips_gen_const_int_vector): Use
7990 gen_const_vec_duplicate.
7991 (mips_expand_vector_init): Use CONST0_RTX.
7992 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
7993 (define_split): Use gen_const_vec_duplicate.
7994 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
7995 (define_split): Use gen_const_vec_duplicate.
7996 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
7997 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
7998 * config/spu/spu.c (spu_const): Likewise.
8000 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
8001 Alan Hayward <alan.hayward@arm.com>
8002 David Sherwood <david.sherwood@arm.com>
8004 * combine.c (can_change_dest_mode): Reject changes in
8005 REGMODE_NATURAL_SIZE.
8007 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
8009 * configure.ac (--enable-libssp): New.
8010 (gcc_cv_libc_provides_ssp): Check for explicit setting before
8011 trying to determine target-specific default. Adjust indentation.
8012 * configure: Regenerated.
8013 * doc/install.texi (Configuration): Expand --disable-libssp
8016 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
8018 config/i386/i386.c (ix86_expand_epilogue): Correct stack
8021 2017-10-31 Martin Jambor <mjambor@suse.cz>
8024 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
8026 2017-10-31 David Malcolm <dmalcolm@redhat.com>
8028 * auto-profile.c (autofdo_source_profile::read): Use
8029 UNKNOWN_LOCATION rather than 0.
8030 * diagnostic-core.h (warning_at_rich_loc): Rename to...
8031 (warning_at): ...this overload.
8032 (warning_at_rich_loc_n): Rename to...
8033 (warning_n): ...this overload.
8034 (error_at_rich_loc): Rename to...
8035 (error_at): ...this overload.
8036 (pedwarn_at_rich_loc): Rename to...
8037 (pedwarn): ...this overload.
8038 (permerror_at_rich_loc): Rename to...
8039 (permerror): ...this overload.
8040 (inform_at_rich_loc): Rename to...
8041 (inform): ...this overload.
8042 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
8043 (diagnostic_n_impl_richloc): Rename to...
8044 (diagnostic_n_impl): ...this rich_location *-based decl.
8045 (inform_at_rich_loc): Rename to...
8046 (inform): ...this, and add an assertion.
8047 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
8048 (warning_at_rich_loc): Rename to...
8049 (warning_at): ...this, and add an assertion.
8050 (warning_at_rich_loc_n): Rename to...
8051 (warning_n): ...this, and add an assertion.
8052 (warning_n): Update location_t-based implementation for removal of
8053 location_t-based diagnostic_n_impl.
8054 (pedwarn_at_rich_loc): Rename to...
8055 (pedwarn): ...this, and add an assertion.
8056 (permerror_at_rich_loc): Rename to...
8057 (permerror): ...this, and add an assertion.
8058 (error_n): Update for removal of location_t-based diagnostic_n_impl.
8059 (error_at_rich_loc): Rename to...
8060 (error_at): ...this, and add an assertion.
8061 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
8062 (driver::do_spec_on_infiles): Likewise.
8063 * substring-locations.c (format_warning_va): Update for renaming
8064 of inform_at_rich_loc.
8066 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
8068 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
8069 _Float<N>X built-in functions so that the variant without the
8070 "__builtin_" prefix is only enabled for the GNU C and Objective C
8071 languages when they are in non-strict ANSI/ISO mode.
8072 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
8073 * target.def (floatn_builtin_p): Add a target hook to control
8074 whether _Float<N> and _Float<N>X built-in functions without the
8075 "__builtin_" prefix are enabled, and return true for C and
8076 Objective C in the default hook. Include langhooks.h in
8078 * targhooks.h (default_floatn_builtin_p): Likewise.
8079 * targhooks.c (default_floatn_builtin_p): Likewise.
8080 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
8081 floatn_builtin_p target hook.
8082 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
8084 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
8085 Eric Botcazou <ebotcazou@adacore.com>
8087 PR rtl-optimization/81803
8088 * lra-constraints.c (curr_insn_transform): Also reload the whole
8089 register for a strict subreg no wider than a word if this is for
8090 a WORD_REGISTER_OPERATIONS target.
8092 2017-10-31 Jason Merrill <jason@redhat.com>
8094 * gdbinit.in: Skip over inlines from timevar.h.
8096 2017-10-31 Martin Liska <mliska@suse.cz>
8098 * doc/gcov.texi: Document new option.
8099 * gcov.c (print_usage): Likewise print it.
8100 (process_args): Support the argument.
8101 (format_count): New function.
8102 (format_gcov): Use the function.
8104 2017-10-31 Martin Liska <mliska@suse.cz>
8106 * gcov.c (struct name_map): do not use typedef.
8107 Define operator== and operator<.
8108 (name_search): Remove.
8109 (name_sort): Remove.
8110 (main): Do not allocate names.
8111 (process_file): Add vertical space.
8112 (generate_results): Use std::find.
8113 (release_structures): Do not release memory.
8114 (find_source): Use std::find.
8116 2017-10-31 Martin Liska <mliska@suse.cz>
8118 * gcov.c (struct line_info): Remove it's typedef.
8119 (line_info::line_info): Add proper ctor.
8120 (line_info::has_block): Do not use a typedef.
8121 (struct source_info): Do not use typedef.
8122 (circuit): Likewise.
8123 (get_cycles_count): Likewise.
8124 (output_intermediate_file): Iterate via vector iterator.
8125 (add_line_counts): Use std::vector methods.
8126 (accumulate_line_counts): Likewise.
8127 (output_lines): Likewise.
8129 2017-10-31 Martin Liska <mliska@suse.cz>
8131 * gcov.c (struct source_info): Remove typedef.
8132 (source_info::source_info): Add proper ctor.
8133 (accumulate_line_counts): Use struct, not it's typedef.
8134 (output_gcov_file): Likewise.
8135 (output_lines): Likewise.
8136 (main): Do not allocate an array.
8137 (output_intermediate_file): Use size of vector container.
8138 (process_file): Resize the vector.
8139 (generate_results): Do not preallocate, use newly added vector
8141 (release_structures): Do not release sources.
8142 (find_source): Use vector methods.
8143 (add_line_counts): Do not use typedef.
8145 2017-10-31 Martin Liska <mliska@suse.cz>
8147 * doc/gcov.texi: Document that.
8148 * gcov.c (add_line_counts): Mark lines with a non-executed
8150 (output_line_beginning): Handle such lines.
8151 (output_lines): Pass new argument.
8152 (output_intermediate_file): Print it in intermediate format.
8154 2017-10-31 Martin Liska <mliska@suse.cz>
8156 * color-macros.h: New file.
8157 * diagnostic-color.c: Factor out color related to macros to
8159 * doc/gcov.texi: Document -k option.
8160 * gcov.c (INCLUDE_STRING): Include string.h.
8161 (print_usage): Add -k option.
8162 (process_args): Parse it.
8163 (pad_count_string): New function.
8164 (output_line_beginning): Likewise.
8165 (DEFAULT_LINE_START): New macro.
8166 (output_lines): Support color output.
8168 2017-10-31 Martin Liska <mliska@suse.cz>
8170 PR gcov-profile/82633
8171 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
8172 their interaction with GCOV infrastructure.
8173 * configure.ac: Add -fkeep-{inline,static}-functions to
8175 * configure: Regenerate.
8177 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
8180 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
8182 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
8185 * config/rs6000/rs6000.md (allocate_stack): Force update interval
8186 into a register if it does not fit into an immediate offset field.
8188 2017-10-31 Olivier Hainque <hainque@adacore.com>
8190 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
8192 2017-10-31 Julia Koval <julia.koval@intel.com>
8194 * config.gcc: Add gfniintrin.h.
8195 * config/i386/gfniintrin.h: New.
8196 * config/i386/i386-builtin-types.def
8197 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8198 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8199 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8200 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8201 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8202 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
8203 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
8204 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
8205 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
8206 V64QI_FTYPE_V64QI_V64QI_INT): New types.
8207 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
8208 * config/i386/immintrin.h: Include gfniintrin.h.
8209 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
8211 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
8213 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
8215 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8217 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
8218 (arm_ashldi3_1bit): Remove pattern.
8219 (ashrdi3): Remove shift by 1 expansion.
8220 (arm_ashrdi3_1bit): Remove pattern.
8221 (lshrdi3): Remove shift by 1 expansion.
8222 (arm_lshrdi3_1bit): Remove pattern.
8223 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
8224 cost of ashldi3 by 1.
8225 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
8226 (<shift>di3_neon): Likewise.
8228 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8230 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
8231 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
8232 and (*aarch64_simd_mov<VQ:mode>).
8233 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
8234 pattern alternative.
8235 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
8236 attributes to match pattern alternative.
8238 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
8240 * config.gcc (powerpc*-*-*): Add emmintrin.h.
8241 * config/rs6000/emmintrin.h: New file.
8242 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
8244 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
8246 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
8247 * (movdi_vfp_cortexa8): Remove pattern.
8249 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8251 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
8253 (Specific, alpha*-dec-osf5.1): Remove.
8254 (Specific, mips-sgi-irix5): Remove.
8255 (Specific, mips-sgi-irix6): Remove.
8257 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8260 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
8261 arguments to clear_bit_region_be.
8263 2017-10-30 Jim Wilson <wilson@tuliptree.org>
8265 * gimplify.c: Include memmodel.h.
8267 2017-10-30 Martin Jambor <mjambor@suse.cz>
8269 * omp-grid.c (grid_attempt_target_gridification): Also insert a
8270 condition whether loop should be executed at all.
8272 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
8274 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
8275 gimple folding of vec_madd() intrinsics.
8276 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
8277 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
8278 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
8280 2017-10-30 Richard Biener <rguenther@suse.de>
8282 PR tree-optimization/82762
8284 2017-10-23 Richard Biener <rguenther@suse.de>
8286 PR tree-optimization/82129
8288 2017-08-01 Richard Biener <rguenther@suse.de>
8290 PR tree-optimization/81181
8291 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8292 (compute_antic): ... end of iteration here.
8294 2017-10-30 Joseph Myers <joseph@codesourcery.com>
8296 * doc/invoke.texi (C Dialect Options): Document -std=c17,
8297 -std=iso9899:2017 and -std=gnu17.
8298 * doc/standards.texi (C Language): Document C17 support.
8299 * doc/cpp.texi (Overview): Mention -std=c17.
8300 (Standard Predefined Macros): Document C11 and C17 values of
8301 __STDC_VERSION__. Do not refer to C99 support as incomplete.
8302 * doc/extend.texi (Inline): Do not list individual options for
8303 standards newer than C99.
8304 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
8306 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
8309 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
8311 * asan.c (asan_finish_file): Align asan globals array by shadow
8314 2017-10-30 Jakub Jelinek <jakub@redhat.com>
8317 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
8318 (struct store_immediate_info): Add bitregion_start and bitregion_end
8320 (store_immediate_info::store_immediate_info): Add brs and bre
8321 arguments and initialize bitregion_{start,end} from those.
8322 (struct merged_store_group): Add bitregion_start, bitregion_end,
8323 align_base and mask fields. Drop unnecessary struct keyword from
8324 struct store_immediate_info. Add do_merge method.
8325 (clear_bit_region_be): Use memset instead of loop storing zeros.
8326 (merged_store_group::do_merge): New method.
8327 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
8328 stores as long as the surrounding bitregions have no gaps.
8329 (merged_store_group::merge_overlapping): Use do_merge.
8330 (merged_store_group::apply_stores): Test that bitregion_{start,end}
8331 is byte aligned, rather than requiring that start and width are
8332 byte aligned. Drop unnecessary struct keyword from
8333 struct store_immediate_info. Allocate and populate also mask array.
8334 Make start of the arrays relative to bitregion_start rather than
8335 start and size them according to bitregion_{end,start} difference.
8336 (struct imm_store_chain_info): Drop unnecessary struct keyword from
8337 struct store_immediate_info.
8338 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
8339 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
8340 struct keyword from struct store_immediate_info.
8341 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
8342 between stores as long as the surrounding bitregions have no gaps.
8344 (struct split_store): Add orig non-static data member.
8345 (split_store::split_store): Initialize orig to false.
8346 (find_constituent_stmts): Return store_immediate_info *, non-NULL
8347 if there is exactly a single original stmt. Change stmts argument
8348 to pointer from reference, if NULL, don't push anything to it. Add
8349 first argument, use it to optimize skipping over orig stmts that
8350 are known to be before bitpos already. Simplify.
8351 (split_group): Return unsigned int count how many stores are or
8352 would be needed rather than a bool. Add allow_unaligned argument.
8353 Change split_stores argument from reference to pointer, if NULL,
8354 only do a dry run computing how many stores would be produced.
8355 Rewritten algorithm to use both alignment and misalign if
8356 !allow_unaligned and handle bitfield stores with gaps.
8357 (imm_store_chain_info::output_merged_store): Set start_byte_pos
8358 from bitregion_start instead of start. Compute allow_unaligned
8359 here, if true, do 2 split_group dry runs to compute which one
8360 produces fewer stores and prefer aligned if equal. Punt if
8361 new count is bigger or equal than original before emitting any
8362 statements, rather than during that. Remove no longer needed
8363 new_ssa_names tracking. Replace num_stmts with
8364 split_stores.length (). Use 32-bit stack allocated entries
8365 in split_stores auto_vec. Try to reuse original store lhs/rhs1
8366 if possible. Handle bitfields with gaps.
8367 (pass_store_merging::execute): Ignore bitsize == 0 stores.
8368 Compute bitregion_{start,end} for the stores and construct
8369 store_immediate_info with that. Formatting fixes.
8371 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
8374 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
8375 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
8377 2017-10-29 Jim Wilson <wilson@tuliptree.org>
8379 * gimplify.c: Include tm_p.h.
8381 * common.opt (gcoff): Re-add as ignored option.
8382 (gcoff1, gcoff2, gcoff3): Likewise.
8384 * Makefile.in (OBJS): Delete sdbout.o.
8385 (GTFILES): Delete $(srcdir)/sdbout.c.
8386 * debug.h: Delete sdb_debug_hooks.
8387 * final.c: Delete sdbout.h include.
8388 (final_scan_insn): Delete SDB_DEBUG check.
8389 (rest_of_clean_state): Likewise.
8390 * output.h: Delete sdb_begin_function_line.
8393 * toplev.c: Delete sdbout.h include.
8394 (process_options): Delete SDB_DEBUG check.
8395 * tree-core.h (tree_type_common): Delete pointer field of
8397 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
8398 TYPE_SYMTAB_POINTER.
8399 * tree.h (TYPE_SYMTAB_POINTER): Delete.
8400 (TYPE_SYMTAB_IS_POINTER): Delete.
8401 (TYPE_SYMTAB_IS_DIE): Renumber.
8402 * xcoffout.c: Refer to former sdbout.c file.
8403 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
8405 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
8406 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
8407 (Debugging Options): Delete -gcoff.
8408 (-gstabs): Delete SDB reference.
8410 (-gcoff@var{level}): Delete.
8411 * doc/passes.texi (Debugging information output): Delete SDB and
8412 sdbout.c references.
8413 * doc/tm.texi: Regenerate.
8414 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
8415 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
8417 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
8418 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
8419 (SDB_DEBUGGING_INFO): Delete.
8420 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
8421 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
8422 * target.def (output_source_filename): Delete COFF reference.
8424 * common.opt (gcoff): Delete.
8425 (gxcoff+): Update Negative chain.
8426 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
8428 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
8429 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
8430 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
8432 (expand_function_start): Change sdb reference to past tense.
8433 (expand_function_end): Change sdb reference to past tense.
8434 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
8435 * opts.c (debug_type_names): Delete coff entry.
8436 (common_handle_option): Delete OPT_gcoff case.
8437 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
8439 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
8440 * config/cris/cris.h: Delete SDB reference in comment.
8441 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
8442 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
8443 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
8444 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
8446 (ix86_expand_prologue): Likewise.
8447 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
8448 * config/ia64/ia64.h: Likewise.
8449 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
8450 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
8452 * config/mmix/mmix.h: Likewise.
8453 * config/nds32/nds32.c: Likewise.
8454 * config/stormy/storym16.h: Likewise.
8455 * config/visium/visium.h: Likewise.
8456 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
8458 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
8460 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
8461 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
8462 FRAME_POINTER_REGNUM point at high end of local var area.
8464 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
8466 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
8467 Move comment around. Do not reset best_edge for a copiable
8468 destination if the copy would cause a partition change.
8469 (better_edge_p): Remove redundant check.
8471 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8473 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
8475 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
8477 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
8478 for math functions that have _Float<N> and _Float<N>X variants.
8479 (mathfn_built_in_2): Add support for math functions that have
8480 _Float<N> and _Float<N>X variants.
8481 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
8482 (expand_builtin_mathfn_ternary): Add support for fma with
8483 _Float<N> and _Float<N>X variants.
8484 (expand_builtin): Likewise.
8485 (fold_builtin_3): Likewise.
8486 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
8487 create math function _Float<N> and _Float<N>X variants as external
8489 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
8490 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
8491 the __builtin_ prefix and if not strict ansi, without the prefix.
8492 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
8493 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
8494 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
8495 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
8496 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
8497 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
8498 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
8499 function signatures for fma _Float<N> and _Float<N>X variants.
8500 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
8501 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
8502 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
8503 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
8504 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
8505 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
8506 * gencfn-macros.c (print_case_cfn): Add support for math functions
8507 that have _Float<N> and _Float<N>X variants.
8508 (print_define_operator_list): Likewise.
8509 (fltfn_suffixes): Likewise.
8511 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
8512 for math functions that have _Float<N> and _Float<N>X variants.
8513 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
8514 and _Float<N>X variants.
8515 (COPYSIGN): Likewise.
8518 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
8519 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
8521 (integer_valued_read_call_p): Likewise.
8522 * fold-const-call.c (fold_const_call_ss): Likewise.
8523 (fold_const_call_sss): Add support for copysign, fmin, and fmax
8524 _Float<N> and _Float<N>X variants.
8525 (fold_const_call_ssss): Add support for fma _Float<N> and
8526 _Float<N>X variants.
8527 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
8528 support for copysign and fma _Float<N> and _Float<N>X variants.
8529 (backprop::process_builtin_call_use): Likewise.
8530 * tree-call-cdce.c (can_test_argument_range); Add support for
8531 sqrt _Float<N> and _Float<N>X variants.
8532 (edom_only_function): Likewise.
8533 (get_no_error_domain): Likewise.
8534 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
8535 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
8536 copysign _Float<N> and _Float<N>X variants.
8537 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
8538 handled by machine independent code.
8539 (FMAF128): Likewise.
8540 * doc/cpp.texi (Common Predefined Macros): Document defining
8541 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
8542 fma _Float<N> and _Float<N>X variants.
8544 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
8547 * config/i386/i386-modes.def (CCFPU): Remove definition.
8548 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
8549 (ix86_cc_modes_compatible): Ditto.
8550 (ix86_expand_carry_flag_compare): Ditto.
8551 (ix86_expand_int_movcc): Ditto.
8552 (ix86_expand_int_addcc): Ditto.
8553 (ix86_reverse_condition): Ditto.
8554 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
8555 Return true/false for unordered/ordered fp comparisons.
8556 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
8557 (ix86_prepare_fp_compare_args): Update for rename.
8558 (ix86_expand_fp_compare): Update for rename. Generate unordered
8559 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
8560 (ix86_expand_sse_compare_and_jump): Ditto.
8561 * config/i386/predicates.md (fcmov_comparison_operator):
8562 Remove CCFPU mode handling.
8563 (ix86_comparison_operator): Ditto.
8564 (ix86_carry_flag_operator): Ditto.
8565 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
8566 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
8567 (*cmpu<mode>_cc_i387): Ditto.
8568 (FPCMP): Remove mode iterator.
8569 (unord): Remove mode attribute.
8570 (unord_subst): New define_subst transformation
8571 (unord): New define_subst attribute.
8573 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
8574 (*cmpi<unord>xf_i387): Ditto.
8575 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
8576 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
8577 using unord_subst transformation.
8578 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
8579 (round_saeonly): Also handle CCFP mode.
8580 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
8581 Remove UNSPEC_SAHF unspec handling.
8583 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8585 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
8587 2017-10-27 Jeff Law <law@redhat.com>
8589 * gimple-ssa-sprintf.c: Include domwalk.h.
8590 (class sprintf_dom_walker): New class, derived from dom_walker.
8591 (sprintf_dom_walker::before_dom_children): New function.
8592 (struct call_info): Moved into sprintf_dom_walker class
8593 (compute_formath_length, handle_gimple_call): Likewise.
8594 (sprintf_length::execute): Call the dominator walker rather
8595 than walking the statements.
8597 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
8598 gimple statement locations.
8599 (check_array_bounds): Corresponding changes. Get the statement's
8600 location directly from wi->stmt.
8602 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
8605 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
8607 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
8609 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
8610 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
8612 2017-10-27 Jakub Jelinek <jakub@redhat.com>
8615 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
8616 * config/i386/i386.c (maybe_get_pool_constant): Removed.
8617 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
8618 maybe_get_pool_constant.
8619 * config/i386/predicates.md (zero_extended_scalar_load_operand):
8622 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8624 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
8625 2.26 caveat. Update gas and gld versions.
8626 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
8629 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8631 * cgraph.h (set_malloc_flag): Declare.
8632 * cgraph.c (set_malloc_flag_1): New function.
8633 (set_malloc_flag): Likewise.
8634 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
8635 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
8637 (read_ipa_call_summary): Add support for reading is_return_callee.
8638 (write_ipa_call_summary): Stream is_return_callee.
8639 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
8640 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
8641 ipa-prop.h, ipa-fnsummary.h.
8642 (pure_const_names): Change to static.
8643 (malloc_state_e): Define.
8644 (malloc_state_names): Define.
8645 (funct_state_d): Add field malloc_state.
8646 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
8647 (check_retval_uses): New function.
8648 (malloc_candidate_p): Likewise.
8649 (analyze_function): Add support for malloc attribute.
8650 (pure_const_write_summary): Stream malloc_state.
8651 (pure_const_read_summary): Add support for reading malloc_state.
8652 (dump_malloc_lattice): New function.
8653 (propagate_malloc): New function.
8654 (warn_function_malloc): New function.
8655 (ipa_pure_const::execute): Call propagate_malloc and
8656 ipa_free_fn_summary.
8657 (pass_local_pure_const::execute): Add support for malloc attribute.
8658 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
8659 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
8661 2017-10-27 Martin Liska <mliska@suse.cz>
8663 PR gcov-profile/82457
8664 * doc/invoke.texi: Document that one needs a non-strict ISO mode
8665 for fork-like functions to be properly instrumented.
8667 2017-10-27 Richard Biener <rguenther@suse.de>
8670 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
8671 info when we redirected EH.
8673 2017-10-26 Michael Collison <michael.collison@arm.com>
8675 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
8677 (<optab>_trunchf<GPI:mode>2: New pattern.
8678 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
8679 * config/aarch64/iterators.md (wv): New mode attribute.
8680 (vf, VF): New mode attributes.
8681 (vgp, VGP): New mode attributes.
8682 (s): Update attribute with SImode and DImode prefixes.
8684 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
8686 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
8687 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
8688 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
8689 (nios2_option_overide): Initialize it. Don't allow R0-relative
8690 addressing with PIC.
8691 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
8692 (nios2_symbolic_constant_p): Likewise.
8693 (nios2_legitimate_address_p): Likewise.
8694 (nios2_r0rel_section_name_p): New.
8695 (nios2_symbol_ref_in_r0rel_data_p): New.
8696 (nios2_emit_move_sequence): Handle r0rel_constant_p.
8697 (r0rel_constant_p): New.
8698 (nios2_print_operand_address): Handle r0rel_constant_p.
8699 (nios2_cdx_narrow_form_p): Likewise.
8700 * config/nios2/nios2.opt (mr0rel-sec=): New option.
8701 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
8702 (Nios II Options): Document -mr0rel-sec.
8704 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
8706 * config/nios2/nios2.c: Include xregex.h.
8707 (nios2_gprel_sec_regex): New.
8708 (nios2_option_overide): Initialize it. Don't allow GP-relative
8709 addressing with PIC.
8710 (nios2_small_section_name_p): Check for regex match.
8711 * config/nios2/nios2.opt (mgprel-sec=): New option.
8712 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
8713 (Nios II Options): Document -mgprel-sec.
8715 2017-10-26 Jim Wilson <wilson@tuliptree.org>
8717 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
8719 2017-10-26 Tom de Vries <tom@codesourcery.com>
8721 PR tree-optimization/82707
8722 * gimple.c (gimple_copy): Fix unsharing of
8723 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
8725 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
8727 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
8728 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
8729 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
8730 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
8731 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
8732 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
8733 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
8734 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
8735 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
8736 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
8737 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
8738 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
8739 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
8740 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
8741 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
8742 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
8743 _mm512_mask_cmpunord_ps_mask): New intrinsics.
8745 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
8747 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
8749 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
8750 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
8751 warning to rs6000.c. Remove the Undocumented flag, since it has
8753 (-mabi=ibmlongdouble): Likewise.
8754 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
8755 already set, set the default format for long double.
8756 (rs6000_debug_reg_global): Print whether long double is IBM or
8758 (rs6000_option_override_internal): Rework setting long double
8759 format. Only warn if the user is changing the long double default
8760 and they did not use -Wno-psabi.
8761 * doc/invoke.texi (PowerPC options): Update the documentation for
8762 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
8764 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8765 Alan Hayward <alan.hayward@arm.com>
8766 David Sherwood <david.sherwood@arm.com>
8768 * rtl.h (wider_subreg_mode): New function.
8769 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
8770 rather than an unsigned int *.
8771 * ira-color.c (regno_max_ref_width): Replace with...
8772 (regno_max_ref_mode): ...this new variable.
8773 (coalesced_pseudo_reg_slot_compare): Update accordingly.
8774 Use wider_subreg_mode.
8775 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
8776 rather than an unsigned int *.
8777 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
8778 (process_alt_operands): Likewise.
8779 (invariant_p): Likewise.
8780 * lra-spills.c (assign_mem_slot): Likewise.
8781 (add_pseudo_to_slot): Likewise.
8782 * lra.c (collect_non_operand_hard_regs): Likewise.
8783 (add_regs_to_insn_regno_info): Likewise.
8784 * reload1.c (regno_max_ref_width): Replace with...
8785 (regno_max_ref_mode): ...this new variable.
8786 (reload): Update accordingly. Update call to
8787 ira_sort_regnos_for_alter_reg.
8788 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
8789 (init_eliminable_invariants): Update to use regno_max_ref_mode.
8790 (scan_paradoxical_subregs): Likewise.
8792 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8794 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
8795 (aarch64_frame): Add emit_frame_chain boolean.
8796 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8797 Move eh_return case to aarch64_layout_frame.
8798 (aarch64_layout_frame): Initialize emit_frame_chain.
8799 (aarch64_expand_prologue): Use emit_frame_chain.
8801 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8803 * config/aarch64/aarch64.c (aarch64_layout_frame):
8804 Ensure LR is always stored at the bottom of the callee-saves.
8805 Remove rarely used frame layout which saves callee-saves at top of
8806 frame, so the store of LR can be used as a valid probe in all cases.
8808 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8810 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
8811 Improve unaligned TImode/TFmode base/offset split.
8813 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8814 Alan Hayward <alan.hayward@arm.com>
8815 David Sherwood <david.sherwood@arm.com>
8817 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
8818 * combine.c (find_single_use_1): Likewise.
8819 (expand_field_assignment): Likewise.
8820 (move_deaths): Likewise.
8821 * lra-constraints.c (simplify_operand_subreg): Likewise.
8822 (curr_insn_transform): Likewise.
8823 * lra.c (collect_non_operand_hard_regs): Likewise.
8824 (add_regs_to_insn_regno_info): Likewise.
8825 * rtlanal.c (reg_referenced_p): Likewise.
8826 (covers_regno_no_parallel_p): Likewise.
8828 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8830 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
8831 Don't print any bits outside the precision of the value.
8832 * wide-int.cc (test_printing): Add some new tests.
8834 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8836 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
8837 supports -xbrace_comment option.
8838 * configure: Regenerate.
8839 * config.in: Regenerate.
8840 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
8841 (ASM_CPU_SPEC): Use it.
8843 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8845 * target.def (static_rtx_alignment): New hook.
8846 * targhooks.h (default_static_rtx_alignment): Declare.
8847 * targhooks.c (default_static_rtx_alignment): New function.
8848 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
8849 * doc/tm.texi: Regenerate.
8850 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
8851 instead of targetm.constant_alignment. Remove call to
8853 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8854 (cris_preferred_mininum_alignment): New function, split out from...
8855 (cris_constant_alignment): ...here.
8856 (cris_static_rtx_alignment): New function.
8857 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
8859 (ix86_constant_alignment): ...here.
8860 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8861 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8862 (mmix_static_rtx_alignment): New function.
8863 * config/spu/spu.c (spu_static_rtx_alignment): New function.
8864 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8866 2017-10-26 Tamar Christina <tamar.christina@arm.com>
8869 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
8870 Add flag_trapping_math and flag_fp_int_builtin_inexact.
8872 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
8874 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
8875 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
8876 mark as a sign-extending load.
8877 (local_pic_load_u): Define.
8879 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
8882 * fold-const.c (operand_equal_for_comparison_p): Also return true
8883 if ARG0 is a simple variant of ARG1 with narrower precision.
8884 (fold_ternary_loc): Always pass unstripped operands to the predicate.
8886 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
8888 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
8890 * i386.h (processor_costs): Add gather_static, gather_per_elt,
8891 scatter_static, scatter_per_elt.
8892 * x86-tune-costs.h: Add new cost entries.
8894 2017-10-25 Richard Biener <rguenther@suse.de>
8896 * tree-ssa-sccvn.h (vn_eliminate): Declare.
8897 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
8898 class pass_fre): Move to ...
8899 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
8900 class pass_fre): ... here and adjust for statistics.
8902 2017-10-25 Jakub Jelinek <jakub@redhat.com>
8905 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
8906 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
8907 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
8908 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
8911 2017-10-25 Richard Biener <rguenther@suse.de>
8913 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
8914 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
8915 eliminate_push_avail, eliminate_insert): Move inside...
8916 (class eliminate_dom_walker): ... this class in preparation
8918 (fini_eliminate): Remove by merging with ...
8919 (eliminate): ... this function. Adjust for class changes.
8920 (pass_pre::execute): Remove fini_eliminate call.
8921 (pass_fre::execute): Likewise.
8923 2017-10-24 Jakub Jelinek <jakub@redhat.com>
8926 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
8927 (VPERMI2, VPERMI2I): New mode iterators.
8928 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
8929 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
8931 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
8932 mode iterator. Remove 3 old define_insn patterns.
8933 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
8934 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
8935 VPERMI2 mode iterator, remove the other two expanders.
8936 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
8937 to use VPERMI2 mode iterator, add another alternative for vpermi2*
8938 instructions, remove the other two patterns.
8939 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
8940 mode iterator, remove the other two patterns.
8941 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
8942 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
8943 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
8944 and adjust argument order accordingly.
8945 (ix86_expand_vec_perm): Adjust caller.
8946 (expand_vec_perm_1): Likewise.
8947 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
8948 (expand_vec_perm_vpermt2_vpshub2): ... this.
8949 (ix86_expand_vec_perm_const_1): Adjust caller.
8950 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
8953 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
8954 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
8955 (vec_shl_<mode>): Remove unused expander.
8956 (avx512bw_<shift_insn><mode>3): New define_insn.
8957 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
8958 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
8960 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
8963 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
8966 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
8968 PR rtl-optimization/82396
8969 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
8970 (autopref_multipass_init): Simplify initialization.
8971 (autopref_rank_data): Simplify sort order.
8972 * gcc/sched-int.h (autopref_multipass_data_): Remove
8973 multi_mem_insn_p, min_offset and max_offset.
8975 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
8978 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8979 Check special value of flag_omit_frame_pointer.
8980 (aarch64_can_eliminate): Likewise.
8981 (aarch64_override_options_after_change_1): Simplify handling of
8982 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
8984 2017-10-24 Richard Biener <rguenther@suse.de>
8986 PR tree-optimization/82697
8987 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
8988 zero for conditional load and unconditional store.
8990 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
8992 * doc/install.texi: Document bootstrap-cet.
8994 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
8997 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
8998 ENDBR instruction at function entrance if function is only
9001 2017-10-24 Jakub Jelinek <jakub@redhat.com>
9004 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
9005 patterns to better describe from which operation the CF is computed.
9006 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
9007 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
9008 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
9009 is 0, use _0 suffixed expanders instead of emitting a comparison
9012 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
9014 * config/i386/i386.md(*movsf_internal, *movdf_internal):
9015 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
9017 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
9020 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
9021 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
9022 * loop-iv.c (iv_get_reaching_def): Likewise.
9023 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
9024 variable is promoted and the partition contains undefined values.
9026 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9028 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
9030 (nios2_address_cost): Define.
9031 (nios2_legitimize_address): Recognize (exp + constant) directly.
9032 (TARGET_ADDRESS_COST): Define.
9034 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9036 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
9037 (nios2_symbolic_memory_operand_p): Declare.
9038 (nios2_split_large_constant): Declare.
9039 (nios2_split_symbolic_memory_operand): Declare.
9040 * config/nios2/nios2.c: Adjust includes.
9041 (nios2_symbolic_constant_allowed): New.
9042 (nios2_symbolic_constant_p): New.
9043 (nios2_plus_symbolic_constant_p): New.
9044 (nios2_valid_addr_expr_p): Recognize addresses involving
9046 (nios2_legitimate_address_p): Likewise, also LO_SUM.
9047 (nios2_symbolic_memory_operand_p): New.
9048 (nios2_large_constant_p): New.
9049 (nios2_split_large_constant): New.
9050 (nios2_split_plus_large_constant): New.
9051 (nios2_split_symbolic_memory_operand): New.
9052 (nios2_legitimize_address): Code refactoring. Handle addresses
9053 involving symbolic constants.
9054 (nios2_emit_move_sequence): Likewise.
9055 (nios2_print_operand): Improve error output.
9056 (nios2_print_operand_address): Handle LO_SUM.
9057 (nios2_cdx_narrow_form_p): Likewise.
9058 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
9059 operands involving symbolic constants.
9060 (movhi_internal, movsi_internal): Likewise.
9061 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
9062 (extendhisi2, extendqi<mode>2): Likewise.
9064 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9066 * tree-pass.h (PROP_rtl_split_insns): Define.
9067 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
9069 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
9071 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
9073 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9076 * target.def (const_not_ok_for_debug_p): Default to
9077 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
9078 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
9079 * targhooks.c (default_const_not_ok_for_debug_p): New function.
9080 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
9081 which targetm.const_not_ok_for_debug_p returned true.
9082 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
9084 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
9086 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
9087 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
9088 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
9090 (ix86_const_not_ok_for_debug_p): New function.
9091 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
9092 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
9094 2017-10-23 David Malcolm <dmalcolm@redhat.com>
9097 * system.h: Conditionally include "unique-ptr.h" if
9098 INCLUDE_UNIQUE_PTR is defined.
9099 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
9100 of defining INCLUDE_UNIQUE_PTR before including "system.h".
9102 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
9104 * config/rl78/rl78.md: New define_expand "subdi3".
9106 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
9109 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
9110 DF_REF_INSN if DF_REF_INSN_INFO is false.
9112 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
9114 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
9115 xmm_move instead of sse_move.
9116 (sse_store_index): New function.
9117 (ix86_register_move_cost): Be more sensible about mismatch stall;
9118 model AVX moves correctly; make difference between sse->integer and
9120 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
9121 moves; make difference between SSE and AVX.
9122 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
9123 and zmm_move. Increase size of sse load and store tables;
9124 add unaligned load and store tables; add ssemmx_to_integer.
9125 * x86-tune-costs.h: Update all entries according to real
9126 move latencies from Agner Fog's manual and chip documentation.
9128 2017-10-23 Jakub Jelinek <jakub@redhat.com>
9131 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
9132 * config/i386/constraints.md (Wf): New constraint.
9133 * config/i386/i386.md (UNSPEC_SBB): New unspec.
9134 (cmp<dwi>_doubleword): Removed.
9135 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
9136 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
9137 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
9138 expand with cmp<dwi>_doubleword. For LTU and GEU use
9139 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
9141 * common.opt (gcolumn-info): Enable by default.
9142 * doc/invoke.texi (gcolumn-info): Document new default.
9144 2017-10-23 Richard Biener <rguenther@suse.de>
9146 PR tree-optimization/82672
9147 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
9148 Fold the stmt if we propagated into it.
9150 2017-10-23 Richard Biener <rguenther@suse.de>
9152 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
9153 (bitmap_remove_expr_from_set): ... this. All callers call this
9154 for non-constant values.
9155 (bitmap_set_subtract): Rename to...
9156 (bitmap_set_subtract_expressions): ... this. Adjust and
9158 (bitmap_set_contains_value): Remove superfluous check.
9159 (bitmap_set_replace_value): Inline into single caller ...
9160 (bitmap_value_replace_in_set): ... here and simplify.
9161 (dependent_clean): Merge into ...
9162 (clean): ... this using an overload. Adjust.
9163 (prune_clobbered_mems): Adjust.
9164 (compute_antic_aux): Likewise.
9165 (compute_partial_antic_aux): Likewise.
9167 2017-10-23 Richard Biener <rguenther@suse.de>
9169 PR tree-optimization/82129
9171 2017-08-01 Richard Biener <rguenther@suse.de>
9173 PR tree-optimization/81181
9174 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9175 (compute_antic): ... end of iteration here.
9177 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9179 * target.def (starting_frame_offset): New hook.
9180 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
9181 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
9182 * doc/tm.texi.in: Regenerate.
9183 * hooks.h (hook_hwi_void_0): Declare.
9184 * hooks.c (hook_hwi_void_0): New function.
9185 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
9186 STARTING_FRAME_OFFSET.
9187 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
9188 * reload1.c (reload): Likewise.
9189 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
9190 instead of STARTING_FRAME_OFFSET.
9191 * function.c (try_fit_stack_local): Likewise.
9192 (assign_stack_local_1): Likewise
9193 (instantiate_virtual_regs): Likewise.
9194 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
9195 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
9196 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
9197 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
9198 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
9199 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
9200 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
9201 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
9202 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
9203 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
9204 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
9205 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
9206 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
9207 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
9208 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
9209 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
9210 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
9211 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
9212 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
9213 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
9214 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
9215 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
9216 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
9217 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
9218 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
9219 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
9220 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
9221 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
9222 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
9223 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
9224 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
9225 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
9226 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
9227 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
9228 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
9229 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
9230 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
9231 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
9232 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
9233 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
9234 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
9235 * config/avr/avr.c (avr_starting_frame_offset): Make static and
9236 return a HOST_WIDE_INT.
9237 (avr_builtin_setjmp_frame_value): Use it instead of
9238 STARTING_FRAME_OFFSET.
9239 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9240 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
9241 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
9243 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9244 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
9245 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
9246 (TARGET_CONSTANT_ALIGNMENT): Redefine.
9247 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
9248 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
9249 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9250 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
9251 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
9252 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9253 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
9254 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
9256 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9257 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
9258 * config/mips/mips.c (mips_compute_frame_info): Refer to
9259 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9260 (mips_starting_frame_offset): New function.
9261 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9262 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
9263 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
9264 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
9265 and return a HOST_WIDE_INT.
9266 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9267 (mmix_initial_elimination_offset): Refer to
9268 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
9269 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
9270 * config/pa/pa.c (pa_starting_frame_offset): New function.
9271 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
9272 (pa_expand_prologue): Likewise.
9273 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9274 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
9275 !FRAME_GROWS_DOWNWARD handling to...
9276 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9277 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
9278 !FRAME_GROWS_DOWNWARD handling to...
9279 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9280 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
9281 !FRAME_GROWS_DOWNWARD handling to...
9282 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9283 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
9285 (rs6000_starting_frame_offset): New function.
9286 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
9287 !FRAME_GROWS_DOWNWARD handling to...
9288 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9289 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
9290 !FRAME_GROWS_DOWNWARD handling to...
9291 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9292 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
9293 !FRAME_GROWS_DOWNWARD handling to...
9294 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
9295 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
9296 (rs6000_starting_frame_offset): New function.
9297 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
9298 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
9299 * config/vax/vax.c (vax_starting_frame_offset): New function.
9300 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
9301 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9302 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
9303 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
9304 (TARGET_STARTING_FRAME_OFFSET): Redefine.
9305 * system.h (STARTING_FRAME_OFFSET): Poison.
9307 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9309 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9310 SCALAR_TYPE_MODE instead of TYPE_MODE.
9312 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9313 Alan Hayward <alan.hayward@arm.com>
9314 David Sherwood <david.sherwood@arm.com>
9316 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
9318 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9319 Alan Hayward <alan.hayward@arm.com>
9320 David Sherwood <david.sherwood@arm.com>
9322 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
9324 2017-10-23 Richard Biener <rguenther@suse.de>
9326 PR tree-optimization/82129
9327 * tree-ssa-pre.c (bitmap_set_and): Remove.
9328 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
9329 canonicalizing expressions in the set to those with lowest
9330 ID rather than taking that from the first edge.
9332 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
9334 * combine.c (rtx_equal_for_field_assignment_p): Use
9335 byte_lowpart_offset.
9337 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9338 Alan Hayward <alan.hayward@arm.com>
9339 David Sherwood <david.sherwood@arm.com>
9341 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
9342 to a SUBREG_PROMOTED_VAR.
9344 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9345 Alan Hayward <alan.hayward@arm.com>
9346 David Sherwood <david.sherwood@arm.com>
9348 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
9349 (expand_debug_source_expr): Likewise.
9350 * combine.c (combine_simplify_rtx): Likewise.
9351 * cse.c (fold_rtx): Likewise.
9352 * optabs.c (expand_float): Likewise.
9353 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9354 (simplify_binary_operation_1): Likewise.
9356 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9357 Alan Hayward <alan.hayward@arm.com>
9358 David Sherwood <david.sherwood@arm.com>
9360 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
9361 (record_promoted_value): Likewise.
9362 * expr.c (expand_expr_real_2): Likewise.
9363 * ree.c (update_reg_equal_equiv_notes): Likewise.
9364 (combine_set_extension): Likewise.
9365 * rtlanal.c (low_bitmask_len): Likewise.
9366 * simplify-rtx.c (neg_const_int): Likewise.
9367 (simplify_binary_operation_1): Likewise.
9369 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9370 Alan Hayward <alan.hayward@arm.com>
9371 David Sherwood <david.sherwood@arm.com>
9373 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
9374 * regcprop.c (maybe_mode_change): Likewise.
9375 * reload1.c (alter_reg): Likewise.
9377 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9379 * inchash.h (inchash::hash::add_wide_int): New function.
9380 * lto-streamer-out.c (hash_tree): Use it.
9382 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
9384 * inchash.h (inchash::hash::add_wide_int): Rename to...
9385 (inchash::hash::add_hwi): ...this.
9386 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
9387 (polymorphic_call_target_hasher::hash): Likewise.
9388 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
9389 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
9390 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
9391 * lto-streamer-out.c (hash_tree): Likewise.
9392 * optc-save-gen.awk: Likewise.
9393 * tree.c (add_expr): Likewise.
9395 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9398 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
9399 for ordered inequality comparisons even with TARGET_IEEE_FP.
9401 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
9404 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
9405 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9406 Expand with cmp<dwi>_doubleword.
9408 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9410 * extend.texi: Add x86 specific to 'nocf_check' attribute.
9411 List CET intrinsics.
9412 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
9413 specific to -fcf-protection option.
9415 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9417 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
9418 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
9419 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
9420 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
9421 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
9422 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
9423 (extra_objs): Add cet.o for Linux/x86 targets.
9424 (tmake_file): Add i386/t-cet for Linux/x86 targets.
9425 * config/i386/cet.c: New file.
9426 * config/i386/cetintrin.h: Likewise.
9427 * config/i386/t-cet: Likewise.
9428 * config/i386/cpuid.h (bit_SHSTK): New.
9429 (bit_IBT): Likewise.
9430 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
9431 pass IBT and SHSTK bits.
9432 * config/i386/i386-builtin-types.def
9433 (VOID_FTYPE_UNSIGNED_PVOID): New.
9434 (VOID_FTYPE_UINT64_PVOID): Likewise.
9435 * config/i386/i386-builtin.def: Add CET intrinsics.
9436 * config/i386/i386-c.c (ix86_target_macros_internal): Add
9437 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
9438 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
9439 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
9441 * config/i386/i386.c (rest_of_insert_endbranch): New.
9442 (pass_data_insert_endbranch): Likewise.
9443 (pass_insert_endbranch): Likewise.
9444 (make_pass_insert_endbranch): Likewise.
9445 (ix86_notrack_prefixed_insn_p): Likewise.
9446 (ix86_target_string): Add -mibt, -mshstk flags.
9447 (ix86_option_override_internal): Add flag_cf_protection
9449 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
9450 (ix86_print_operand): Add 'notrack' prefix output.
9451 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
9452 (ix86_expand_builtin): Expand CET intrinsics.
9453 (x86_output_mi_thunk): Add 'endbranch' instruction.
9454 * config/i386/i386.h (TARGET_IBT): New.
9455 (TARGET_IBT_P): Likewise.
9456 (TARGET_SHSTK): Likewise.
9457 (TARGET_SHSTK_P): Likewise.
9458 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
9459 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
9460 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
9461 (builtin_setjmp_setup): New pattern.
9462 (builtin_longjmp): Likewise.
9463 (rdssp<mode>): Likewise.
9464 (incssp<mode>): Likewise.
9465 (saveprevssp): Likewise.
9466 (rstorssp): Likewise.
9467 (wrss<mode>): Likewise.
9468 (wruss<mode>): Likewise.
9469 (setssbsy): Likewise.
9470 (clrssbsy): Likewise.
9471 (nop_endbr): Likewise.
9472 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
9474 * config/i386/immintrin.h: Include <cetintrin.h>.
9475 * config/i386/linux-common.h
9476 (file_end_indicate_exec_stack_and_cet): New prototype.
9477 (TARGET_ASM_FILE_END): New.
9479 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9481 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
9482 latencies instead of having separate table; make difference between
9483 integer and float costs.
9484 * i386.h (processor_costs): Remove scalar_stmt_cost,
9485 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
9486 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
9488 * x86-tune-costs.h: Remove entries which has been removed in
9489 procesor_costs from all tables; make cond_taken_branch_cost
9490 and cond_not_taken_branch_cost COST_N_INSNS based.
9492 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
9494 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
9496 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9498 * config/i386/i386.md (isa): Remove fma_avx512f.
9499 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
9500 <avx512>_fmadd_<mode>_mask3<round_name>,
9501 <avx512>_fmsub_<mode>_mask<round_name>,
9502 <avx512>_fmsub_<mode>_mask3<round_name>,
9503 <avx512>_fnmadd_<mode>_mask<round_name>,
9504 <avx512>_fnmadd_<mode>_mask3<round_name>,
9505 <avx512>_fnmsub_<mode>_mask<round_name>,
9506 <avx512>_fnmsub_<mode>_mask3<round_name>,
9507 <avx512>_fmaddsub_<mode>_mask<round_name>,
9508 <avx512>_fmaddsub_<mode>_mask3<round_name>,
9509 <avx512>_fmsubadd_<mode>_mask<round_name>,
9510 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
9511 (*vec_widen_umult_even_v16si<mask_name>,
9512 *vec_widen_smult_even_v16si<mask_name>): Likewise.
9513 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
9515 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9517 * extend.texi: Add 'nocf_check' documentation.
9518 * gimple.texi: Add second parameter to
9519 gimple_build_call_from_tree.
9520 * invoke.texi: Add -fcf-protection documentation.
9521 * rtl.texi: Add REG_CALL_NOTRACK documenation.
9523 2017-10-20 Richard Biener <rguenther@suse.de>
9525 PR tree-optimization/82473
9526 * tree-vect-loop.c (vectorizable_reduction): Properly get at
9527 the largest input type.
9529 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
9531 * c-attribs.c (handle_nocf_check_attribute): New function.
9532 (c_common_attribute_table): Add 'nocf_check' handling.
9533 * gimple-parser.c: Add second argument NULL to
9534 gimple_build_call_from_tree.
9535 * attrib.c (comp_type_attributes): Check nocf_check attribute.
9536 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
9538 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
9539 * common.opt: Add fcf-protection flag.
9540 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
9541 * flag-types.h: Add enum cf_protection_level.
9542 * gimple.c (gimple_build_call_from_tree): Add second parameter.
9543 Add 'nocf_check' attribute propagation to gimple call.
9544 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
9545 (gimple_build_call_from_tree): Update prototype.
9546 (gimple_call_nocf_check_p): New function.
9547 (gimple_call_set_nocf_check): Likewise.
9548 * gimplify.c: Add second argument to gimple_build_call_from_tree.
9549 * ipa-icf.c: Add nocf_check attribute in statement hash.
9550 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
9551 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
9552 * toplev.c (process_options): Add flag_cf_protection handling.
9554 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9556 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
9558 2017-10-20 Richard Biener <rguenther@suse.de>
9560 PR tree-optimization/82603
9561 * tree-if-conv.c (predicate_mem_writes): Make sure to only
9562 remove false predicated stores.
9564 2017-10-20 Richard Biener <rguenther@suse.de>
9566 * graphite-isl-ast-to-gimple.c
9567 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
9568 Remove return value and simplify, dump copied stmt after lhs
9570 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
9571 Reduce dump verbosity.
9572 (gsi_insert_earliest): Likewise.
9573 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
9574 * graphite.c (print_global_statistics): Adjust dumping.
9575 (print_graphite_scop_statistics): Likewise.
9576 (print_graphite_statistics): Do not dump loops here.
9577 (graphite_transform_loops): But here.
9579 2017-10-20 Nicolas Roche <roche@adacore.com>
9581 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
9582 * configure: Regenerate.
9584 2017-10-20 Jakub Jelinek <jakub@redhat.com>
9587 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
9588 functions when optimizing replace GIMPLE_RETURN stmts with
9589 calls to __builtin_unreachable ().
9592 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
9593 for -fsanitize=thread link of executables.
9594 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
9595 link of executables.
9598 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
9600 (<shift_insn><mode>3<mask_name>): Change the last of the 3
9601 define_insns for logical vector shifts to use VI248_AVX512BW
9602 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
9603 condition, useless isa and prefix attributes. Change the first
9604 2 of these define_insns to ...
9605 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
9606 define_insn for avx512vl.
9607 (<shift_insn><mode>3): ... and this, new define_insn without
9608 masking for non-avx512vl.
9611 * config/i386/sse.md (*andnot<mode>3,
9612 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
9613 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
9614 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
9615 not applied use empty suffix even for TARGET_AVX512VL.
9616 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
9617 is applied, supply evex,evex or evex,evex,evex instead of just
9620 2017-10-20 Julia Koval <julia.koval@intel.com>
9622 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
9623 (OPTION_MASK_ISA_GFNI_UNSET): New.
9624 (ix86_handle_option): Handle OPT_mgfni.
9625 * config/i386/cpuid.h (bit_GFNI): New.
9626 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
9627 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
9628 * config/i386/i386.c (ix86_target_string): Add -mgfni.
9629 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
9630 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
9631 * config/i386/i386.opt: Add mgfni.
9633 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
9635 * config/msp430/msp430.c (msp430_option_override): Disable
9636 -fdelete-null-pointer-checks.
9637 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
9639 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9641 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
9642 of x87 and SSE instructions.
9644 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
9646 * asan.c (create_cond_insert_point): Do not update edge count.
9647 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
9648 (afdo_propagate_circuit): Likewise.
9649 (afdo_calculate_branch_prob): Likewise.
9650 (afdo_annotate_cfg): Likewise.
9651 * basic-block.h (struct edge_def): Remove count.
9652 (edge_def::count): New accessor.
9653 * bb-reorder.c (rotate_loop): Update.
9654 (find_traces_1_round): Update.
9655 (connect_traces): Update.
9656 (sanitize_hot_paths): Update.
9657 * cfg.c (unchecked_make_edge): Update.
9658 (make_single_succ_edge): Update.
9659 (check_bb_profile): Update.
9660 (dump_edge_info): Update.
9661 (update_bb_profile_for_threading): Update.
9662 (scale_bbs_frequencies_int): Update.
9663 (scale_bbs_frequencies_gcov_type): Update.
9664 (scale_bbs_frequencies_profile_count): Update.
9665 (scale_bbs_frequencies): Update.
9666 * cfganal.c (connect_infinite_loops_to_exit): Update.
9667 * cfgbuild.c (compute_outgoing_frequencies): Update.
9668 (find_many_sub_basic_blocks): Update.
9669 * cfgcleanup.c (try_forward_edges): Update.
9670 (try_crossjump_to_edge): Update
9671 * cfgexpand.c (expand_gimple_cond): Update
9672 (expand_gimple_tailcall): Update
9673 (construct_exit_block): Update
9674 * cfghooks.c (verify_flow_info): Update
9675 (redirect_edge_succ_nodup): Update
9676 (split_edge): Update
9677 (make_forwarder_block): Update
9678 (duplicate_block): Update
9679 (account_profile_record): Update
9680 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
9681 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
9682 * cfgloopmanip.c (scale_loop_profile): Update.
9684 (lv_adjust_loop_entry_edge): Update.
9685 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9686 (force_nonfallthru_and_redirect): Update.
9687 (purge_dead_edges): Update.
9688 (rtl_flow_call_edges_add): Update.
9689 * cgraphunit.c (init_lowered_empty_function): Update.
9690 (cgraph_node::expand_thunk): Update.
9691 * gimple-pretty-print.c (dump_probability): Update.
9692 (dump_edge_probability): Update.
9693 * gimple-ssa-isolate-paths.c (isolate_path): Update.
9694 * haifa-sched.c (sched_create_recovery_edges): Update.
9695 * hsa-gen.c (convert_switch_statements): Update.
9696 * ifcvt.c (dead_or_predicable): Update.
9697 * ipa-inline-transform.c (inline_transform): Update.
9698 * ipa-split.c (split_function): Update.
9699 * ipa-utils.c (ipa_merge_profiles): Update.
9700 * loop-doloop.c (add_test): Update.
9701 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
9702 * lto-streamer-in.c (input_cfg): Update.
9703 (input_function): Update.
9704 * lto-streamer-out.c (output_cfg): Update.
9705 * modulo-sched.c (sms_schedule): Update.
9706 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
9707 * predict.c (maybe_hot_edge_p): Update.
9708 (unlikely_executed_edge_p): Update.
9709 (probably_never_executed_edge_p): Update.
9710 (dump_prediction): Update.
9711 (drop_profile): Update.
9712 (propagate_unlikely_bbs_forward): Update.
9713 (determine_unlikely_bbs): Update.
9714 (force_edge_cold): Update.
9715 * profile.c (compute_branch_probabilities): Update.
9716 * reg-stack.c (better_edge): Update.
9717 * shrink-wrap.c (handle_simple_exit): Update.
9718 * tracer.c (better_p): Update.
9719 * trans-mem.c (expand_transaction): Update.
9720 (split_bb_make_tm_edge): Update.
9721 * tree-call-cdce.c: Update.
9722 * tree-cfg.c (gimple_find_sub_bbs): Update.
9723 (gimple_split_edge): Update.
9724 (gimple_duplicate_sese_region): Update.
9725 (gimple_duplicate_sese_tail): Update.
9726 (gimple_flow_call_edges_add): Update.
9727 (insert_cond_bb): Update.
9728 (execute_fixup_cfg): Update.
9729 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
9730 * tree-complex.c (expand_complex_div_wide): Update.
9731 * tree-eh.c (lower_resx): Update.
9732 (unsplit_eh): Update.
9733 (cleanup_empty_eh_move_lp): Update.
9734 * tree-inline.c (copy_edges_for_bb): Update.
9735 (freqs_to_counts): Update.
9736 (copy_cfg_body): Update.
9737 * tree-ssa-dce.c (remove_dead_stmt): Update.
9738 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
9739 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
9740 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
9741 (unloop_loops): Update.
9742 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9743 * tree-ssa-loop-split.c (connect_loops): Update.
9744 (split_loop): Update.
9745 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
9746 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
9747 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
9748 * tree-ssa-reassoc.c (branch_fixup): Update.
9749 * tree-ssa-tail-merge.c (replace_block_by): Update.
9750 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
9751 (compute_path_counts): Update.
9752 (update_profile): Update.
9753 (recompute_probabilities): Update.
9754 (update_joiner_offpath_counts): Update.
9755 (estimated_freqs_path): Update.
9756 (freqs_to_counts_path): Update.
9757 (clear_counts_path): Update.
9758 (ssa_fix_duplicate_block_edges): Update.
9759 (duplicate_thread_path): Update.
9760 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
9761 (case_bit_test_cmp): Update.
9762 (collect_switch_conv_info): Update.
9763 (gen_inbound_check): Update.
9764 (do_jump_if_equal): Update.
9765 (emit_cmp_and_jump_insns): Update.
9766 * tree-tailcall.c (decrease_profile): Update.
9767 (eliminate_tail_call): Update.
9768 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
9769 (vect_do_peeling): Update.
9770 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9771 * ubsan.c (ubsan_expand_null_ifn): Update.
9772 (ubsan_expand_ptr_ifn): Update.
9773 * value-prof.c (gimple_divmod_fixed_value): Update.
9774 (gimple_mod_pow2): Update.
9775 (gimple_mod_subtract): Update.
9776 (gimple_ic): Update.
9777 (gimple_stringop_fixed_value): Update.
9779 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9782 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
9784 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
9786 PR rtl-optimization/82395
9787 * ira-color.c (allocno_priority_compare_func): Fix comparison step
9788 based on non_spilled_static_chain_regno_p.
9790 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9792 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
9793 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
9794 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
9796 2017-10-19 Martin Sebor <msebor@redhat.com>
9798 PR tree-optimization/82596
9799 * tree.c (array_at_struct_end_p): Handle STRING_CST.
9801 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9803 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
9804 (get_mem_refs_of_builtin_call): Likewise.
9805 * builtins.c (expand_builtin_apply): Adjust call to
9806 allocate_dynamic_stack_space.
9807 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
9808 the third argument to allocate_dynamic_stack_space, otherwise -1.
9809 (expand_builtin): Deal with all alloca variants.
9810 (is_inexpensive_builtin): Likewise.
9811 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
9812 * calls.c (special_function_p): Deal with all alloca variants.
9813 (initialize_argument_information): Adjust call to
9814 allocate_dynamic_stack_space.
9815 (expand_call): Likewise.
9816 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
9817 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
9818 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
9819 use it for the stack usage computation.
9820 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
9821 * function.c (gimplify_parameters): Call build_alloca_call_expr.
9822 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
9823 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
9824 (in_loop_p): Remove first argument and useless check.
9825 (pass_walloca::execute): Remove useless test and adjust call to above.
9826 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
9827 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
9828 (gimplify_call_expr): Deal with all alloca variants.
9829 * hsa-gen.c (gen_hsa_alloca): Likewise.
9830 (gen_hsa_insns_for_call): Likewise.
9831 * ipa-pure-const.c (special_builtin_state): Likewise.
9832 * tree-chkp.c (chkp_build_returned_bound): Likewise.
9833 * tree-object-size.c (alloc_object_size): Likewise.
9834 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
9835 (call_may_clobber_ref_p_1): Likewise.
9836 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
9837 (ccp_fold_stmt): Likewise.
9838 (optimize_stack_restore): Likewise.
9839 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
9840 (mark_all_reaching_defs_necessary_1): Likewise.
9841 (propagate_necessity): Likewise.
9842 (eliminate_unnecessary_stmts): Likewise.
9843 * tree.c (build_common_builtin_nodes): Build
9844 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
9845 (build_alloca_call_expr): New function.
9846 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
9847 (CASE_BUILT_IN_ALLOCA): Likewise.
9848 (build_alloca_call_expr): Declare.
9849 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
9851 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9854 * dwarf2out.c (new_die_raw): New static inline function.
9855 (new_die): Use it to create the DIE.
9856 (add_AT_external_die_ref): Likewise.
9857 (clone_die): Likewise.
9858 (clone_as_declaration): Likewise.
9859 (dwarf2out_vms_debug_main_pointer): Likewise.
9860 (base_type_die): Likewise. Remove early return for corner cases.
9861 Do not call add_pubtype on the DIE here.
9862 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
9863 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
9864 typedefs for base types with DW_AT_endianity. Make sure a DIE with
9865 native order exists for base types, attach the DIE manually and call
9866 add_pubtype on it. Do not equate a reverse order DIE to the type.
9868 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
9870 * config/arm/arm.c (align_ok_ldrd_strd): New function.
9871 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
9873 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
9875 2017-10-19 Jakub Jelinek <jakub@redhat.com>
9877 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
9878 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
9879 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
9880 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
9881 * opts.c (sanitizer_opts): Add builtin.
9882 * ubsan.c (instrument_builtin): New function.
9883 (pass_ubsan::execute): Call it.
9884 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
9885 * doc/invoke.texi: Document -fsanitize=builtin.
9887 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
9888 builtins, store max (log2 (align), 0) into uchar field instead of
9889 align into uptr field.
9890 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
9891 store uchar 0 field instead of uptr 0 field.
9892 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
9893 instead of passing one address of struct with 2 locations pass
9894 two addresses of structs with 1 location each.
9895 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
9896 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
9897 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
9898 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
9899 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
9900 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
9901 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
9902 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
9904 2017-10-19 Martin Liska <mliska@suse.cz>
9907 * file-find.c (remove_prefix): Remove.
9908 * file-find.h (remove_prefix): Likewise.
9909 * gcc-ar.c: Remove smartness of lookup.
9911 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
9913 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
9914 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
9915 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
9917 2017-10-19 Jakub Jelinek <jakub@redhat.com>
9920 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
9921 (setcc + and to xor + setcc): New peephole2.
9923 2017-10-19 Tom de Vries <tom@codesourcery.com>
9925 * doc/sourcebuild.texi (Test Directives, Variants of
9926 dg-require-support): Add dg-require-stack-size.
9928 2017-10-19 Martin Liska <mliska@suse.cz>
9931 * gimplify.c (gimplify_decl_expr): Do not instrument variables
9932 that have a large alignment.
9933 (gimplify_target_expr): Likewise.
9935 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
9937 PR rtl-optimization/82602
9938 * ira.c (rtx_moveable_p): Return false for volatile asm.
9940 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
9943 * config/i386/i386-modes.def (CCGZ): New CC mode.
9944 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
9945 * config/i386/predicates.md (ix86_comparison_operator):
9947 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9948 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
9949 with double-word subtraction.
9950 (put_condition_code): Handle CCGZmode.
9952 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
9954 * wide-int.cc (debug (const wide_int &)): New.
9955 (debug (const wide_int *)): New.
9956 (debug (const widest_int &)): New.
9957 (debug (const widest_int *)): New.
9959 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
9962 * lra-constraints.c (curr_insn_transform): Use non-input operand
9963 instead of output one for matched reload.
9965 2017-10-18 Bin Cheng <bin.cheng@arm.com>
9967 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
9968 (tree-ssa-loop-ivopts.h): New header file.
9969 (struct builtin_info): New fields.
9970 (classify_builtin_1): Compute and record base and offset parts for
9971 memset builtin partition by calling strip_offset.
9972 (offset_cmp, fuse_memset_builtins): New functions.
9973 (finalize_partitions): Fuse adjacent memset partitions by calling
9975 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
9976 Expose the interface.
9977 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
9979 2017-10-18 Bin Cheng <bin.cheng@arm.com>
9981 PR tree-optimization/82574
9982 * tree-loop-distribution.c (find_single_drs): New parameter. Check
9983 that data reference must be executed exactly once per iteration
9984 against the outermost loop in nest.
9985 (classify_partition): Update call to above function.
9987 2017-10-18 Richard Biener <rguenther@suse.de>
9989 PR tree-optimization/82591
9990 * graphite.c (graphite_transform_loops): Move code gen message
9992 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9993 Here. Handle scop_to_isl_ast failing.
9994 (scop_to_isl_ast): Limit the number of ISL operations.
9996 2017-10-18 Richard Biener <rguenther@suse.de>
9998 * graphite-isl-ast-to-gimple.c
9999 (translate_isl_ast_to_gimple::set_rename): Simplify.
10000 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
10001 (graphite_copy_stmts_from_block): ... here.
10002 (copy_bb_and_scalar_dependences): Simplify.
10003 (add_parameters_to_ivs_params): Canonicalize.
10004 (generate_entry_out_of_ssa_copies): Simplify.
10005 * graphite-sese-to-poly.c (extract_affine_name): Simplify
10006 by passing in ISL dimension.
10007 (parameter_index_in_region_1): Rename to ...
10008 (parameter_index_in_region): ... this.
10009 (extract_affine): Adjust assert, pass down parameter index.
10010 (add_param_constraints): Use range-info when available.
10011 (build_scop_context): Adjust.
10012 * sese.c (new_sese_info): Adjust.
10013 (free_sese_info): Likewise.
10014 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
10015 Remove unused typedefs.
10016 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
10018 2017-10-18 Martin Liska <mliska@suse.cz>
10020 * combine.c (simplify_compare_const): Add gcc_fallthrough.
10022 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10024 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
10025 (s390_sched_init): Do not reset s390_sched_state if we entered the
10026 current basic block via a fallthru edge and all others are unlikely.
10028 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
10030 * config/s390/s390.c (NUM_SIDES): New variable.
10031 (LONGRUNNING_THRESHOLD): New variable.
10032 (LATENCY_FACTOR): New variable.
10033 (s390_sched_score): Decrease score for long-running instructions on
10035 (s390_sched_variable_issue): Perform bookkeeping for long-running
10038 2017-10-18 Richard Biener <rguenther@suse.de>
10040 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
10041 Simplify with removal of the parameter rename map.
10042 (set_rename): Likewise.
10043 (should_copy_to_new_region): Likewise.
10044 (graphite_copy_stmts_from_block): Likewise.
10045 (copy_bb_and_scalar_dependences): Remove initialization of
10046 unused copied_bb_map.
10047 (copy_def): Remove.
10048 (copy_internal_parameters): Likewise.
10049 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
10050 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
10051 Use INTEGRAL_TYPE_P.
10052 (parameter_index_in_region_1): Rename to ...
10053 (assign_parameter_index_in_region): ... this. Assert we have
10054 a parameter we handle.
10055 (scan_tree_for_params): Adjust.
10056 * sese.h (parameter_rename_map_t): Remove.
10057 (struct sese_info_t): Remove unused parameter_rename_map and
10058 copied_bb_map members.
10059 * sese.c (new_sese_info): Adjust.
10060 (free_sese_info): Likewise.
10062 2017-10-18 Martin Liska <mliska@suse.cz>
10065 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
10066 on an abnormal edge.
10068 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
10070 * doc/invoke.texi (ffunction-sections and fdata-sections):
10073 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10075 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
10076 the use statement can throw internally.
10078 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
10080 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
10081 any RTX present on the RHS of a SET.
10082 * compare-elim.c (try_eliminate_compare): Restore comment.
10084 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10086 * langhooks.h (struct lang_hooks): Document that tree_size langhook
10087 may be also called on tcc_type nodes.
10088 * langhooks.c (lhd_tree_size): Likewise.
10090 2017-10-17 David Malcolm <dmalcolm@redhat.com>
10092 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
10093 format_warning_at_substring.
10094 (maybe_warn): Convert source_range * param to a location_t. Pass
10095 UNKNOWN_LOCATION rather than NULL to fmtwarn.
10096 (format_directive): Remove code to extract source_ranges and
10097 source_range * in favor of just a location_t.
10098 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
10100 * substring-locations.c (format_warning_va): Convert
10101 source_range * param to a location_t.
10102 (format_warning_at_substring): Likewise.
10103 * substring-locations.h (format_warning_va): Likewise.
10104 (format_warning_at_substring): Likewise.
10106 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
10108 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
10110 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
10111 and scatter/gather ops.
10113 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
10114 vec_gather_load and vec_scatter_store.
10115 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
10116 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
10117 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
10118 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
10119 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
10120 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
10122 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
10124 * reg-stack.c (compare_for_stack_reg): Add bool argument.
10125 Detect FTST instruction and handle its register pops. Only pop
10126 second operand if can_pop_second_op is true.
10127 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
10128 set can_pop_second_op to false in the compare_for_stack_reg call.
10130 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
10131 output_fp_compare for stack register operands.
10132 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
10133 instructions here. Do not emit stack register pops here. Assert
10134 that FCOMPP pops next to top stack register. Rewrite function.
10136 2017-10-17 Nathan Sidwell <nathan@acm.org>
10138 PR middle-end/82577
10139 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
10140 use DECL_ASSEMBLER_NAME_RAW.
10142 PR middle-end/82546
10143 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
10146 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
10147 Wilco Dijkstra <wilco.dijkstra@arm.com>
10149 * builtins.c (expand_builtin_update_setjmp_buf): Add a
10150 converstion to Pmode from the buf_addr.
10152 2017-10-17 Richard Biener <rguenther@suse.de>
10154 * graphite-dependences.c (scop_get_reads_and_writes): Change
10155 output parameters to references.
10157 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
10159 PR 71026/tree-optimization
10160 * fold-const.c (distribute_real_division): Removed.
10161 (fold_binary_loc): Remove calls to distribute_real_divison.
10163 2017-10-17 Richard Biener <rguenther@suse.de>
10165 * graphite-scop-detection.c
10166 (scop_detection::stmt_has_simple_data_refs_p): Always use
10167 the full nest as region.
10168 (try_generate_gimple_bb): Likewise.
10169 * sese.c (scalar_evolution_in_region): Simplify now that
10170 SCEV can handle instantiation in regions.
10171 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
10172 in the non-loop part of a function if requested.
10174 2017-10-17 Richard Biener <rguenther@suse.de>
10176 PR tree-optimization/82563
10177 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
10179 (graphite_regenerate_ast_isl): Call it.
10180 * graphite-scop-detection.c (build_scops): Remove entry edge split.
10182 2017-10-17 Jakub Jelinek <jakub@redhat.com>
10184 PR tree-optimization/82549
10185 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
10186 Formatting fixes. Instead of calling make_bit_field_ref with negative
10189 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
10191 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
10192 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
10193 _mm_maskz_reduce_ss): New.
10194 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
10195 __builtin_ia32_reducess_mask): Ditto..
10196 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
10197 * config/i386/sse.md (reduces<mode>): Renamed to ...
10198 (reduces<mode><mask_scalar_name>): ... this.
10199 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
10201 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
10202 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
10204 2017-10-16 David Malcolm <dmalcolm@redhat.com>
10206 * Makefile.in (OBJS): Add unique-ptr-tests.o.
10207 * selftest-run-tests.c (selftest::run_tests): Call
10208 selftest::unique_ptr_tests_cc_tests.
10209 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
10210 * unique-ptr-tests.cc: New file.
10212 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
10215 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10217 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10218 (make_hard_regno_born, make_hard_regno_dead): Update
10219 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
10221 2017-10-16 Jeff Law <law@redhat.com>
10223 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10225 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
10227 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
10229 2017-10-16 Olivier Hainque <hainque@adacore.com>
10231 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
10232 with_cpu if we were configured for an e500v2 target cpu name.
10234 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
10236 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
10237 * doc/invoke.texi: Document +nodsp as a valid extension for
10240 2017-10-16 Martin Liska <mliska@suse.cz>
10242 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
10243 (test_set_range): Likewise.
10244 (test_range_functions): Rename to ...
10245 (test_bit_in_range): ... this.
10246 (sbitmap_c_tests): Add new test.
10248 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10250 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
10252 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
10253 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
10255 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10257 * config/aarch64/aarch64-builtins.c
10258 (aarch64_types_quadopu_lane_qualifiers): New.
10259 (TYPES_QUADOPU_LANE): New.
10260 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
10261 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
10262 (aarch64_<sur>dot_laneq<vsi2qi>): New.
10263 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
10264 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
10265 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
10266 (Vdottype, DOTPROD): New.
10267 (sur): Add SDOT and UDOT.
10269 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10271 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
10272 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
10273 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
10274 Add TARGET_DOTPROD.
10275 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
10276 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
10277 Enable TARGET_DOTPROD.
10278 (cortex-a75.cortex-a55): Likewise.
10279 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
10281 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10283 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
10284 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
10286 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
10288 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
10289 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
10290 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
10291 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
10292 * config/arm/types.md (neon_dot, neon_dot_q): New.
10293 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
10295 2017-10-16 Tamar Christina <tamar.christina@arm.com>
10297 * config/arm/arm.h (TARGET_DOTPROD): New.
10298 * config/arm/arm.c (arm_arch_dotprod): New.
10299 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
10300 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
10301 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
10302 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
10303 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
10304 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
10305 * doc/invoke.texi (armv8.2-a): Document dotprod
10307 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10309 * i386.c (ix86_vec_cost): New function.
10310 (ix86_rtx_costs): Handle vector operations better.
10311 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
10312 * x86-tune-costs.h: Add new costs to all tables.
10314 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10316 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
10318 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
10319 divsd, sqrtss and sqrtsd
10320 * x86-tune-costs.h: Add new entries to all costs.
10321 (znver1_cost): Fix to match real instruction latencies.
10323 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10324 Michael Collison <michael.collison@arm.com>
10326 * compare-elim.c: Include emit-rtl.h.
10327 (can_merge_compare_into_arith): New function.
10328 (try_validate_parallel): Likewise.
10329 (try_merge_compare): Likewise.
10330 (try_eliminate_compare): Call the above when no previous clobber
10332 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
10335 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10337 PR middle-end/62263
10338 PR middle-end/82498
10339 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
10340 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
10342 PR middle-end/62263
10343 PR middle-end/82498
10344 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
10345 to be any operand_equal_p operands. For & (B - 1) require
10346 B to be power of 2. Recognize
10347 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
10349 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10352 * optabs.c (expand_memory_blockage): Fix call of
10353 targetm.have_memory_blockage.
10355 2017-10-14 Jakub Jelinek <jakub@redhat.com>
10358 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
10359 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
10360 objects to extra_objs instead of overwriting it.
10362 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10364 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
10365 Use any_fp_register_operand as operand[3] predicate. Simplify
10366 equality test for operands[2] and operands[4] memory location.
10367 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
10368 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
10369 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
10370 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
10371 any_fp_register_operand as operand[1] predicate. Simplify
10372 equality test for operands[0] and operands[3] memory location.
10373 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
10374 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
10375 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
10377 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
10379 * target-insns.def: Add memory_blockage.
10380 * optabs.c (expand_memory_blockage): New function.
10381 (expand_asm_memory_barrier): Rename ...
10382 (expand_asm_memory_blockage): ... to this.
10383 (expand_mem_thread_fence): Call expand_memory_blockage
10384 instead of expand_asm_memory_barrier.
10385 (expand_mem_singnal_fence): Ditto.
10386 (expand_atomic_load): Ditto.
10387 (expand_atomic_store): Ditto.
10388 * doc/md.texi (Standard Pattern Names For Generation):
10389 Document memory_blockage instruction pattern.
10391 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
10393 * config/rl78/rl78.c (rl78_emit_libcall): New function.
10394 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
10395 * config/rl78/rl78.md: New define_expand "adddi3".
10397 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10399 * cfghooks.c (verify_flow_info): Disable check that all probabilities
10402 2017-10-13 Jeff Law <law@redhat.com>
10404 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
10406 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10409 * internal-fn.c (expand_mul_overflow): If both operands have
10410 the same highpart of -1 or 0 and the topmost bit of lowpart
10411 is different, overflow is if res <= 0 rather than res < 0.
10413 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
10415 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
10416 TARGET_P9_VECTOR code for unaligned_load case.
10418 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
10420 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
10422 2017-10-13 Nathan Sidwell <nathan@acm.org>
10424 * tree-core.h (tree_contains_struct): Make bool.
10425 * tree.c (tree_contains_struct): Likewise.
10426 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
10427 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
10428 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
10429 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
10430 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
10432 2017-10-13 Richard Biener <rguenther@suse.de>
10434 * graphite-isl-ast-to-gimple.c
10435 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
10436 parameters and dominance check.
10437 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
10438 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
10439 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
10440 Do not update SSA form here or do intermediate IL verification.
10441 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
10442 (graphite_initialize): Remove check on the number of loops in
10443 the function and inline into graphite_transform_loops.
10444 (graphite_finalize): Inline into graphite_transform_loops.
10445 (graphite_transform_loops): Perform SSA update and IL verification
10447 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
10449 2017-10-13 Richard Biener <rguenther@suse.de>
10451 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
10452 graphite_expression_type_precision): Avoid global constructor
10454 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
10455 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
10456 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
10457 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
10458 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
10459 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
10461 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
10464 * config/i386/i386.h (ix86_red_zone_size): New.
10465 * config/i386/i386.md (push peephole2s): Replace
10466 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
10468 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10469 Alan Hayward <alan.hayward@arm.com>
10470 David Sherwood <david.sherwood@arm.com>
10472 * combine.c (can_change_dest_mode): Reject changes in
10473 REGMODE_NATURAL_SIZE.
10475 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10476 Alan Hayward <alan.hayward@arm.com>
10477 David Sherwood <david.sherwood@arm.com>
10479 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
10480 (expand_debug_source_expr): Likewise.
10481 * combine.c (combine_simplify_rtx): Likewise.
10482 * cse.c (fold_rtx): Likewise.
10483 * fwprop.c (canonicalize_address): Likewise.
10484 * targhooks.c (default_shift_truncation_mask): Likewise.
10486 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10487 Alan Hayward <alan.hayward@arm.com>
10488 David Sherwood <david.sherwood@arm.com>
10490 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
10491 (widened_mode): Likewise.
10492 (expand_unop): Likewise.
10493 * ree.c (transform_ifelse): Likewise.
10494 (merge_def_and_ext): Likewise.
10495 (combine_reaching_defs): Likewise.
10496 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10498 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10499 Alan Hayward <alan.hayward@arm.com>
10500 David Sherwood <david.sherwood@arm.com>
10502 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
10503 * combine.c (gen_lowpart_for_combine): Likewise.
10504 * dwarf2out.c (rtl_for_decl_location): Likewise.
10505 * final.c (alter_subreg): Likewise.
10506 * rtlhooks.c (gen_lowpart_general): Likewise.
10507 (gen_lowpart_if_possible): Likewise.
10509 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
10510 Alan Hayward <alan.hayward@arm.com>
10511 David Sherwood <david.sherwood@arm.com>
10513 * calls.c (expand_call): Use subreg_lowpart_offset.
10514 * cse.c (cse_insn): Likewise.
10515 * regcprop.c (copy_value): Likewise.
10516 (copyprop_hardreg_forward_1): Likewise.
10518 2017-10-13 Jakub Jelinek <jakub@redhat.com>
10521 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
10522 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
10523 =Q constraints to +Q and into insn condition add check
10524 that operands[0] and operands[1] are equal.
10525 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
10526 =Q constraints to +Q and into insn condition add check
10527 that operands[0] is equal to either operands[1] or operands[2].
10530 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
10531 instead of handling MINUS_EXPR twice (once for each argument),
10532 canonicalize operand order and handle just once, use rtype where
10533 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
10536 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
10537 any values of __C while still being pattern recognizable as a simple
10538 rotate instruction.
10540 2017-10-13 Richard Biener <rguenther@suse.de>
10542 PR tree-optimization/82451
10544 2017-10-02 Richard Biener <rguenther@suse.de>
10546 PR tree-optimization/82355
10547 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
10548 a mapping for the enclosing loop but avoid generating one for
10549 the loop tree root.
10550 (copy_bb_and_scalar_dependences): Remove premature codegen
10551 error on PHIs in blocks duplicated into multiple places.
10552 * graphite-scop-detection.c
10553 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
10554 in the region use it as loop and nest to analyze the DR in.
10555 (try_generate_gimple_bb): Likewise.
10556 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
10557 (add_loop_constraints): For blocks in a loop not in the region
10558 create a dimension with a single iteration.
10559 * sese.h (gbb_loop_at_index): Remove assert.
10561 * cfgloop.c (loop_preheader_edge): For the loop tree root
10562 return the single successor of the entry block.
10563 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
10564 Reset the SCEV hashtable and niters.
10565 * graphite-scop-detection.c
10566 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
10567 assert that we only have POLYNOMIAL_CHREC that vary in loops
10568 contained in the region.
10569 (scop_detection::graphite_can_represent_expr): Adjust.
10570 (scop_detection::stmt_has_simple_data_refs_p): For loops
10571 not in the region set loop to NULL. The nest is now the
10572 entry edge to the region.
10573 (try_generate_gimple_bb): Likewise.
10574 * sese.c (scalar_evolution_in_region): Adjust for
10575 instantiate_scev change.
10576 * tree-data-ref.h (graphite_find_data_references_in_stmt):
10577 Make nest parameter the edge into the region.
10578 (create_data_ref): Likewise.
10579 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
10580 entry edge into a region and adjust instantiate_scev calls.
10581 (create_data_ref): Likewise.
10582 (graphite_find_data_references_in_stmt): Likewise.
10583 (find_data_references_in_stmt): Pass the loop preheader edge
10584 from the nest argument.
10585 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
10586 parameter the edge into the region.
10587 (instantiate_parameters): Use the loop preheader edge as entry.
10588 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
10590 (get_instantiated_value_entry): Make instantiate_below parameter
10591 the edge into the region.
10592 (instantiate_scev_name): Likewise. Adjust dominance checks,
10593 when we cannot use loop-based instantiation instantiate by
10594 walking use-def chains.
10595 (instantiate_scev_poly): Adjust.
10596 (instantiate_scev_binary): Likewise.
10597 (instantiate_scev_convert): Likewise.
10598 (instantiate_scev_not): Likewise.
10599 (instantiate_array_ref): Remove.
10600 (instantiate_scev_3): Likewise.
10601 (instantiate_scev_2): Likewise.
10602 (instantiate_scev_1): Likewise.
10603 (instantiate_scev_r): Do not blindly handle N-operand trees.
10604 Do not instantiate array-refs. Handle all constants and invariants.
10605 (instantiate_scev): Make instantiate_below parameter
10606 the edge into the region.
10607 (resolve_mixers): Use the loop preheader edge for the region
10608 parameter to instantiate_scev_r.
10609 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
10611 2017-10-13 Richard Biener <rguenther@suse.de>
10613 PR tree-optimization/82525
10614 * graphite-isl-ast-to-gimple.c
10615 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
10617 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
10618 Fail code generation when we cannot represent the isl integer.
10619 (binary_op_to_tree): Elide modulo operations that are no-ops
10620 in the type we code generate. Remove now superfluous code
10623 2017-10-13 Richard Biener <rguenther@suse.de>
10625 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
10626 (scop_detection::harmful_loop_in_region): Remove premature
10627 IV type restriction.
10628 (scop_detection::graphite_can_represent_scev): We can handle
10629 pointer IVs just fine.
10631 2017-10-13 Alan Modra <amodra@gmail.com>
10633 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
10634 "Clobbers and Scratch Registers". Add paragraph on
10635 alternative to clobbers for scratch registers and OpenBLAS
10638 2017-10-13 Alan Modra <amodra@gmail.com>
10640 * doc/extend.texi (Clobbers): Correct vax example. Delete old
10641 example of a memory input for a string of known length. Move
10642 commentary out of table. Add a number of new examples
10643 covering array memory inputs.
10645 2017-10-12 Martin Liska <mliska@suse.cz>
10647 PR tree-optimization/82493
10648 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
10649 (test_range_functions): New function.
10650 (sbitmap_c_tests): Likewise.
10651 * selftest-run-tests.c (selftest::run_tests): Run new tests.
10652 * selftest.h (sbitmap_c_tests): New function.
10654 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
10656 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
10658 * config/rs6000/amo.h: Fix spacing issue.
10660 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10663 * config/i386/i386.md (*ashl<mode>3_mask_1,
10664 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
10665 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
10668 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
10670 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
10671 (profile_probability): Set max_probability
10672 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
10674 * profile-count.c (profile_probability::differs_from_p): Do not
10675 rely on max_probaiblity == 10000
10677 2017-10-12 Jeff Law <law@redhat.com>
10679 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
10682 2017-10-12 Martin Sebor <msebor@redhat.com>
10686 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
10687 (handle_alias_pairs): Call it.
10688 * common.opt (-Wattribute-alias): New option.
10689 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
10690 * doc/invoke.texi (-Wattribute-alias): Document.
10692 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
10695 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10697 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10699 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10700 (make_hard_regno_born, make_hard_regno_dead): Update
10701 bb_killed_pseudos and bb_gen_pseudos.
10703 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
10705 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
10707 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
10709 * config/alpha/alpha.c (alpha_split_conditional_move):
10710 Use std::swap instead of manually swapping.
10711 (alpha_stdarg_optimize_hook): Ditto.
10712 (alpha_canonicalize_comparison): Ditto.
10714 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10716 * tree-loop-distribution.c (struct builtin_info): New struct.
10717 (struct partition): Refactor fields into struct builtin_info.
10718 (partition_free): Free struct builtin_info.
10719 (build_size_arg_loc, build_addr_arg_loc): Delete.
10720 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
10721 information from struct builtin_info.
10722 (find_single_drs): New function refactored from classify_partition.
10723 Also moved builtin validity checks to this function.
10724 (compute_access_range, alloc_builtin): New functions.
10725 (classify_builtin_st, classify_builtin_ldst): New functions.
10726 (classify_partition): Refactor code into functions find_single_drs,
10727 classify_builtin_st and classify_builtin_ldst.
10728 (distribute_loop): Don't do runtime alias check when distributing
10730 (find_seed_stmts_for_distribution): New function.
10731 (pass_loop_distribution::execute): Refactor code finding seed
10732 stmts into above function. Support distribution for the innermost
10733 two-level loop nest. Adjust dump information.
10735 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10737 * tree-loop-distribution.c: Adjust the general comment.
10738 (NUM_PARTITION_THRESHOLD): New macro.
10739 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
10740 (classify_partition): Skip builtin pattern of loop nest's inner loop.
10741 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
10742 in call to build_partition_graph.
10743 (finalize_partitions): New parameter. Make loop distribution more
10744 conservative by fusing more partitions.
10745 (distribute_loop): Don't do runtime alias check in case of loop nest
10747 (find_seed_stmts_for_distribution): New function.
10748 (prepare_perfect_loop_nest): New function.
10749 (pass_loop_distribution::execute): Refactor code finding seed stmts
10750 and loop nest into above functions. Support loop nest distribution.
10751 Adjust dump information accordingly.
10753 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10755 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
10756 and set PTYPE_SEQUENTIAL for merged partition.
10758 2017-10-12 Richard Biener <rguenther@suse.de>
10760 PR tree-optimization/69728
10762 2017-09-19 Richard Biener <rguenther@suse.de>
10764 PR tree-optimization/69728
10765 * graphite-sese-to-poly.c (schedule_error): New global.
10766 (add_loop_schedule): Handle empty domain by failing the
10768 (build_original_schedule): Handle schedule_error.
10770 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
10771 domain by returning an unchanged schedule.
10773 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10775 * genrecog.c (validate_pattern): For VEC_SELECT verify that
10776 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
10778 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
10780 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
10783 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10786 * expr.c (store_field): Don't optimize away bitsize == 0 store
10787 from CALL_EXPR with addressable return type.
10789 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10791 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
10792 * config/rs6000/rs6000.md (sel): Delete mode attribute.
10793 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
10794 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
10795 TARGET_ISEL instead of TARGET_ISEL<sel>.
10797 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
10799 * config/rs6000/rs6000.c
10800 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
10802 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10804 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
10806 (reg_or_cint_operand): Fix comment.
10807 (reg_or_zero_operand): New predicate.
10808 * config/rs6000/rs6000-protos.h (output_isel): Delete.
10809 * config/rs6000/rs6000.c (output_isel): Delete.
10810 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
10811 instead of reg_or_cint_operand. Output instruction directly (not via
10813 (isel_unsigned_<mode>): Ditto.
10814 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
10815 gpc_reg_operand. Add an instruction alternative for this. Output
10816 instruction directly.
10817 (*isel_reversed_unsigned_<mode>): Ditto.
10819 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10821 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
10822 (TARGET_CANONICALIZE_COMPARISON): Define.
10824 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
10827 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
10828 Check whether the dest is REG before adding REG_EQUIV note.
10830 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10833 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10835 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10836 (make_hard_regno_born, make_hard_regno_dead): Update
10837 bb_killed_pseudos and bb_gen_pseudos.
10839 2017-10-11 Nathan Sidwell <nathan@acm.org>
10841 * incpath.h (enum incpath_kind): Name enum, prefix values.
10842 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
10843 * incpath.c (heads, tails): Use INC_MAX.
10844 (add_env_var_paths, add_standard_paths): Use incpath_kind.
10845 (merge_include_chains, split_quote_chain,
10846 register_include_chains): Update incpath_kind names.
10847 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
10848 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
10850 (add_framework_path, darwin_register_objc_includes): Likewise.
10851 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
10853 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10855 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
10856 Do not use float_operator operator predicate.
10857 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
10858 * config/i386/predicates.md (float_operator): Remove predicate.
10860 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10862 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
10863 (*jccxf_i387): Ditto.
10864 (*jcc<mode>_i387): Ditto.
10865 (*jccu<mode>_i387): Ditto.
10866 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
10867 (*jcc_*_i387 splitters): Remove.
10868 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
10869 * config/i386/i386.c (ix86_split_fp_branch): Remove.
10870 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
10873 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10875 * profile-count.h (slow_safe_scale_64bit): New function.
10876 (safe_scale_64bit): New inline.
10877 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
10878 * profile-count.c: Include wide-int.h
10879 (slow_safe_scale_64bit): New.
10881 2017-10-11 Nathan Sidwell <nathan@acm.org>
10883 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
10884 HAS_DECL_ASSEMBLER_NAME_P.
10885 * gimple-expr.c (gimple_decl_printable_name: Check
10886 HAS_DECL_ASSEMBLER_NAME_P too.
10887 * ipa-utils.h (type_in_anonymous_namespace_p): Check
10888 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
10889 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
10890 * passes.c (rest_of_decl_compilation): Check
10891 HAS_DECL_ASSEMBLER_NAME_P too.
10892 * recog.c (verify_changes): Likewise.
10893 * tree-pretty-print.c (dump_decl_name): Likewise.
10894 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
10896 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
10897 (SET_DECL_ASSEMBLER_NAME): Use it.
10898 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
10899 (COPY_DECL_ASSEMBLER_NAME): Likewise.
10900 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
10902 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10904 * config.gcc (i386, x86_64): Add extra objects.
10905 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
10906 (ix86_min_insn_size): Declare.
10907 (ix86_issue_rate): Declare.
10908 (ix86_adjust_cost): Declare.
10909 (ia32_multipass_dfa_lookahead): Declare.
10910 (ix86_macro_fusion_p): Declare.
10911 (ix86_macro_fusion_pair_p): Declare.
10912 (ix86_bd_has_dispatch): Declare.
10913 (ix86_bd_do_dispatch): Declare.
10914 (ix86_core2i7_init_hooks): Declare.
10915 (ix86_atom_sched_reorder): Declare.
10916 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
10917 (COSTS_N_BYTES): Move to x86-tune-costs.h.
10918 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
10919 (rip_relative_addr_p): Rename to ...
10920 (ix86_rip_relative_addr_p): ... this one; export.
10921 (memory_address_length): Update.
10922 (ix86_issue_rate): Move to x86-tune-sched.c.
10923 (ix86_flags_dependent): Move to x86-tune-sched.c.
10924 (ix86_agi_dependent): Move to x86-tune-sched.c.
10925 (exact_dependency_1): Move to x86-tune-sched.c.
10926 (exact_store_load_dependency): Move to x86-tune-sched.c.
10927 (ix86_adjust_cost): Move to x86-tune-sched.c.
10928 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
10929 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
10930 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
10931 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
10932 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
10933 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
10934 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
10935 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
10936 (min_insn_size): Rename to ...
10937 (ix86_min_insn_size): ... this one; export.
10938 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
10939 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
10940 (core2i7_first_cycle_multipass_backtrack): Move to
10941 x86-tune-sched-core.c.
10942 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
10943 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
10944 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
10945 (ix86_avoid_jump_mispredicts): Update.
10946 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
10947 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
10948 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
10949 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
10950 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
10951 (MAX_INSN): Move to ix86-tune-sched-bd.c.
10952 (MAX_IMM): Move to ix86-tune-sched-bd.c.
10953 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
10954 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
10955 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
10956 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
10957 (MAX_STORE): Move to ix86-tune-sched-bd.c.
10958 (BIG): Move to ix86-tune-sched-bd.c.
10959 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
10960 (enum insn_path): Move to ix86-tune-sched-bd.c.
10961 (get_mem_group): Move to ix86-tune-sched-bd.c.
10962 (is_cmp): Move to ix86-tune-sched-bd.c.
10963 (dispatch_violation): Move to ix86-tune-sched-bd.c.
10964 (is_branch): Move to ix86-tune-sched-bd.c.
10965 (is_prefetch): Move to ix86-tune-sched-bd.c.
10966 (init_window): Move to ix86-tune-sched-bd.c.
10967 (allocate_window): Move to ix86-tune-sched-bd.c.
10968 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
10969 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
10970 (process_end_window): Move to ix86-tune-sched-bd.c.
10971 (allocate_next_window): Move to ix86-tune-sched-bd.c.
10972 (find_constant): Move to ix86-tune-sched-bd.c.
10973 (get_num_immediates): Move to ix86-tune-sched-bd.c.
10974 (has_immediate): Move to ix86-tune-sched-bd.c.
10975 (get_insn_path): Move to ix86-tune-sched-bd.c.
10976 (get_insn_group): Move to ix86-tune-sched-bd.c.
10977 (count_num_restricted): Move to ix86-tune-sched-bd.c.
10978 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
10979 (add_insn_window): Move to ix86-tune-sched-bd.c.
10980 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
10981 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
10982 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
10983 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
10984 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
10985 (do_dispatch): Move to ix86-tune-sched-bd.c.
10986 (has_dispatch): Move to ix86-tune-sched-bd.c.
10987 * config/i386/t-i386: Add new object files.
10988 * config/i386/x86-tune-costs.h: New file.
10989 * config/i386/x86-tune-sched-atom.c: New file.
10990 * config/i386/x86-tune-sched-bd.c: New file.
10991 * config/i386/x86-tune-sched-core.c: New file.
10992 * config/i386/x86-tune-sched.c: New file.
10994 2017-10-11 Liu Hao <lh_mouse@126.com>
10996 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
10997 the generic version below instead.
10998 (should_colorize): Recognize Windows consoles as terminals
11000 * pretty-print.c [__MINGW32__] (write_all): New function.
11001 [__MINGW32__] (find_esc_head): Likewise.
11002 [__MINGW32__] (find_esc_terminator): Likewise.
11003 [__MINGW32__] (eat_esc_sequence): Likewise.
11004 [__MINGW32__] (mingw_ansi_fputs): New function that handles
11006 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
11009 2017-10-11 Richard Biener <rguenther@suse.de>
11011 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
11012 Properly call analyze_scalar_evolution with the loop of the stmt.
11014 2017-10-11 Richard Biener <rguenther@suse.de>
11016 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
11017 * tree-core.h (tree_base): Add chrec_var union member.
11018 * tree.h (CHREC_VAR): Remove.
11019 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
11020 * tree-chrec.h (build_polynomial_chrec): Adjust.
11021 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
11022 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
11024 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
11026 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
11027 * match.pd: ... here.
11028 ((T) X == (T) Y): Relax condition.
11030 2017-10-11 Bin Cheng <bin.cheng@arm.com>
11032 PR tree-optimization/82472
11033 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
11035 (break_alias_scc_partitions): Update postorder number.
11037 2017-10-11 Martin Liska <mliska@suse.cz>
11040 * opts.c (parse_no_sanitize_attribute): Do not use error_value
11042 * opts.h (parse_no_sanitize_attribute): Remove last argument.
11044 2017-10-11 Martin Liska <mliska@suse.cz>
11046 * print-rtl.c (print_insn): Move declaration of idbuf
11047 to same scope as name.
11049 2017-10-11 Martin Liska <mliska@suse.cz>
11054 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11055 volatile arguments.
11057 2017-10-11 Martin Liska <mliska@suse.cz>
11060 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
11061 volatile arguments.
11063 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
11065 * config.gcc (default_gnu_indirect_function): Default to yes for
11066 arm*-*-linux* with glibc.
11068 2017-10-11 Richard Biener <rguenther@suse.de>
11070 * tree-scalar-evolution.c (get_scalar_evolution): Handle
11071 default-defs and types we do not want to analyze.
11072 (interpret_loop_phi): Replace unreachable code with an assert.
11073 (compute_scalar_evolution_in_loop): Remove and inline ...
11074 (analyze_scalar_evolution_1): ... here, replacing condition with
11075 what makes the intent clearer. Remove handling of cases
11076 get_scalar_evolution now handles.
11078 2017-10-10 Jim Wilson <wilson@tuliptree.org>
11080 PR rtl-optimization/81434
11081 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
11082 comment for main loop. In sched_group_found if, also add checks for
11083 pass and min_cost_group.
11085 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
11087 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11088 (rs6000_insn_cost): New function.
11089 * config/rs6000/rs6000.md (cost): New attribute.
11091 2017-10-10 Jakub Jelinek <jakub@redhat.com>
11092 H.J. Lu <hongjiu.lu@intel.com>
11096 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
11097 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
11098 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
11099 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
11100 to OPTION_MASK_ISA_AVX512VL - builtins that have both
11101 OPTION_MASK_ISA_MMX and some other bit set require both
11102 mmx and the ISAs without the mmx bit.
11103 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
11104 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
11105 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
11106 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
11107 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
11108 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
11109 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
11110 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
11111 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
11112 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
11113 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
11114 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
11115 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
11116 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
11117 Add OPTION_MASK_ISA_MMX.
11119 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
11121 * config.gcc (armv7*-*-freebsd*): New target.
11122 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
11124 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
11126 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
11127 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
11130 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
11132 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
11133 a template parameter.
11134 (WIDE_INT_REF_FOR): Update accordingly.
11135 * tree.h (wi::int_traits <const_tree>): Delete.
11136 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
11137 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
11138 (wi::tree_to_wide_ref): New typedef.
11139 (wi::to_wide): New function.
11140 * calls.c (get_size_range): Use wi::to_wide when operating on
11141 trees as wide_ints.
11142 * cgraph.c (cgraph_node::create_thunk): Likewise.
11143 * config/i386/i386.c (ix86_data_alignment): Likewise.
11144 (ix86_local_alignment): Likewise.
11145 * dbxout.c (stabstr_O): Likewise.
11146 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
11147 * expr.c (const_vector_from_tree): Likewise.
11148 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
11149 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
11150 (fold_negate_expr_1, int_const_binop_1, const_binop)
11151 (fold_convert_const_int_from_real, optimize_bit_field_compare)
11152 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
11153 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
11154 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
11155 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
11156 (fold_not_const, round_up_loc): Likewise.
11157 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
11158 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
11159 (alloca_call_type): Likewise.
11160 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
11161 * godump.c (go_output_typedef): Likewise.
11162 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
11163 * internal-fn.c (get_min_precision): Likewise.
11164 * ipa-cp.c (ipcp_store_vr_results): Likewise.
11165 * ipa-polymorphic-call.c
11166 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
11167 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
11168 (ipa_modify_call_arguments): Likewise.
11169 * match.pd: Likewise.
11170 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
11171 * print-tree.c (print_node_brief, print_node): Likewise.
11172 * stmt.c (expand_case): Likewise.
11173 * stor-layout.c (layout_type): Likewise.
11174 * tree-affine.c (tree_to_aff_combination): Likewise.
11175 * tree-cfg.c (group_case_labels_stmt): Likewise.
11176 * tree-data-ref.c (dr_analyze_indices): Likewise.
11177 (prune_runtime_alias_test_list): Likewise.
11178 * tree-dump.c (dequeue_and_dump): Likewise.
11179 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
11180 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
11181 * tree-pretty-print.c (dump_generic_node): Likewise.
11182 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
11183 (simple_iv_with_niters): Likewise.
11184 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
11185 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
11186 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
11187 * tree-ssa-loop-niter.c (split_to_var_and_offset)
11188 (refine_value_range_using_guard, number_of_iterations_ne_max)
11189 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
11190 (get_cst_init_from_scev, record_nonwrapping_iv)
11191 (scev_var_range_cant_overflow): Likewise.
11192 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
11193 * tree-ssa-pre.c (compute_avail): Likewise.
11194 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
11195 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
11196 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
11197 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
11198 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
11199 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
11200 (dump_case_nodes, try_switch_expansion): Likewise.
11201 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
11202 (vect_do_peeling): Likewise.
11203 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
11204 * tree-vect-stmts.c (vectorizable_load): Likewise.
11205 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
11206 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
11207 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
11208 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
11209 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
11210 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
11211 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
11212 (evrp_dom_walker::before_dom_children): Likewise.
11213 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
11214 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
11215 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
11216 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
11217 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
11218 (get_range_pos_neg): Likewise.
11219 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
11220 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
11221 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
11222 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
11223 * config/avr/avr.c (avr_fold_builtin): Likewise.
11224 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
11225 * config/msp430/msp430.c (msp430_attr): Likewise.
11226 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
11227 * config/powerpcspe/powerpcspe-c.c
11228 (altivec_resolve_overloaded_builtin): Likewise.
11229 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
11230 (rs6000_expand_ternop_builtin): Likewise.
11231 * config/rs6000/rs6000-c.c
11232 (altivec_resolve_overloaded_builtin): Likewise.
11233 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
11234 (rs6000_expand_ternop_builtin): Likewise.
11235 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
11237 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11239 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
11240 when copying loop nest with only one inner loop.
11242 2017-10-10 Richard Biener <rguenther@suse.de>
11244 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
11245 blocks if SCEV is active.
11246 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
11248 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
11249 (scev_initialize): Assert we are not yet initialized.
11251 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11253 * tree-loop-distribution.c (generate_loops_for_partition): Remove
11254 inner loop's exit stmt by making it always exit the loop, otherwise
11255 we would generate an infinite empty loop.
11257 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11259 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
11260 renaming variables in new preheader if it's deleted.
11262 2017-10-10 Bin Cheng <bin.cheng@arm.com>
11264 * tree-loop-distribution.c (struct partition): Remove unused field
11265 loops of the structure.
11266 (partition_alloc, partition_free): Ditto.
11267 (build_rdg_partition_for_vertex): Ditto.
11269 2017-10-09 Jeff Law <law@redhat.com>
11271 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
11272 return type to match prototype and documentation.
11274 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11276 * config/rs6000/rs6000.c (processor_costs): Move to ...
11277 * config/rs6000/rs6000.h: ... here.
11278 (rs6000_cost): Declare.
11280 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
11282 * except.c (setjmp_fn): New global variable.
11283 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
11284 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
11285 if DONT_USE_BUILTIN_SETJMP is defined.
11287 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11289 * target.def (insn_cost): New hook.
11290 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
11291 * doc/tm.texi: Regenerate.
11292 * rtlanal.c (insn_cost): Use the new hook.
11294 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11296 * combine.c (combine_validate_cost): Compute the new insn_cost,
11297 not just pattern_cost.
11298 (try_combine): Adjust comment.
11300 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
11302 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
11304 * combine.c (uid_insn_cost): Adjust comment.
11305 (combine_validate_cost): Adjust comment. Use pattern_cost instead
11307 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
11308 * dse.c (find_shift_sequence): Ditto.
11309 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
11310 (bb_valid_for_noce_process_p): Use pattern_cost.
11311 * rtl.h (insn_rtx_cost): Delete.
11312 (pattern_cost): New prototype.
11313 (insn_cost): New prototype.
11314 * rtlanal.c (insn_rtx_cost): Rename to...
11315 (pattern_cost): ... this.
11318 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
11320 * config/i386/i386.md (*jcc_2): Remove insn pattern.
11321 (*jcc<mode>_0_r_i387): Ditto.
11322 (*jccxf_r_i387): Ditto.
11323 (*jcc<mode>_r_i387): Ditto.
11324 (*jccu<mode>_r_i387): Ditto.
11325 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
11326 (*jcc): Rename from *jcc_1.
11328 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11330 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
11331 deferred rescans after the lvx/stvx recombination pre-pass.
11333 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
11335 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
11336 memory operation instruction support.
11337 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
11338 (rs6000-ibm-aix[789]*): Likewise.
11339 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
11340 Document new functions.
11342 2017-10-09 Richard Biener <rguenther@suse.de>
11344 PR tree-optimization/82397
11345 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
11346 equality only for semantically equal trees.
11348 2017-10-09 Richard Biener <rguenther@suse.de>
11350 PR tree-optimization/82449
11351 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
11352 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
11353 allow constant addresses.
11354 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
11357 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11359 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
11362 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11365 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
11368 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11371 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
11373 2017-10-09 Jakub Jelinek <jakub@redhat.com>
11376 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
11377 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
11379 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
11381 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
11382 (WI_BINARY_PREDICATE_RESULT): Likewise.
11383 (wi::binary_traits::operator_result): New type.
11384 (wi::binary_traits::predicate_result): Likewise.
11385 (generic_wide_int::operator~, unary generic_wide_int::operator-)
11386 (generic_wide_int::operator==, generic_wide_int::operator!=)
11387 (generic_wide_int::operator&, generic_wide_int::and_not)
11388 (generic_wide_int::operator|, generic_wide_int::or_not)
11389 (generic_wide_int::operator^, generic_wide_int::operator+
11390 (binary generic_wide_int::operator-, generic_wide_int::operator*):
11392 (operator~, unary operator-, operator==, operator!=, operator&)
11393 (operator|, operator^, operator+, binary operator-, operator*): New
11395 * expr.c (get_inner_reference): Use wi::bit_and_not.
11396 * fold-const.c (fold_binary_loc): Likewise.
11397 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
11398 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
11399 (bit_value_binop): Likewise.
11400 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
11401 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
11402 (extract_range_from_binary_expr_1): Likewise.
11403 (masked_increment): Likewise.
11404 (simplify_bit_ops_using_ranges): Likewise.
11406 2017-10-09 Martin Jambor <mjambor@suse.cz>
11409 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
11410 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
11411 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
11413 (hsa_fixup_mov_insn_type): New function.
11414 (hsa_op_with_type::get_in_type): Use it.
11415 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
11416 immediates in an assert.
11417 (hsa_op_with_type::extend_int_to_32bit): New method.
11418 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
11419 types. Convert to dest type if necessary.
11420 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
11421 (reg_for_gimple_ssa): Pass false as min32int to
11422 hsa_type_for_scalar_tree_type.
11423 (gen_hsa_addr): Fixup type when creating addresable temporary.
11424 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
11425 (gen_hsa_unary_operation): Extend operands and convert to dest type if
11426 necessary. Call hsa_fixup_mov_insn_type.
11427 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
11428 extend operands and convert to dest type if necessary.
11429 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
11430 to dest type if necessary.
11431 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
11432 if conversion nt necessary and size matches.
11433 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
11434 to dest type if necessary.
11435 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
11436 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
11438 (gen_hsa_clrsb): Likewise.
11439 (gen_hsa_ffs): Likewise.
11440 (gen_hsa_divmod): Extend operands and convert to dest type if
11442 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
11444 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11446 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
11447 Remove empty default arguments. Use a brace block as output
11449 (conditional return): Ditto.
11451 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
11452 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
11454 (group_ending_nop): Ditto.
11455 (doloop_end): Ditto.
11456 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
11457 (splitters for those): Ditto.
11459 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
11461 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
11462 a conditional jump (and the compare for it) so that pc_rtx is the
11464 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
11465 for the deleted and renamed ctr<mode>_internal[234] patterns.
11466 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
11467 Delete second conditional return pattern.
11468 (ctr<mode>_internal2): Delete this second bdnz pattern.
11469 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
11470 (ctr<mode>_internal4): Delete this second bdz pattern.
11472 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11474 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
11475 (always_initialized_rtx_for_ssa_name_p): New predicate.
11476 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
11477 (finish_out_of_ssa): Free new field of SA.
11478 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
11479 * tree-ssa-coalesce.c: Include tree-ssa.h.
11480 (get_parm_default_def_partitions): Remove extern keyword.
11481 (get_undefined_value_partitions): New function.
11482 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
11483 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
11484 uninitialized bits.
11485 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
11487 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
11489 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
11491 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11493 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
11494 for targets that preffer 128bit.
11496 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
11498 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
11500 2017-10-08 Olivier Hainque <hainque@adacore.com>
11502 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
11503 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
11505 (thumb_set_return_address): Likewise.
11507 2017-10-08 Olivier Hainque <hainque@adacore.com>
11509 * common/config/arm/arm-common.c (arm_except_unwind_info):
11510 Handle DWARF2_UNWIND_INFO.
11512 2017-10-07 Michael Collison <michael.collison@arm.com>
11514 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
11517 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
11519 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
11520 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
11521 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
11522 defined, force the creation of a new block for a dispatch label.
11524 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
11526 * invoke.texi (Wsuggest-attribute=cold): Document.
11527 * common.opt (Wsuggest-attribute=cold): New
11528 * ipa-pure-const.c (warn_function_cold): New function.
11529 * predict.c (compute_function_frequency): Use it.
11530 * predict.h (warn_function_cold): Declare.
11532 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
11534 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
11537 2017-10-06 Martin Liska <mliska@suse.cz>
11539 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
11540 keyword for member functions.
11541 (struct sanopt_tree_couple): New struct.
11542 (struct sanopt_tree_couple_hash): New function.
11543 (struct sanopt_ctx): Add new hash_map.
11544 (has_dominating_ubsan_ptr_check): New function.
11545 (record_ubsan_ptr_check_stmt): Likewise.
11546 (maybe_optimize_ubsan_ptr_ifn): Likewise.
11547 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
11548 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
11550 2017-10-06 Sudakshina Das <sudi.das@arm.com>
11553 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
11554 aarch64_simd_valid_immediate on CONST_VECTORs.
11555 (aarch64_reg_or_bic_imm): Likewise.
11557 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
11559 PR rtl-optimization/82396
11560 * haifa-sched.c (ready_sort_real): Disable qsort checking.
11562 2017-10-06 Sebastian Pop <sebpop@gmail.com>
11564 * graphite-dependences.c (scop_get_reads): Move code to...
11565 (scop_get_must_writes): Move code to...
11566 (scop_get_may_writes): Move code to...
11567 (scop_get_reads_and_writes): ... here.
11568 (scop_get_dependences): Call scop_get_reads_and_writes.
11570 2017-10-06 Jakub Jelinek <jakub@redhat.com>
11572 PR tree-optimization/82434
11573 * fold-const.h (can_native_encode_type_p,
11574 can_native_encode_string_p): Remove.
11575 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
11576 don't encode anything, just return what would be otherwise returned.
11577 (native_encode_fixed, native_encode_complex, native_encode_vector):
11579 (native_encode_string): Likewise. Inline by hand
11580 can_native_encode_string_p.
11581 (can_native_encode_type_p): Remove.
11582 (can_native_encode_string_p): Remove.
11583 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
11584 STRING_CSTs using can_native_encode_string_p, test all
11585 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
11586 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
11587 argument from native_encode_expr.
11588 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
11589 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
11592 2017-10-06 Richard Biener <rguenther@suse.de>
11594 PR tree-optimization/82397
11595 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
11596 operand_equal_p but rely on data_ref_compare_tree for detecting
11598 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
11599 to match up with dr_group_sort_cmp.
11601 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11604 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
11606 * config/s390/s390-builtin-types.def: Regenerate.
11608 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11611 * config/s390/s390-builtin-types.def: Regenerate.
11612 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
11613 Change flag from B_VXE to B_VX.
11614 (s390_vec_min_dbl): Remove B_VXE flag.
11616 2017-10-06 Richard Biener <rguenther@suse.de>
11618 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
11619 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
11620 translate_isl_ast_to_gimple::is_valid_rename,
11621 translate_isl_ast_to_gimple::get_rename,
11622 translate_isl_ast_to_gimple::get_def_bb_for_const,
11623 translate_isl_ast_to_gimple::get_new_name,
11624 translate_isl_ast_to_gimple::collect_all_ssa_names,
11625 translate_isl_ast_to_gimple::copy_loop_phi_args,
11626 translate_isl_ast_to_gimple::collect_all_ssa_names,
11627 translate_isl_ast_to_gimple::copy_loop_phi_args,
11628 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
11629 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
11630 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
11631 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
11632 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
11633 translate_isl_ast_to_gimple::copy_cond_phi_args,
11634 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
11635 translate_isl_ast_to_gimple::edge_for_new_close_phis,
11636 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
11637 translate_isl_ast_to_gimple::rename_uses,
11638 translate_isl_ast_to_gimple::rename_all_uses): Remove.
11639 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
11640 (set_rename_for_each_def): Likewise.
11641 (graphite_copy_stmts_from_block): Handle debug stmt resetting
11642 here. Handle rewriting SCEV analyzable uses here.
11643 (copy_bb_and_scalar_dependences): Generate code for PHI
11645 (graphite_regenerate_ast_isl): Adjust.
11646 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
11647 (add_write, add_read): New functions.
11648 (build_cross_bb_scalars_def): Use it and simplify.
11649 (build_cross_bb_scalars_use): Likewise.
11650 (graphite_find_cross_bb_scalar_vars): Inline into...
11651 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
11652 simulating out-of-SSA. Compute liveout and add dependencies.
11653 (build_scops): Force an empty entry block.
11654 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
11656 (sese_build_liveouts): Declare.
11657 (sese_trivially_empty_bb_p): Likewise.
11658 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
11659 compute liveout and debug_liveout.
11660 (sese_bad_liveouts_use): Remove.
11661 (sese_reset_debug_liveouts_bb): Likewise.
11662 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
11663 (sese_build_liveouts): Build liveout and debug_liveout and store
11665 (new_sese_info): Adjust.
11666 (free_sese_info): Likewise.
11667 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
11668 do not build liveout here.
11669 (move_sese_in_condition): Adjust region entry.
11670 (scev_analyzable_p): Match up with chrec_apply requirements.
11671 (sese_trivially_empty_bb_p): New.
11672 * tree-into-ssa.c (get_reaching_def): Properly support generating
11673 default-defs for incremental rewrite of anonymous names.
11675 2017-10-06 Richard Biener <rguenther@suse.de>
11677 * graphite-sese-to-poly.c (extract_affine): For casts increasing
11678 precision do not perform modulo reduction.
11680 2017-10-06 Richard Biener <rguenther@suse.de>
11682 PR tree-optimization/82436
11683 * tree-vect-slp.c (vect_supported_load_permutation_p): More
11684 conservatively choose the vectorization factor when checking
11685 whether we can perform the required load permutation.
11686 (vect_transform_slp_perm_load): Assert when we may not fail.
11688 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
11690 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
11691 message for incompatible -msdata=* and -mcall-* options.
11693 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11695 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
11696 rate for post-reload scheduling.
11698 2017-10-05 Tamar Christina <tamar.christina@arm.com>
11700 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
11702 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11704 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
11705 to improve monte carlo in scimark.
11707 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
11709 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
11710 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
11711 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
11712 pentium4_cost, nocona_cost): Set reassociation width to 1.
11713 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
11714 width to 2 for fp operations and 1 otherwise.
11715 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
11717 (atom_cost): Set reassociation width to 2.
11718 (slm_cost, generic_cost): Set fp reassociation width
11719 to 2 and 1 otherwise.
11720 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
11721 (core_cost): Set fp reassociation width to 4 and vector to 2.
11722 (ix86_reassociation_width): Rewrite using cost table; special case
11723 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
11724 and TARGET_AVX128_OPTIMAL.
11725 * config/i386/i386.h (processor_costs): Add
11726 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
11727 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
11728 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
11729 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
11730 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
11731 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
11733 2017-10-05 Nathan Sidwell <nathan@acm.org>
11735 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
11737 2017-10-05 Tamar Christina <tamar.christina@arm.com>
11739 * config/arm/arm.c (arm_test_fpu_data): New.
11740 (arm_run_selftests): Call arm_test_fpu_data.
11742 2017-10-04 Nathan Sidwell <nathan@acm.org>
11744 * toplev.c (toplev::main): Remove excess parens on pretty_printer
11746 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
11748 2017-10-04 Sudakshina Das <sudi.das@arm.com>
11750 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
11751 check type for aarch64_simd_valid_immediate.
11752 (aarch64_output_simd_mov_immediate): Update prototype.
11753 (aarch64_simd_valid_immediate): Update prototype.
11754 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
11755 support for ORR-immediate.
11756 (and<mode>3): modified pattern to add support for BIC-immediate.
11757 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
11758 now checks for valid immediate for BIC and ORR based on new enum
11760 (aarch64_output_simd_mov_immediate): Function now used to output
11761 BIC/ORR imm as well based on new enum argument.
11762 * config/aarch64/constraints.md (Do): New vector immediate constraint.
11764 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
11765 (aarch64_reg_or_bic_imm): Likewise.
11767 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11769 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
11770 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
11772 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11776 PR rtl-optimization/82396
11777 * haifa-sched.c (autopref_multipass_init): Simplify
11779 (autopref_rank_data): Simplify sort order.
11780 * sched-int.h (autopref_multipass_data_): Remove
11781 multi_mem_insn_p, min_offset and max_offset.
11783 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11785 * doc/sourcebuild.texi: Document vect_peeling_profitable.
11787 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11789 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
11790 vect_doubleint_cvt.
11792 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11794 * doc/sourcebuild.texi: Document vect_long_mult.
11796 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
11798 PR tree-optimization/82413
11799 * fold-const.c (build_range_check): Use widest_int when comparing
11800 the maximum ETYPE value with HIGH.
11802 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11804 PR rtl-optimization/82396
11805 * haifa-sched.c (autopref_multipass_init): Simplify
11807 (autopref_rank_data): Simplify sort order.
11808 * sched-int.h (autopref_multipass_data_): Remove
11809 multi_mem_insn_p, min_offset and max_offset.
11811 2017-10-04 Jakub Jelinek <jakub@redhat.com>
11813 PR tree-optimization/82381
11814 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
11815 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
11818 PR tree-optimization/82374
11819 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
11820 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
11821 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
11822 current_function_decl to the new decl.
11824 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
11826 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
11827 helper macro for IEEE float128 hardware built-in functions.
11828 (SQRTF128_ODD): Add built-in functions with the round-to-odd
11830 (TRUNCF128_ODD): Likewise.
11831 (ADDF128_ODD): Likewise.
11832 (SUBF128_ODD): Likewise.
11833 (MULF128_ODD): Likewise.
11834 (DIVF128_ODD): Likewise.
11835 (FMAF128_ODD): Likewise.
11836 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
11837 UNSPEC_TRUNC_ROUND_TO_ODD.
11838 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
11839 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
11840 floating point round to odd instructions.
11841 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
11842 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
11843 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
11844 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
11845 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
11846 (trunc<mode>sf2_hw): Change the truncate with round to odd
11847 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
11848 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
11849 to odd hardware instructions.
11850 (sub<mode>3_odd): Likewise.
11851 (mul<mode>3_odd): Likewise.
11852 (div<mode>3_odd): Likewise.
11853 (sqrt<mode>2_odd): Likewise.
11854 (fma<mode>4_odd): Likewise.
11855 (fms<mode>4_odd): Likewise.
11856 (nfma<mode>4_odd): Likewise.
11857 (nfms<mode>4_odd): Likewise.
11858 (trunc<mode>df2_odd): Change the truncate with round to odd
11859 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
11861 * doc/extend.texi (PowerPC built-in functions): Update documentation
11862 for existing IEEE float128-bit built-in functions. Add built-in
11863 functions that generate the IEEE 128-bit floating point round to
11866 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
11868 PR rtl-optimization/77729
11869 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
11870 to (X&(C1&~C2))|C2 transformations.
11872 2017-10-03 Martin Jambor <mjambor@suse.cz>
11874 PR tree-optimization/82363
11875 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
11876 mismatch, mark lacc written regardless of racc.
11878 2017-10-03 Jakub Jelinek <jakub@redhat.com>
11880 PR tree-optimization/82381
11881 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
11882 stmt_to_insert nor wheather SSA_NAMEs are default defs.
11883 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
11884 fallthrough into reassoc_stmt_dominates_stmt_p.
11887 * combine.c (combine_instructions): Don't combine in unreachable
11890 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
11893 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
11894 function to not use the have_cpu variable. Do not set cpu_index,
11895 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
11896 or the default cpu.
11897 (rs6000_valid_attribute_p): Remove duplicate initializations of
11898 old_optimize and func_optimize.
11899 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
11900 (rs6000_activate_target_options): Make global.
11901 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
11904 2017-10-02 Jakub Jelinek <jakub@redhat.com>
11906 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
11907 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
11908 Set *poffset to 0 and *psize and *pmax_size to -1 if
11909 *poffset + *psize overflows in HOST_WIDE_INT.
11911 PR tree-optimization/82387
11912 PR tree-optimization/82388
11913 PR tree-optimization/82389
11914 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
11915 instead of live_bytes non-NULL.
11917 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
11920 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
11923 2017-10-02 Richard Biener <rguenther@suse.de>
11925 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
11926 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
11927 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
11929 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
11931 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
11932 requested precision matches the type's.
11933 * calls.c (alloc_max_size): Calculate the new candidate size as
11934 a widest_int and use wi::to_widest when comparing it with the
11935 current candidate size.
11936 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
11937 zero rather than integer_zero_node.
11938 * match.pd: Check for a no-op conversion before using wi::add
11939 rather than after. Use tree_to_uhwi when summing small shift
11940 counts into an unsigned int.
11942 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
11943 Alan Hayward <alan.hayward@arm.com>
11944 David Sherwood <david.sherwood@arm.com>
11947 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
11948 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
11949 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
11950 POINTER_AND_FP_REGS.
11952 2017-10-02 Richard Biener <rguenther@suse.de>
11954 PR tree-optimization/82355
11955 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11956 a mapping for the enclosing loop but avoid generating one for
11957 the loop tree root.
11958 (copy_bb_and_scalar_dependences): Remove premature codegen
11959 error on PHIs in blocks duplicated into multiple places.
11960 * graphite-scop-detection.c
11961 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11962 in the region use it as loop and nest to analyze the DR in.
11963 (try_generate_gimple_bb): Likewise.
11964 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11965 (add_loop_constraints): For blocks in a loop not in the region
11966 create a dimension with a single iteration.
11967 * sese.h (gbb_loop_at_index): Remove assert.
11969 2017-10-01 Kevin Buettner <kevinb@redhat.com>
11971 * omp-expand.c (adjust_context_scope): New function.
11972 (expand_parallel_call): Call adjust_context_scope.
11974 2017-10-01 Jeff Law <law@redhat.com>
11976 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
11977 dom_opt_dom_walker class with direct access to private members.
11978 Add comments. Call test_for_singularity.
11979 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
11980 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
11981 m_dummy_cond anymore.
11982 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
11984 (pass_dominator:execute): Build the dummy_cond here and pass it
11985 to the dom_opt_dom_walker ctor.
11986 (test_for_singularity): New function.
11988 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
11989 Maya Rashish <coypu@sdf.org>
11991 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
11992 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
11993 (alpha*-*-netbsd*) Use nbsd_tm_file.
11994 (arm*-*-netbsdelf*) Likewise.
11995 (i[34567]86-*-netbsdelf*) Likewise.
11996 (x86_64-*-netbsd*) Likewise.
11997 (mips*-*-netbsd*) Likewise.
11998 (powerpc-*-netbsd*) Likewise.
11999 (sh*-*-netbsd*) Likewise.
12000 (sparc-*-netbsdelf*) Likewise.
12001 (sparc64-*-netbsd*) Likewise.
12002 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
12004 (vax-*-netbsdelf*) Likewise.
12005 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
12006 (UINT_FAST8_TYPE) Likewise.
12007 (INT_FAST16_TYPE) Check CHAR_FAST16.
12008 (UINT_FAST16_TYPE) Likewise.
12010 2017-09-30 Jakub Jelinek <jakub@redhat.com>
12013 * config/i386/i386.md
12014 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
12015 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
12016 *divmodsi4_zext_2): New define_insn_and_split.
12017 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
12018 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
12019 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
12020 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
12021 New define_insn_and_split.
12022 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
12023 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
12024 operands[1] having DImode when mode is SImode.
12026 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
12027 always SImode for DIV and MOD in REG_EQUAL notes.
12029 2017-09-29 Yury Gribov <tetra2005@gmail.com>
12031 PR middle-end/82319
12032 * match.pd: Fix handling of NaNs in pattern.
12034 2017-09-29 Jeff Law <law@redhat.com>
12036 * sbitmap.c (bitmap_bit_in_range_p): New function.
12037 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
12038 * tree-ssa-dse.c (live_bytes_read): New function.
12039 (dse_classify_store): Ignore reads of dead bytes.
12041 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
12042 typos and whitespace errors.
12043 * config/i386/predicates.md (address_no_seg_operand): Likewise.
12044 * config/s390/s390.c (s390_emit_prologue): Likewise.
12046 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12049 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
12050 with a symbol for LRA.
12052 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
12054 PR rtl-optimization/82338
12055 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
12057 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
12059 * genmodes.c (calc_wider_mode): Suppress qsort macro.
12060 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
12061 (qsort_chk): Declare.
12062 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
12063 (qsort_chk): New function.
12065 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12067 PR tree-optimization/82337
12068 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
12069 phi definition if the PHI result appears in an abnormal PHI.
12070 (find_basis_for_base_expr): Don't record a basis if the LHS of the
12071 basis appears in an abnormal PHI.
12073 2017-09-29 Richard Biener <rguenther@suse.de>
12075 * graphite-isl-ast-to-gimple.c
12076 (translate_isl_ast_to_gimple::set_codegen_error): New function.
12077 (binary_op_to_tree): Use it.
12078 (get_rename_from_scev): Likewise.
12079 (copy_loop_phi_nodes): Likewise.
12080 (copy_bb_and_scalar_dependences): Likewise.
12081 (translate_pending_phi_nodes): Likewise.
12083 2017-09-29 Jakub Jelinek <jakub@redhat.com>
12086 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
12087 for movabsq $(i32 << shift), r64.
12089 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
12091 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
12092 index when encoding %esp as %rsp to avoid 0x67 prefix.
12094 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
12096 * config/i386/i386.md (*movsf_internal, *movdf_internal):
12097 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
12099 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
12101 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
12102 Extensions with more than 16 double VFP registers.
12103 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
12104 to_clear_mask and all code related to it. Replace the remaining
12105 entry by a sbitmap and adapt code accordingly.
12107 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
12109 * brig-builtins.def: Change pure attributes to const.
12111 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12113 * config.gcc (default_gnu_indirect_function): Default to yes for
12114 sparc*-*-linux* with glibc.
12116 2017-09-28 Joseph Myers <joseph@codesourcery.com>
12118 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
12119 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
12120 when creating .init_array and .fini_array sections with priority
12123 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
12126 * config/aarch64/aarch64.c
12127 (aarch64_builtin_support_vector_misalignment): Always return false
12128 when misalignment is unknown.
12130 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
12132 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
12133 this function to return false if the definition used by the swap
12134 instruction is artificial, or if the memory address from which the
12135 constant value is loaded is not represented by a base address held
12136 in a register or if the base address register is a frame or stack
12137 pointer. Additionally, return false if the base address of the
12138 loaded constant is a SYMBOL_REF but is not considered to be a
12140 (replace_swapped_load_constant): New function.
12141 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
12142 loaded constant vector with a load of a swapped constant vector.
12144 2017-09-27 Carl Love <cel@us.ibm.com>
12146 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
12147 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
12148 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
12151 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
12153 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
12154 first, always call autopref_rank_data otherwise.
12156 2017-09-27 Richard Biener <rguenther@suse.de>
12158 * graphite-scop-detection.c (find_scop_parameters): Move
12159 loop bound handling ...
12160 (gather_bbs::before_dom_children): ... here, avoiding the need
12161 to build scop_info->loop_nest.
12162 (record_loop_in_sese): Remove.
12163 * sese.h (sese_info_t::loop_nest): Remove.
12164 * sese.c (new_sese_info): Do not allocate loop_nest.
12165 (free_sese_info): Do not free loop_nest.
12167 2017-09-27 Jakub Jelinek <jakub@redhat.com>
12170 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
12171 lhs from calls if the lhs has addressable type.
12173 2017-09-27 Richard Biener <rguenther@suse.de>
12175 * graphite.h (scop::max_alias_set): New member.
12176 * graphite-scop-detection.c: Remove references to non-existing
12177 --param in comments.
12178 (build_alias_sets): Record the maximum alias set used for drs.
12179 (build_scops): Support zero as unlimited for
12180 --param graphite-max-arrays-per-scop.
12181 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
12182 and inline into ...
12183 (build_poly_sr_1): ... here. Compute alias set based on the
12184 maximum alias set used for drs rather than
12185 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
12187 2017-09-27 Richard Biener <rguenther@suse.de>
12189 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
12190 --param loop-block-tile-size=0 to disable tiling.
12192 2017-09-27 Richard Biener <rguenther@suse.de>
12194 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
12195 (graphite-max-nb-scop-params): Document special value zero.
12196 * domwalk.h (dom_walker::STOP): New symbolical constant.
12197 (dom_walker::dom_walker): Add optional parameter for bb to
12199 (dom_walker::~dom_walker): Declare.
12200 (dom_walker::before_dom_children): Document STOP return value.
12201 (dom_walker::m_user_bb_to_rpo): New member.
12202 (dom_walker::m_bb_to_rpo): Likewise.
12203 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
12204 mapping here if not provided by the user.
12205 (dom_walker::~dom_walker): Free bb to RPO mapping if not
12206 provided by the user.
12207 (dom_walker::STOP): Define.
12208 (dom_walker::walk): Do not compute bb to RPO mapping here.
12209 Support STOP return value from before_dom_children to stop
12211 * graphite-optimize-isl.c (optimize_isl): If the schedule
12212 is the same still generate code if -fgraphite-identity
12213 or -floop-parallelize-all are given.
12214 * graphite-scop-detection.c: Include cfganal.h.
12215 (gather_bbs::gather_bbs): Get and pass through bb to RPO
12217 (gather_bbs::before_dom_children): Return STOP for BBs
12219 (build_scops): Compute bb to RPO mapping and pass it to
12220 the domwalk. Treat --param graphite-max-nb-scop-params=0
12221 as not limiting the number of params.
12222 * graphite.c (graphite_initialize): Remove limit on the
12223 number of basic-blocks in a function.
12224 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
12225 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
12226 default value of 10.
12228 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12230 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
12231 Adjust code to eliminate needing to do the shift right 32-bits
12232 operation after XSCVDPSPN.
12234 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12236 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
12237 ((X / Y) != 0 -> X >= Y): Likewise.
12239 2017-09-26 Carl Love <cel@us.ibm.com>
12241 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
12242 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
12243 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
12244 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
12245 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
12246 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
12247 definitions and overloading.
12248 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
12249 statement for P9V_BUILTIN_XST_LEN_R.
12250 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
12251 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
12252 define_expand and define_insn for the instructions and builtins.
12253 * doc/extend.texi: Update the built-in documentation file for the new
12254 built-in functions.
12255 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
12256 define_insn for the instructions
12258 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
12261 * gcc/config/netbsd-protos.h: New file.
12262 * gcc/config/netbsd.c: New file.
12263 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
12264 * gcc/config/t-netbsd: New file.
12265 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
12266 (tmake_file) Add t-netbsd.
12267 (extra_objs) Add netbsd.o.
12269 2017-09-26 Janus Weil <janus@gcc.gnu.org>
12273 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
12275 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
12277 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
12278 sign extension from a vector register to a GPR by doing a 32-bit
12279 direct move and then an EXTSW.
12280 (extendsi<mode>2 splitter): Likewise.
12281 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
12282 right or vector extract after doing XSCVDPSPN. Use
12283 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
12285 (movdi_from_sf_zero_ext): Likewise.
12286 (reload_gpr_from_vsxsf): Likewise.
12287 (p8_mfvsrd_4_disf): Delete, no longer used.
12288 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
12289 then needing to move the SFmode to a GPR to use the XSCVDPSP
12290 instruction instead of FRSP and XSCVDPSPN.
12291 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
12292 it is adjacent to the other XSCVSPDP insns.
12293 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
12294 SFmode to be in traditional Altivec registers.
12295 (vsx_xscvdpspn): Eliminate useless alternative constraint.
12296 (vsx_xscvspdpn): Likewise.
12297 (vsx_xscvspdpn_scalar): Likewise.
12299 2017-09-26 Martin Jambor <mjambor@suse.cz>
12301 * tree-sra.c (compare_access_positions): Put integral types first,
12302 stabilize sorting of integral types, remove conditions putting
12303 non-full-precision integers last.
12304 (sort_and_splice_var_accesses): Disable scalarization if a
12305 non-integert would be represented by a non-full-precision integer.
12307 2017-09-26 Joseph Myers <joseph@codesourcery.com>
12309 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
12310 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
12311 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
12312 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
12313 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
12314 conditionals inside the function instead of around it. Call
12315 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
12316 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
12318 2017-09-26 Richard Biener <rguenther@suse.de>
12320 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
12322 (scop_detection::build_scop_breadth): ... this. Removed.
12323 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
12324 (scop_detection::harmful_stmt_in_bb): Likewise.
12325 (scop_detection::graphite_can_represent_stmt): Likewise.
12326 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
12327 (scop_detection::can_represent_loop): Remove recursion, fold in ...
12328 (scop_detection::can_represent_loop_1): ... this. Removed.
12329 (scop_detection::harmful_loop_in_region): Simplify after inlining
12330 the above and remove more quadraticness.
12331 (build_scops): Adjust.
12332 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
12335 2017-09-26 Jakub Jelinek <jakub@redhat.com>
12338 * config/i386/i386.c (ix86_print_operand_address_as): Only test
12339 REGNO (base) == SP_REG if base is a REG.
12341 PR middle-end/35691
12342 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
12343 if it is different SSA_NAME.
12344 (optimize_range_tests_cmp_bitwise): New function.
12345 (optimize_range_tests): Call it.
12347 2017-09-26 Richard Biener <rguenther@suse.de>
12349 PR tree-optimization/82321
12350 * graphite.c (canonicalize_loop_closed_ssa): Properly check
12351 for the def being inside the loop.
12353 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12355 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
12357 * config/s390/s390-builtins.def: Fix constraint on op4.
12359 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12361 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
12362 independent expanders.
12363 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
12364 ("vec_ordered", "vec_unordered"): New expanders.
12366 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12368 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
12371 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12373 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
12374 vec_unpacks_lo_v16qi.
12375 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
12377 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12379 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
12380 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
12381 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
12383 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12385 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
12387 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
12389 ("*vec_slb<mode>"): New insn pattern.
12390 ("vec_shr_<mode>"): New expander.
12391 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
12392 and force the shift count operand to V16QImode.
12393 ("vec_srb<mode>"): Set shift count mode to V16QI.
12395 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12397 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
12398 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
12399 ("vec_widen_smult_hi_<mode>"): New expander definitions.
12401 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
12404 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
12406 2017-09-26 Richard Biener <rguenther@suse.de>
12408 PR tree-optimization/82320
12409 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
12412 2017-09-25 Jeff Law <law@redhat.com>
12414 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
12415 prototype for new argument.
12416 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
12417 mostly extracted from rs6000_emit_allocate_stack.
12418 (rs6000_emit_probe_stack_range_stack_clash): New function.
12419 (rs6000_emit_allocate_stack): Call
12420 rs6000_emit_probe_stack_range_stack_clash as needed.
12421 (rs6000_emit_probe_stack_range): Add additional argument
12422 to call to gen_probe_stack_range{si,di}.
12423 (output_probe_stack_range): New.
12424 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
12425 (output_probe_stack_range_stack_clash): New.
12426 (rs6000_emit_prologue): Emit notes into dump file as requested.
12427 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
12428 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
12429 Add additional operand and pass it to output_probe_stack_range.
12431 2017-09-25 Bin Cheng <bin.cheng@arm.com>
12433 PR tree-optimization/82163
12434 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
12435 (checking_verify_loop_closed_ssa): New parameter.
12436 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
12437 (check_loop_closed_ssa_stmt): Delete.
12438 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
12439 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
12440 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
12443 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
12445 * brig-builtins.def: Treat HSAIL barrier builtins as
12446 setjmp/longjump style functions.
12448 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12450 * target.def (constant_alignment): New hook.
12451 * defaults.h (CONSTANT_ALIGNMENT): Delete.
12452 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
12453 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
12454 * doc/tm.texi: Regenerate.
12455 * targhooks.h (default_constant_alignment): Declare.
12456 (constant_alignment_word_strings): Likewise.
12457 * targhooks.c (default_constant_alignment): New function.
12458 (constant_alignment_word_strings): Likewise.
12459 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
12460 instead of CONSTANT_ALIGNMENT.
12461 * varasm.c (align_variable, get_variable_align, build_constant_desc)
12462 (force_const_mem): Likewise.
12463 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
12464 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
12465 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
12466 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12467 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
12469 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
12470 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12471 constant_alignment_word_strings.
12472 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
12473 (CONSTANT_ALIGNMENT): Likewise.
12474 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12475 (arm_constant_alignment): New function.
12476 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
12477 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12478 constant_alignment_word_strings.
12479 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
12480 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12481 constant_alignment_word_strings.
12482 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
12483 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12484 (cris_constant_alignment): New function.
12485 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
12486 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12487 (epiphany_constant_alignment): New function.
12488 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
12489 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12490 constant_alignment_word_strings.
12491 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
12492 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12493 constant_alignment_word_strings.
12494 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
12495 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12496 constant_alignment_word_strings.
12497 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
12498 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
12499 * config/i386/i386.c (ix86_constant_alignment): Make static.
12500 Use the same interface as the target hook.
12501 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12502 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
12503 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12504 constant_alignment_word_strings.
12505 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
12506 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
12507 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12508 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
12509 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12510 constant_alignment_word_strings.
12511 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
12512 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12513 constant_alignment_word_strings.
12514 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
12515 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12516 constant_alignment_word_strings.
12517 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
12518 * config/microblaze/microblaze.c (microblaze_constant_alignment):
12520 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12521 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
12522 * config/mips/mips.c (mips_constant_alignment): New function.
12523 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12524 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
12525 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
12526 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12527 (mmix_constant_alignment): Make static. Use the same interface
12528 as the target hook.
12529 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
12530 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12531 constant_alignment_word_strings.
12532 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
12533 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12534 constant_alignment_word_strings.
12535 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
12536 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12537 constant_alignment_word_strings.
12538 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
12539 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12540 (rs6000_constant_alignment): New function.
12541 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
12542 * config/riscv/riscv.c (riscv_constant_alignment): New function.
12543 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12544 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
12545 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12546 (rs6000_constant_alignment): New function.
12547 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
12548 * config/s390/s390.c (s390_constant_alignment): New function.
12549 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12550 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
12551 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12552 constant_alignment_word_strings.
12553 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
12554 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12555 (sparc_constant_alignment): New function.
12556 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
12557 * config/spu/spu.c (spu_constant_alignment): New function.
12558 (TARGET_CONSTANT_ALIGNMENT): Redefine.
12559 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
12560 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12561 constant_alignment_word_strings.
12562 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
12563 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12564 constant_alignment_word_strings.
12565 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
12566 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
12567 constant_alignment_word_strings.
12568 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
12569 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12570 (visium_constant_alignment): New function.
12571 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
12572 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
12573 (xtensa_constant_alignment): New function.
12574 * system.h (CONSTANT_ALIGNMENT): Poison.
12576 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
12578 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12579 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
12580 (rs6000_builtin_valid_without_lhs): New helper function.
12581 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12582 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
12584 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
12586 * target.h (vec_perm_indices): Use unsigned short rather than
12588 (auto_vec_perm_indices): Likewise.
12589 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
12590 Use unsigned int rather than unsigned char.
12591 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
12593 2017-09-25 Richard Biener <rguenther@suse.de>
12595 * cfgloop.h (sort_sibling_loops): Declare.
12596 * cfgloop.c (sort_sibling_loops_cmp): New helper.
12597 (sort_sibling_loops): New function sorting the sibling loop list
12599 * graphite.c (graphite_transform_loops): Sort sibling loops.
12601 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
12603 * target.def (vec_perm_const_ok): Change sel parameter to
12605 * optabs-query.c (can_vec_perm_p): Update accordingly.
12606 * doc/tm.texi: Regenerate.
12607 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
12608 auto_vec_perm_indices and remove separate nelt field.
12609 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
12610 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
12611 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
12612 (aarch64_expand_vec_perm_const): Update accordingly.
12613 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
12614 to vec_perm_indices.
12615 * config/arm/arm.c (expand_vec_perm_d): Change perm to
12616 auto_vec_perm_indices and remove separate nelt field.
12617 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
12618 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
12619 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
12621 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
12622 to vec_perm_indices.
12623 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
12624 sel to vec_perm_indices.
12625 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
12626 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
12627 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
12629 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
12632 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
12635 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
12636 on the FUNCTION_DECL function context if it has a DIE that is a
12639 2017-09-25 Richard Biener <rguenther@suse.de>
12641 PR tree-optimization/82285
12642 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
12645 2017-09-25 Tom de Vries <tom@codesourcery.com>
12649 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
12652 2017-09-25 Richard Biener <rguenther@suse.de>
12654 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
12655 ISL errors other than isl_error_quota happen. Dump if the
12656 schedule is the same.
12657 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
12658 errors instead of aborting inside ISL.
12660 2017-09-25 Iain Sandoe <iain@codesourcery.com>
12663 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
12664 of libgcc_eh for m64.
12665 * config/i386/darwin64.h: Likewise.
12667 2017-09-25 Richard Biener <rguenther@suse.de>
12669 PR middle-end/82144
12670 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
12671 attribute for incomplete types nor twice for complete ones.
12673 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
12676 * config/i386/i386.c (ix86_print_operand_address_as): Encode
12677 %esp as %rsp to avoid 0x67 prefix if there is no index or base
12680 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
12683 * config/i386/i386.opt (mprefer-avx256): Use
12684 ix86_target_flags variable.
12685 * config/i386/i386.c (ix86_target_string): Move
12686 -mprefer-avx256 to flag2_opts.
12688 2017-09-22 Jakub Jelinek <jakub@redhat.com>
12690 PR middle-end/35691
12691 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
12692 and x != -1 | y != -1 into (x & y) != -1.
12694 2017-09-22 Steve Ellcey <sellcey@cavium.com>
12696 * config.gcc: Add new case statement to set
12697 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
12698 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
12699 s390x-*-linux* case statements. Added aarch64 to the list of
12700 supported architectures.
12702 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12704 PR tree-optimization/82289
12705 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
12706 STMT_VINFO_RELEVANT_P.
12708 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12709 Alan Hayward <alan.hayward@arm.com>
12710 David Sherwood <david.sherwood@arm.com>
12712 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
12713 for VR_RANGE only; don't allow VR_ANTI_RANGE.
12714 (extract_range_from_binary_expr_1): Don't call
12715 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
12717 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12718 Alan Hayward <alan.hayward@arm.com>
12719 David Sherwood <david.sherwood@arm.com>
12721 * target.def (preferred_vector_alignment): New hook.
12722 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
12724 * doc/tm.texi: Regenerate.
12725 * targhooks.h (default_preferred_vector_alignment): Declare.
12726 * targhooks.c (default_preferred_vector_alignment): New function.
12727 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
12729 (DR_TARGET_ALIGNMENT): New macro.
12730 (aligned_access_p): Update commentary.
12731 (vect_known_alignment_in_bytes): New function.
12732 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
12734 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
12735 Calculate the misalignment based on the target alignment rather than
12737 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
12738 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
12739 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
12740 the target alignment, rather than masking the element misalignment
12741 with the number of elements in a vector. Also use the target
12742 alignment when calculating the maximum number of peels.
12743 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
12744 instead of TYPE_ALIGN_UNIT.
12745 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
12746 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
12747 (vect_create_addr_base_for_vector_ref): Update call accordingly.
12748 (vect_create_data_ref_ptr): Likewise.
12749 (vect_setup_realignment): Realign by ANDing with
12750 -DR_TARGET_MISALIGNMENT.
12751 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
12752 the number of peels based on DR_TARGET_ALIGNMENT.
12753 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
12754 with the guaranteed alignment boundary when deciding whether
12756 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
12757 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
12758 (ensure_base_align): Remove stmt_info parameter. Get the
12759 target base alignment from DR_TARGET_ALIGNMENT.
12760 (vectorizable_store): Update call accordingly. Interpret
12761 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
12763 (vectorizable_load): Likewise.
12765 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12766 Alan Hayward <alan.hayward@arm.com>
12767 David Sherwood <david.sherwood@arm.com>
12769 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
12770 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
12771 (vect_enhance_data_refs_alignment): Likewise.
12773 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12775 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
12776 error. Only quit immediately if parsing is complete.
12777 (BEGIN): Initialize fatal_err and parse_done.
12778 (begin fpu, end fpu): Check number of arguments.
12779 (begin arch, end arch): Likewise.
12780 (begin cpu, end cpu): Likewise.
12781 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12782 (optalias): Likewise.
12784 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12786 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12787 * config/arm/arm-isa.h: Delete. Move definitions to ...
12788 * arm-cpus.in: ... here. Use new feature and fgroup values.
12789 * config/arm/arm.c (arm_option_override): Use lower case for feature
12791 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12792 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12793 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12795 (print_isa_bits_for): New function.
12796 (gen_isa): New function.
12797 (gen_comm_data): Use print_isa_bits_for.
12798 (define feature): New keyword.
12799 (define fgroup): New keyword.
12800 * config/arm/t-arm (TM_H): Remove.
12801 (GTM_H): Add arm-isa.h.
12802 (arm-isa.h): Add rule to generate file.
12803 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12804 case for feature bit names.
12806 2017-09-22 Richard Biener <rguenther@suse.de>
12808 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
12810 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
12811 print of no dependency loops ...
12812 * graphite.c (graphite_transform_loops): ... here.
12813 (canonicalize_loop_closed_ssa_form): Work from inner to outer
12815 (same_close_phi_node, remove_duplicate_close_phi,
12816 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
12817 (canonicalize_loop_closed_ssa): ... here and simplify.
12818 * graphite-optimize-isl.c: Include tree-vectorizer.h.
12819 (optimize_isl): Use dump_printf_loc to tell when we stopped
12820 optimizing because of an ISL timeout.
12822 2017-09-22 Richard Biener <rguenther@suse.de>
12824 PR tree-optimization/82291
12825 * tree-if-conv.c (predicate_mem_writes): Make sure to
12826 remove writes in blocks predicated with false.
12828 2017-09-22 Richard Biener <rguenther@suse.de>
12830 * sese.c: Include cfganal.h.
12831 (if_region_set_false_region): Remove.
12832 (create_if_region_on_edge): Likewise.
12833 (move_sese_in_condition): Re-implement without destroying
12836 2017-09-22 Richard Biener <rguenther@suse.de>
12838 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
12839 Verify both BBs contain loop PHI nodes before dispatching to
12840 copy_loop_phi_args.
12841 (graphite_regenerate_ast_isl): Do not recompute dominators,
12842 do not verify three times. Restructure for clarity.
12843 * graphite-scop-detection.c (same_close_phi_node,
12844 remove_duplicate_close_phi, make_close_phi_nodes_unique,
12845 defined_in_loop_p, canonicalize_loop_closed_ssa,
12846 canonicalize_loop_closed_ssa_form): Simplify, remove excess
12847 checking and SSA rewrite, move to ...
12848 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
12849 (graphite_initialize): Do not pass in ctx, do not reset the
12850 SCEV cache, compute only dominators.
12851 (graphite_transform_loops): Allocate ISL ctx after
12852 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
12853 Maintain post-dominators only around build_scops.
12854 * sese.c (if_region_set_false_region): Make static. Free
12855 and recompute dominators.
12856 (move_sese_in_condition): Assert we don't get called with
12857 post-dominators computed.
12858 * sese.h (if_region_set_false_region): Remove.
12860 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
12862 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
12863 mode attribute for TARGET_AVX512VL.
12865 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
12867 * config/i386/i386.opt (mprefer-avx256): New option.
12868 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
12870 (ix86_preferred_simd_mode): Return 256-bit AVX modes
12871 for TARGET_PREFER_AVX256.
12872 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
12874 2017-09-21 Jeff Law <law@redhat.com>
12876 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
12877 Fix dump output if the only stack space is for pushed registers.
12879 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12881 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
12884 2017-09-21 Martin Sebor <msebor@redhat.com>
12887 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
12888 code (in C++) or code that triggers warnings.
12890 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
12892 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
12894 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
12896 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
12897 * sched-rgn.c: Ditto.
12898 * sel-sched-ir.c: Ditto.
12900 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
12902 * toplev.h (set_random_seed): Adjust return type.
12903 * toplev.c (init_local_tick): Move eager initialization of random_seed
12904 to get_random_seed. Adjust comment.
12905 (init_random_seed): Inline to get_random_seed, delete.
12906 (get_random_seed): Initialize random_seed lazily.
12907 (set_random_seed): Do not return previous value.
12908 (print_switch_value): Do not call get_random_seed.
12910 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
12912 * cgraph.c (delete_function_version): New, broken out from...
12913 (cgraph_node::delete_function_version): ...here. Rename to
12914 cgraph_node::delete_function_version_by_decl. Update all uses.
12915 (cgraph_node::remove): Call delete_function_version.
12917 2017-09-21 Jakub Jelinek <jakub@redhat.com>
12920 * tree-inline.c (expand_call_inline): Emit clobber stmts for
12921 VAR_DECLs to which addressable non-volatile parameters are mapped
12922 and for id->retvar after the return value assignment. Clear
12923 id->retval and id->retbnd after inlining.
12925 2017-09-21 Richard Biener <rguenther@suse.de>
12927 PR tree-optimization/82276
12928 PR tree-optimization/82244
12929 * tree-vrp.c (build_assert_expr_for): Set
12930 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
12932 (remove_range_assertions): Revert earlier change.
12934 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
12937 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
12939 2017-09-21 Richard Biener <rguenther@suse.de>
12941 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12942 Restore valid IL after code generation errors.
12943 * graphite.c (graphite_transform_loops): Diagnose code
12944 generation issues as MSG_MISSED_OPTIMIZATION and continue
12945 with processing SCOPs.
12947 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12948 Alan Hayward <alan.hayward@arm.com>
12949 David Sherwood <david.sherwood@arm.com>
12951 * calls.c (compute_argument_addresses): Use simplify_gen_binary
12952 rather than choosing between plus_constant and gen_rtx_<CODE>.
12953 * expr.c (emit_push_insn): Likewise.
12954 (expand_expr_real_2): Likewise.
12956 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12957 Alan Hayward <alan.hayward@arm.com>
12958 David Sherwood <david.sherwood@arm.com>
12960 * loop-unroll.c (split_iv): Call copy_rtx on the step.
12962 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12963 Alan Hayward <alan.hayward@arm.com>
12964 David Sherwood <david.sherwood@arm.com>
12966 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
12967 calling tree_to_uhwi.
12969 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12970 Alan Hayward <alan.hayward@arm.com>
12971 David Sherwood <david.sherwood@arm.com>
12973 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
12974 INTEGER_CST rather than a negative test for ADDR_EXPR.
12976 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12977 Alan Hayward <alan.hayward@arm.com>
12978 David Sherwood <david.sherwood@arm.com>
12980 * tree-vrp.c (extract_range_from_binary_expr_1): Check
12981 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
12983 2017-09-21 Richard Biener <rguenther@suse.de>
12985 PR tree-optimization/71351
12986 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
12987 graphite_create_new_loop_guard): Remove, fold remaining parts
12989 (translate_isl_ast_node_for): ... here and simplify.
12991 2017-09-21 Jakub Jelinek <jakub@redhat.com>
12994 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
12995 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
12996 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
12997 alternative always use QI mode, for -Os imov (=R,R) alternative
12998 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
13001 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13002 Jeff Law <law@redhat.com>
13004 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
13005 (allocate_stack_space): New function, partially extracted from
13006 s390_emit_prologue.
13007 (s390_emit_prologue): Track offset to most recent stack probe.
13008 Code to allocate space moved into allocate_stack_space.
13009 Dump actions when no stack is allocated.
13010 (s390_prologue_plus_offset): New function.
13011 (s390_emit_stack_probe): Likewise.
13013 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
13015 * common.opt (Wa, Wl, Wp, g, gz=): Add
13017 (gno-column-info): Remove.
13018 (gcolumn-info): Drop RejectNegative.
13019 (gno-): New prefix.
13020 (gno-record-gcc-switches): Remove.
13021 (grecord-gcc-switches): Drop RejectNegative.
13022 (gno-split-dwarf): Remove.
13023 (gsplit-dwarf): Drop RejectNegative.
13024 (gno-strict-dwarf): Remove.
13025 (gstrict-dwarf): Drop RejectNegative.
13026 * config/darwin.opt (gfull, gused): Add RejectNegative.
13027 * dwarf2out.c (gen_producer_string): Drop
13028 gno-record-gcc-switches handler.
13029 * optc-gen.awk: Add g to prefixes with negative forms.
13030 * opts-common.c (remapping_prefix_p): New.
13031 (find_opt): Check it.
13032 (generate_canonical_option): Test g prefix.
13033 (option_map): Add -gno- mapping.
13034 (add_misspelling_candidates): Check remapping_prefix_p.
13036 2017-09-20 Jeff Law <law@redhat.com>
13038 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
13039 thinko in stack clash protection support.
13041 * explow.c (compute_stack_clash_protection_loop_data): Use
13042 CONST_INT_P instead of explicit test. Verify object is a
13043 CONST_INT_P before looking at INTVAL.
13044 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
13045 instead of explicit test.
13047 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
13050 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
13051 address instead of to r1 and r11.
13053 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
13055 * config.gcc: Support "knm".
13056 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
13057 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13059 * config/i386/i386.c (m_KNM): Define.
13060 (processor_target_table): Add "knm".
13062 (ix86_option_override_internal): Add "knm".
13063 (ix86_issue_rate): Add PROCESSOR_KNM.
13064 (ix86_adjust_cost): Ditto.
13065 (ia32_multipass_dfa_lookahead): Ditto.
13066 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
13067 (fold_builtin_cpu): Add M_INTEL_KNM.
13068 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
13069 (processor_type): Add PROCESSOR_KNM.
13070 * config/i386/x86-tune.def: Add m_KNM.
13071 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
13073 2017-09-20 Richard Biener <rguenther@suse.de>
13075 PR tree-optimization/80213
13076 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
13077 are allowed in empty BBs as well.
13078 (canonicalize_loop_closed_ssa): Also look for other complex
13080 (scop_detection::get_sese): Include the loop-closed PHI block
13082 (scop_detection::merge_sese): Remove code adding extra blocks.
13083 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
13084 (build_scops): Assert the final returned scop is invalid.
13086 2017-09-20 Richard Biener <rguenther@suse.de>
13088 PR tree-optimization/82264
13089 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
13091 (vn_phi_lookup): Likewise.
13092 (vn_phi_insert): Likewise.
13094 2017-09-20 Jakub Jelinek <jakub@redhat.com>
13096 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
13097 that fits into uhwi or shwi, add DW_AT_const_value regardless
13098 of early_dwarf without going through RTL, using add_AT_unsigned
13101 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
13102 (DEBUG_LTO_ABBREV_SECTION): Likewise.
13103 (DEBUG_LTO_MACINFO_SECTION): Likewise.
13104 (DEBUG_MACRO_SECTION): Likewise.
13105 (DEBUG_LTO_MACRO_SECTION): Likewise.
13106 (DEBUG_STR_DWO_SECTION): Likewise.
13107 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
13108 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
13109 (DEBUG_LTO_DWO_LINE_SECTION): Define.
13110 (DEBUG_LTO_LINE_STR_SECTION): Define.
13111 (init_sections_and_labels): Initialize debug_line_str_section
13112 variable. Initialize debug_loc_section for -gdwarf-5 to
13113 DEBUG_LOCLISTS_SECTION. Formatting fixes.
13115 2017-09-20 Richard Biener <rguenther@suse.de>
13117 * graphite-sese-to-poly.c (extract_affine): Properly handle
13118 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
13120 2017-09-20 Richard Biener <rguenther@suse.de>
13122 PR tree-optimization/81373
13123 * graphite-scop-detection.c (build_cross_bb_scalars_def):
13124 Force SESE live-out defs to be handled even if they are
13127 2017-09-19 Jeff Law <law@redhat.com>
13129 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
13130 nothing for stack adjustments with REG_STACK_CHECK.
13131 * sched-deps.c (parse_add_or_inc): Reject insns with
13132 REG_STACK_CHECK from dependency breaking.
13133 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
13134 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
13135 * reg-notes.def (STACK_CHECK): New note.
13137 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
13138 (ix86_expand_prologue): Dump stack clash info as needed.
13139 Call ix86_adjust_stack_and_probe_stack_clash as needed.
13141 * function.c (dump_stack_clash_frame_info): New function.
13142 * function.h (dump_stack_clash_frame_info): Prototype.
13143 (enum stack_clash_probes): New enum.
13145 * config/alpha/alpha.c (alpha_expand_prologue): Also check
13146 flag_stack_clash_protection.
13147 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
13148 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
13149 (arm_frame_pointer_required): Likewise.
13150 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
13151 (ia64_expand_prologue): Likewise.
13152 * config/mips/mips.c (mips_expand_prologue): Likewise.
13153 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
13154 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13155 (sparc_flat_expand_prologue): Likewise.
13156 * config/spu/spu.c (spu_expand_prologue): Likewise.
13158 * explow.c: Include "params.h".
13159 (anti_adjust_stack_and_probe_stack_clash): New function.
13160 (get_stack_check_protect): Likewise.
13161 (compute_stack_clash_protection_loop_data): Likewise.
13162 (emit_stack_clash_protection_loop_start): Likewise.
13163 (emit_stack_clash_protection_loop_end): Likewise.
13164 (allocate_dynamic_stack_space): Use get_stack_check_protect.
13165 Use anti_adjust_stack_and_probe_stack_clash.
13166 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
13167 (emit_stack_clash_protection_loop_start): Likewise.
13168 (emit_stack_clash_protection_loop_end): Likewise.
13169 * rtl.h (get_stack_check_protect): Prototype.
13170 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
13171 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
13172 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
13174 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
13176 * doc/tm.texi: Rebuilt.
13177 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
13178 get_stack_check_protect.
13179 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
13180 * config/arm/arm.c (arm_expand_prologue): Likewise.
13181 (arm_frame_pointer_required): Likewise.
13182 * config/i386/i386.c (ix86_expand_prologue): Likewise.
13183 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
13184 * config/mips/mips.c (mips_expand_prologue): Likewise.
13185 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
13186 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
13187 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
13188 (sparc_flat_expand_prologue): Likewise.
13190 * common.opt (-fstack-clash-protection): New option.
13191 * flag-types.h (enum stack_check_type): Note difference between
13192 -fstack-check= and -fstack-clash-protection.
13193 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
13194 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
13195 * toplev.c (process_options): Issue warnings/errors for cases
13196 not handled with -fstack-clash-protection.
13197 * doc/invoke.texi (-fstack-clash-protection): Document new option.
13198 (-fstack-check): Note additional problem with -fstack-check=generic.
13199 Note that -fstack-check is primarily for Ada and refer users
13200 to -fstack-clash-protection for stack-clash-protection.
13201 Document new params for stack clash protection.
13203 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
13205 * config/i386/i386.c (ix86_split_long_move): Do not handle
13206 address used for LEA in a special way.
13208 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
13210 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
13212 2017-09-19 Martin Sebor <msebor@redhat.com>
13215 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
13216 of incompatible types.
13218 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
13220 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
13221 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
13222 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13223 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
13225 2017-09-19 Richard Biener <rguenther@suse.de>
13227 PR tree-optimization/82244
13228 * tree-vrp.c (remove_range_assertions): Do not propagate
13229 a constant to abnormals but replace the assert with a copy.
13231 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
13233 PR rtl-optimization/57878
13234 PR rtl-optimization/68988
13235 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
13236 avoidance test involving non_reload_pseudos. Move frequency test
13237 below the general fragmentation avoidance test.
13239 2017-09-19 Richard Biener <rguenther@suse.de>
13241 PR tree-optimization/69728
13242 * graphite-sese-to-poly.c (schedule_error): New global.
13243 (add_loop_schedule): Handle empty domain by failing the
13245 (build_original_schedule): Handle schedule_error.
13247 2017-09-19 Richard Biener <rguenther@suse.de>
13249 * graphite-scop-detection.c (scop_detection::can_represent_loop):
13250 Do not iterate to sibling loops but only to siblings of inner
13253 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
13256 * config/m68k/m68k.md (moveq feeding equality comparison): Check
13257 that the registers are different.
13259 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
13261 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
13262 to processor_model and "amdfam17h" to arch_names_table.
13263 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
13265 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13268 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
13270 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13271 Alan Hayward <alan.hayward@arm.com>
13272 David Sherwood <david.sherwood@arm.com>
13274 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
13276 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
13278 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
13279 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
13280 vect_schedule_slp_instance.
13281 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
13282 Update call to vect_slp_analyze_node_operations. Simplify return
13284 (vect_slp_analyze_bb_1): Update call accordingly.
13285 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
13286 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
13287 (vect_schedule_slp): Update call accordingly.
13289 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13290 Alan Hayward <alan.hayward@arm.com>
13291 David Sherwood <david.sherwood@arm.com>
13293 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
13294 with types that aren't in fact scalar.
13296 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13298 * tree-vect-slp.c (vect_record_max_nunits): New function,
13300 (vect_build_slp_tree_1): ...here.
13301 (vect_build_slp_tree_2): Call it for phis too.
13303 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13305 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
13306 to vect_get_vec_def_for_operand when getting the mask operand.
13308 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13309 Alan Hayward <alan.hayward@arm.com>
13310 David Sherwood <david.sherwood@arm.com>
13312 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
13315 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13316 Alan Hayward <alan.hayward@arm.com>
13317 David Sherwood <david.sherwood@arm.com>
13319 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
13320 calculation for vector booleans.
13322 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
13323 Alan Hayward <alan.hayward@arm.com>
13324 David Sherwood <david.sherwood@arm.com>
13326 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
13328 (vect_transform_stmt): ...here.
13329 (vect_analyze_stmt): Use it instead of calling
13330 vectorizable_live_operation directly.
13332 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
13334 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
13335 non-SIMT targets in acc vector loops.
13337 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
13339 * configure.ac: Add arc and check if assembler supports gdwarf2.
13340 * configure: Regenerate.
13342 2017-09-18 Richard Biener <rguenther@suse.de>
13344 PR tree-optimization/82220
13345 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
13346 epilogue niters from the min_profitable_iters compute.
13348 2017-09-18 Jakub Jelinek <jakub@redhat.com>
13351 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
13352 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
13353 (ix86_init_pic_reg): Revert 2017-09-01 changes.
13355 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
13358 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
13359 switching to a new text section.
13361 2017-09-18 Richard Biener <rguenther@suse.de>
13363 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
13365 (build_alias_set): Reject aliases with no access function.
13367 2017-09-18 Richard Biener <rguenther@suse.de>
13369 PR tree-optimization/79622
13370 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
13372 (build_cross_bb_scalars_use): Likewise.
13374 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
13376 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
13378 2017-09-18 Alan Modra <amodra@gmail.com>
13381 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
13382 stack_pointer_rtx for count 0. Update comments. Break up
13383 large rtl expression.
13385 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
13387 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
13388 Increase to 20 bytes.
13389 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
13390 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
13391 or avx version of the stub.
13393 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
13396 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
13397 compute the minimum stack alignment. Also update preferred stack
13398 boundary for leaf functions.
13400 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
13402 PR tree-optimization/82228
13403 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
13406 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13408 * common/config/nds32/nds32-common.c
13409 (nds32_option_optimization_table): Refine formatting.
13410 (nds32_option_optimization_table): Use -fsched-pressure and
13411 -fomit-frame-pointer for specific optimization level.
13413 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
13415 * config/nds32/nds32.c: Refine formatting and comments.
13416 * config/nds32/nds32.h: Likewise.
13417 * config/nds32/nds32.md: Likewise.
13418 * config/nds32/nds32-cost.c: Likewise.
13419 * config/nds32/nds32-isr.c: Likewise.
13420 * config/nds32/nds32-md-auxiliary.c: Likewise.
13421 * config/nds32/nds32-multiple.md: Likewise.
13422 * config/nds32/nds32-predicates.c: Likewise.
13424 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
13425 Jakub Jelinek <jakub@redhat.com>
13427 Add support for -std=c++2a.
13428 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
13430 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
13432 2017-09-15 Steve Ellcey <sellcey@cavium.com>
13435 * doc/extend.texi (Common Function Attributes): Add
13436 references to ARM, AArch64, and S/390 specific attributes.
13437 (Function Specific Option Pragmas): Add AArch64 and S/390
13438 to list of back ends that support the target pragma.
13440 2017-09-15 Nathan Sidwell <nathan@acm.org>
13442 * doc/standards.texi: Fix C++17 description. Update URLs for
13445 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
13447 * common.opt (Wcast-align=strict): New warning option.
13448 * doc/invoke.texi: Document -Wcast-align=strict.
13450 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
13452 * cgraph.h (cgraph_thunk_info): Add comments.
13453 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
13454 assert for VIRTUAL_* arguments stricter.
13456 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
13458 PR tree-optimization/71026
13459 * match.pd: Move RDIV patterns from fold-const.c
13460 * fold-const.c (distribute_real_division): Removed.
13461 (fold_binary_loc): Remove calls to distribute_real_divison.
13463 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13465 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
13466 c++1z and gnu++1z as deprecated. Change other references to
13467 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
13468 Change -Wc++1z-compat to -Wc++17-compat.
13469 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
13470 * dwarf2out.c (highest_c_language): Handle C++17.
13471 (gen_compile_unit_die): Likewise.
13473 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13475 PR rtl-optimization/82192
13476 * combine.c (make_extraction): Don't look through non-paradoxical
13477 SUBREGs or TRUNCATE if pos + len is or might be bigger than
13480 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13481 Alan Hayward <alan.hayward@arm.com>
13482 David Sherwood <david.sherwood@arm.com>
13484 * target.def (function_arg_offset): New hook.
13485 * targhooks.h (default_function_arg_offset): Declare.
13486 * targhooks.c (default_function_arg_offset): New function.
13487 * function.c (locate_and_pad_parm): Use
13488 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
13489 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
13490 (TARGET_FUNCTION_ARG_OFFSET): ...this.
13491 * doc/tm.texi: Regenerate.
13492 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
13493 * config/spu/spu.c (spu_function_arg_offset): New function.
13494 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
13495 * system.h (FUNCTION_ARG_OFFSET): Poison.
13497 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13498 Alan Hayard <alan.hayward@arm.com>
13499 David Sherwood <david.sherwood@arm.com>
13501 * target.def (truly_noop_truncation): New hook.
13502 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
13503 than TRULY_NOOP_TRUNCATION.
13504 * hooks.h (hook_bool_uint_uint_true): Declare.
13505 * hooks.c (hook_bool_uint_uint_true): New function.
13506 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
13507 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
13508 * doc/tm.texi: Regenerate.
13509 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
13510 rather than TRULY_NOOP_TRUNCATION in comments.
13511 (simplify_comparison): Likewise.
13512 (record_truncated_value): Likewise.
13513 * expmed.c (extract_bit_field_1): Likewise.
13514 (extract_split_bit_field): Likewise.
13515 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
13516 instead of TRULY_NOOP_TRUNCATION.
13517 * function.c (assign_parm_setup_block): Likewise.
13518 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
13519 * rtlhooks.c: Include target.h.
13520 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
13521 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
13522 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
13523 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
13524 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
13525 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
13526 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
13527 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
13528 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
13529 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
13530 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
13531 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
13532 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
13533 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
13534 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
13535 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
13536 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
13537 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
13538 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
13539 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
13540 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
13541 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
13542 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
13543 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
13544 * config/mips/mips.c (mips_truly_noop_truncation): New function.
13545 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13546 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13547 rather than TRULY_NOOP_TRUNCATION in comments.
13548 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
13549 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
13550 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
13551 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
13552 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
13553 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
13554 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
13555 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
13556 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
13557 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
13558 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
13559 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13560 rather than TRULY_NOOP_TRUNCATION in comments.
13561 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
13562 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
13563 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
13564 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
13565 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
13566 TRULY_NOOP_TRUNCATION condition.
13567 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
13568 (TRULY_NOOP_TRUNCATION): Delete.
13569 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
13570 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
13571 * config/spu/spu.c (spu_truly_noop_truncation): New function.
13572 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13573 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
13574 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
13575 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
13576 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
13577 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
13578 rather than TRULY_NOOP_TRUNCATION in comments.
13579 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
13580 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
13581 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
13582 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
13583 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
13584 * system.h (TRULY_NOOP_TRUNCATION): Poison.
13586 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
13589 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
13590 (*cmp_ior): Likewise.
13591 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
13592 (*ior_scc_scc_cmp): Likewise.
13593 (*and_scc_scc): Likewise.
13594 (*and_scc_scc_cmp): Likewise.
13596 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
13597 Alan Hayard <alan.hayward@arm.com>
13598 David Sherwood <david.sherwood@arm.com>
13600 * target.def (can_change_mode_class): New hook.
13601 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
13602 (hard_regno_nregs): Likewise.
13603 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
13604 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
13605 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
13606 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
13607 (LOAD_EXTEND_OP): Update accordingly.
13608 * doc/tm.texi: Regenerate.
13609 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
13610 CANNOT_CHANGE_MODE_CLASS.
13611 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
13612 (REG_CAN_CHANGE_MODE_P): ...this new macro.
13613 * combine.c (simplify_set): Update accordingly.
13614 * emit-rtl.c (validate_subreg): Likewise.
13615 * recog.c (general_operand): Likewise.
13616 * regcprop.c (mode_change_ok): Likewise.
13617 * reload1.c (choose_reload_regs): Likewise.
13618 (inherit_piecemeal_p): Likewise.
13619 * rtlanal.c (simplify_subreg_regno): Likewise.
13620 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
13621 instead of CANNOT_CHANGE_MODE_CLASS.
13622 (reload_cse_simplify_operands): Likewise.
13623 * reload.c (push_reload): Use targetm.can_change_mode_class
13624 instead of CANNOT_CHANGE_MODE_CLASS.
13625 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
13626 REG_CANNOT_CHANGE_MODE_P.
13627 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13628 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
13629 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13630 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13631 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13632 (arm_can_change_mode_class): New function.
13633 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
13634 than CANNOT_CHANGE_MODE_CLASS in comments.
13635 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13636 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
13637 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
13638 (ix86_can_change_mode_class): ...this new function, inverting the
13639 sense of the return value.
13640 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13641 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13642 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13643 (ia64_can_change_mode_class): New function.
13644 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13645 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
13646 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
13647 (m32c_can_change_mode_class): ...this new function, inverting the
13648 sense of the return value.
13649 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13650 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13651 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
13652 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
13653 (mips_can_change_mode_class): ...this new function, inverting the
13654 sense of the return value.
13655 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13656 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13657 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13658 (msp430_can_change_mode_class): New function.
13659 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13660 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
13661 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13662 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13663 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13664 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
13665 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13666 (pa_cannot_change_mode_class): Replace with...
13667 (pa_can_change_mode_class): ...this new function, inverting the
13668 sense of the return value.
13669 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
13670 than CANNOT_CHANGE_MODE_CLASS in comments.
13671 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13672 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
13673 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13674 (pdp11_cannot_change_mode_class): Replace with...
13675 (pdp11_can_change_mode_class): ...this new function, inverting the
13676 sense of the return value.
13677 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13678 * config/powerpcspe/powerpcspe-protos.h
13679 (rs6000_cannot_change_mode_class_ptr): Delete.
13680 * config/powerpcspe/powerpcspe.c
13681 (rs6000_cannot_change_mode_class_ptr): Delete.
13682 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13683 (rs6000_option_override_internal): Assign to
13684 targetm.can_change_mode_class instead of
13685 rs6000_cannot_change_mode_class_ptr.
13686 (rs6000_cannot_change_mode_class): Replace with...
13687 (rs6000_can_change_mode_class): ...this new function, inverting the
13688 sense of the return value.
13689 (rs6000_debug_cannot_change_mode_class): Replace with...
13690 (rs6000_debug_can_change_mode_class): ...this new function.
13691 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13692 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
13693 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13694 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13695 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
13697 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
13698 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13699 (rs6000_option_override_internal): Assign to
13700 targetm.can_change_mode_class instead of
13701 rs6000_cannot_change_mode_class_ptr.
13702 (rs6000_cannot_change_mode_class): Replace with...
13703 (rs6000_can_change_mode_class): ...this new function, inverting the
13704 sense of the return value.
13705 (rs6000_debug_cannot_change_mode_class): Replace with...
13706 (rs6000_debug_can_change_mode_class): ...this new function.
13707 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13708 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
13709 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
13710 (s390_can_change_mode_class): ...this new function, inverting the
13711 sense of the return value.
13712 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13713 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13714 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
13715 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13716 (sh_cannot_change_mode_class): Replace with...
13717 (sh_can_change_mode_class): ...this new function, inverting the
13718 sense of the return value.
13719 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13720 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13721 (sparc_can_change_mode_class): New function.
13722 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13723 * config/spu/spu.c (spu_can_change_mode_class): New function.
13724 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13725 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
13726 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
13727 (visium_can_change_mode_class): New function.
13728 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
13730 2017-09-15 Richard Biener <rguenther@suse.de>
13732 PR tree-optimization/82217
13733 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
13734 but not undefined case.
13736 2017-09-15 Jakub Jelinek <jakub@redhat.com>
13739 * postreload.c (reload_cse_simplify_operands): Skip
13740 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
13742 2017-09-15 Richard Biener <rguenther@suse.de>
13744 PR tree-optimization/68823
13745 * graphite-scop-detection.c (build_alias_set): If we have a
13746 possible dependence check whether we can handle them by just
13747 looking at the DRs DR_ACCESS_FNs.
13748 (build_scops): If build_alias_set fails, fail the SCOP.
13750 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
13752 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
13753 to support float128 built-in functions that require the ISA 3.0
13755 (BU_FLOAT128_3_HW): Likewise.
13756 (SQRTF128): Add support for the IEEE 128-bit square root and fma
13757 built-in functions.
13758 (FMAF128): Likewise.
13760 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13761 support for built-in functions that need the ISA 3.0 IEEE 128-bit
13762 floating point instructions.
13763 (rs6000_invalid_builtin): Likewise.
13764 (rs6000_builtin_mask_names): Likewise.
13765 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
13766 (RS6000_BTM_FLOAT128_HW): Likewise.
13767 (RS6000_BTM_COMMON): Likewise.
13768 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
13770 * doc/extend.texi (RS/6000 built-in functions): Document the
13771 IEEE 128-bit floating point square root and fused multiply-add
13772 built-in functions.
13774 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
13776 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
13777 reg (r2) isn't in the set of registers defined in the prologue.
13779 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13780 Alan Hayward <alan.hayward@arm.com>
13781 David Sherwood <david.sherwood@arm.com>
13783 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
13784 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
13785 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
13786 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
13787 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
13789 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
13790 max_vectorization_factor.
13791 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
13793 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13794 Alan Hayward <alan.hayward@arm.com>
13795 David Sherwood <david.sherwood@arm.com>
13797 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
13798 (vect_worthwhile_without_simd_p): Declare.
13799 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
13800 (vectorizable_reduction): Use it.
13801 * tree-vect-stmts.c (vectorizable_shift): Likewise.
13802 (vectorizable_operation): Likewise.
13804 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13805 Alan Hayward <alan.hayward@arm.com>
13806 David Sherwood <david.sherwood@arm.com>
13808 * tree-vectorizer.h (vect_get_num_copies): New function.
13809 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
13810 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13811 (vectorizable_induction): Likewise.
13812 (vectorizable_live_operation): Likewise.
13813 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13814 (vectorizable_bswap): Likewise.
13815 (vectorizable_call): Likewise.
13816 (vectorizable_conversion): Likewise.
13817 (vectorizable_assignment): Likewise.
13818 (vectorizable_shift): Likewise.
13819 (vectorizable_operation): Likewise.
13820 (vectorizable_store): Likewise.
13821 (vectorizable_load): Likewise.
13822 (vectorizable_condition): Likewise.
13823 (vectorizable_comparison): Likewise.
13824 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
13826 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13827 Alan Hayward <alan.hayward@arm.com>
13828 David Sherwood <david.sherwood@arm.com>
13830 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
13831 of vect_init_vector.
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 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
13838 an inline wrapper that provides a location.
13839 (gimple_build_vector): Likewise.
13840 * gimple-fold.c (gimple_build_vector_from_val): New function.
13841 (gimple_build_vector): Likewise.
13842 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
13843 functions to build the initial value. Always return a gimple value.
13844 (get_initial_defs_for_reduction): Likewise. Only compute
13846 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
13847 vect_init_vector on the results from get_initial_def(s)_for_reduction.
13848 (vectorizable_induction): Use gimple_build_vector rather than
13851 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13852 Alan Hayward <alan.hayward@arm.com>
13853 David Sherwood <david.sherwood@arm.com>
13855 * target.h (vec_perm_indices): New typedef.
13856 (auto_vec_perm_indices): Likewise.
13857 * optabs-query.h: Include target.h
13858 (can_vec_perm_p): Take a vec_perm_indices *.
13859 * optabs-query.c (can_vec_perm_p): Likewise.
13860 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
13861 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13862 * tree-vect-generic.c (lower_vec_perm): Likewise.
13863 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
13864 (vect_grouped_load_supported): Likewise.
13865 (vect_shift_permute_load_chain): Likewise.
13866 (vect_permute_store_chain): Use auto_vec_perm_indices.
13867 (vect_permute_load_chain): Likewise.
13868 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
13869 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
13870 Update uses of can_vec_perm_p.
13871 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
13872 mode with a number of elements. Take a vec_perm_indices *.
13873 (vect_create_epilog_for_reduction): Update accordingly.
13874 Use auto_vec_perm_indices.
13875 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
13876 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
13877 (vect_transform_slp_perm_load): Likewise.
13878 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
13879 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
13880 (vect_gen_perm_mask_checked): Likewise.
13881 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
13882 (vect_gen_perm_mask_checked): Likewise.
13883 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
13884 (vectorizable_store): Likewise.
13885 (vectorizable_load): Likewise.
13886 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
13887 (vectorizable_bswap): Likewise.
13889 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13890 Alan Hayward <alan.hayward@arm.com>
13891 David Sherwood <david.sherwood@arm.com>
13893 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
13894 * tree.c (build_vector): Likewise.
13895 (build_vector_from_ctor): Update accordingly.
13896 (build_vector_from_val): Likewise.
13897 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
13898 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13899 * tree-vect-generic.c (add_rshift): Likewise.
13900 (expand_vector_divmod): Likewise.
13901 (optimize_vector_constructor): Likewise.
13902 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
13903 (vect_transform_slp_perm_load): Likewise.
13904 (vect_schedule_slp_instance): Likewise.
13905 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
13906 (vectorizable_call): Likewise.
13907 (vect_gen_perm_mask_any): Likewise. Add elements in order.
13908 * expmed.c (make_tree): Likewise.
13909 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
13910 a vector passed to build_vector.
13911 (fold_convert_const): Likewise.
13912 (exact_inverse): Likewise.
13913 (fold_ternary_loc): Likewise.
13914 (fold_relational_const): Likewise.
13915 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
13916 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
13917 (const_unop): Likewise. Store the reduction accumulator in a
13918 variable rather than an array.
13919 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
13920 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
13921 the new vector, rather than constructing it after the input arrays.
13922 (native_interpret_vector): Use auto_vec<tree> when building
13923 a vector passed to build_vector. Add elements in order.
13924 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
13925 auto_vec<tree> when building a vector passed to build_vector.
13926 (vect_create_epilog_for_reduction): Likewise.
13927 (vectorizable_induction): Likewise.
13928 (get_initial_def_for_reduction): Likewise. Fix indentation of
13930 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
13932 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
13933 passed to build_vector.
13935 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13936 Alan Hayward <alan.hayward@arm.com>
13937 David Sherwood <david.sherwood@arm.com>
13939 * tree-core.h (tree_base::u): Add an "nelts" field.
13940 (tree_vector): Use VECTOR_CST_NELTS as the length.
13941 * tree.c (tree_size): Likewise.
13942 (make_vector): Initialize VECTOR_CST_NELTS.
13943 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
13944 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
13945 TYPE_VECTOR_SUBPARTS.
13946 * expr.c (const_vector_mask_from_tree): Consistently use "units"
13947 as the number of units, setting it from VECTOR_CST_NELTS.
13948 (const_vector_from_tree): Likewise.
13949 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
13950 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
13951 (fold_negate_expr_1): Likewise.
13952 (fold_convert_const): Likewise.
13953 (const_binop): Likewise. Differentiate the number of output and
13955 (const_unop): Likewise.
13956 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
13957 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
13958 in cases that did the opposite.
13960 2017-09-14 Richard Biener <rguenther@suse.de>
13962 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
13965 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
13967 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
13969 2017-09-14 Jakub Jelinek <jakub@redhat.com>
13972 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
13973 if and where to split a bb, except for splitting before debug insn
13974 sequences followed by non-label real insn. Delete debug insns
13975 in between basic blocks.
13977 * combine.c (make_compound_operation_int): Formatting fixes.
13979 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
13980 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13981 * config/netbsd.h (LINK_EH_SPEC): Likewise.
13982 * config/sol2.h (LINK_EH_SPEC): Likewise.
13983 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13984 * config/s390/linux.h (LINK_SPEC): Likewise.
13985 * config/freebsd.h (LINK_EH_SPEC): Likewise.
13986 * config/openbsd.h (LINK_EH_SPEC): Likewise.
13987 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13988 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
13989 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
13990 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13991 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
13992 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
13994 2017-09-13 Jakub Jelinek <jakub@redhat.com>
13996 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
13998 (ENDFILE_LINUX_SPEC): Likewise.
13999 (LINK_EH_SPEC): Likewise.
14000 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
14001 (LINK_OS_LINUX_SPEC32): Likewise.
14002 (LINK_OS_LINUX_SPEC64): Likewise.
14003 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
14004 (LINK_OS_LINUX_SPEC): Likewise.
14006 2017-09-13 Martin Liska <mliska@suse.cz>
14008 PR middle-end/82154
14009 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
14010 CASE_HIGH is NULL_TREE.
14012 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14013 Alan Hayward <alan.hayward@arm.com>
14014 David Sherwood <david.sherwood@arm.com>
14016 * target.def (secondary_memory_needed): New hook.
14017 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
14018 instead of SECONDARY_MEMORY_NEEDED.
14019 (secondary_memory_needed_mode): Likewise.
14020 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
14021 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
14022 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
14023 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
14024 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
14025 * doc/tm.texi: Regenerate.
14026 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
14027 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
14028 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14029 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
14030 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
14031 * config/i386/i386.c (inline_secondary_memory_needed): Put the
14032 mode argument first and change the reg_class arguments to reg_class_t.
14033 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
14034 Make static. Update the call to inline_secondary_memory_needed.
14035 (ix86_register_move_cost): Update the call to
14036 inline_secondary_memory_needed.
14037 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14038 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
14040 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
14041 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14043 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
14044 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
14045 * config/mips/mips.c (mips_secondary_memory_needed): Make static
14046 and match hook interface. Add comment from mips.h.
14047 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14048 * config/mmix/mmix.md (truncdfsf2): Refer to
14049 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14051 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
14052 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
14053 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14054 (pa_secondary_memory_needed): New function.
14055 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
14056 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
14057 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14058 (pdp11_secondary_memory_needed): Make static and match hook interface.
14059 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
14060 * config/powerpcspe/powerpcspe-protos.h
14061 (rs6000_secondary_memory_needed_ptr): Delete.
14062 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
14064 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14065 (rs6000_option_override_internal): Assign to
14066 targetm.secondary_memory_needed rather than
14067 rs6000_secondary_memory_needed_ptr.
14068 (rs6000_secondary_memory_needed): Match hook interface.
14069 (rs6000_debug_secondary_memory_needed): Likewise.
14070 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
14071 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
14072 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
14073 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14074 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
14075 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
14077 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
14078 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14079 (rs6000_option_override_internal): Assign to
14080 targetm.secondary_memory_needed rather than
14081 rs6000_secondary_memory_needed_ptr.
14082 (rs6000_secondary_memory_needed): Match hook interface.
14083 (rs6000_debug_secondary_memory_needed): Likewise.
14084 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
14085 * config/s390/s390.c (s390_secondary_memory_needed): New function.
14086 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14087 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
14088 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
14089 (sparc_secondary_memory_needed): New function.
14090 * lra-constraints.c (check_and_process_move): Refer to
14091 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
14093 (curr_insn_transform): Likewise.
14094 (process_alt_operands): Use targetm.secondary_memory_needed
14095 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14096 (check_secondary_memory_needed_p): Likewise.
14097 (choose_split_class): Likewise.
14098 * reload.c: Unconditionally include code that was previously
14099 conditional on SECONDARY_MEMORY_NEEDED.
14100 (push_secondary_reload): Use targetm.secondary_memory_needed
14101 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14102 (push_reload): Likewise.
14103 * reload1.c: Unconditionally include code that was previously
14104 conditional on SECONDARY_MEMORY_NEEDED.
14105 (choose_reload_regs): Use targetm.secondary_memory_needed
14106 instead of TARGET_SECONDARY_MEMORY_NEEDED.
14107 (gen_reload): Likewise.
14108 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
14110 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
14111 Alan Hayward <alan.hayward@arm.com>
14112 David Sherwood <david.sherwood@arm.com>
14114 * target.def (secondary_memory_needed_mode): New hook:
14115 * targhooks.c (default_secondary_memory_needed_mode): Declare.
14116 * targhooks.h (default_secondary_memory_needed_mode): New function.
14117 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
14118 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
14119 * doc/tm.texi: Regenerate.
14120 * lra-constraints.c (check_and_process_move): Use
14121 targetm.secondary_memory_needed_mode instead of
14122 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
14123 (curr_insn_transform): Likewise.
14124 * reload.c (get_secondary_mem): Likewise.
14125 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14126 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
14128 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14129 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14130 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
14131 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14132 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
14134 * config/powerpcspe/powerpcspe-protos.h
14135 (rs6000_secondary_memory_needed_mode): Delete.
14136 * config/powerpcspe/powerpcspe.c
14137 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14138 (rs6000_secondary_memory_needed_mode): Make static.
14139 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14140 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
14142 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14144 (rs6000_secondary_memory_needed_mode): Make static.
14145 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14146 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
14147 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
14148 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
14149 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
14151 (sparc_secondary_memory_needed_mode): New function.
14152 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
14154 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
14156 * config/aarch64/constraints.md (Umq): New constraint.
14157 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
14159 (mov<mode>): Update condition.
14161 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14163 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
14164 when bitposition is the same.
14166 2017-09-13 Richard Biener <rguenther@suse.de>
14168 * dwarf2out.c (output_die_symbol): Remove.
14169 (output_die): Do not output a DIEs symbol.
14171 2017-09-13 Richard Biener <rguenther@suse.de>
14173 PR middle-end/82128
14174 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
14175 default-def to avoid breaking iterator update with the weird
14176 interaction with cgraph_update_edges_for_call_stmt_node.
14178 2017-09-13 Richard Biener <rguenther@suse.de>
14180 * tree-cfg.c (verify_gimple_assign_binary): Add verification
14181 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
14182 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
14183 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
14185 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14187 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
14188 Disable pc relative literal load irrespective of
14189 TARGET_FIX_ERR_A53_84341 for default.
14191 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
14193 * config/sparc/sparc.c (output_return): Output the source location of
14194 the insn in the delay slot, if any.
14195 (output_sibcall): Likewise.
14197 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
14200 * common.opt (-static-pie): New alias.
14201 (shared): Negate static-pie.
14202 (-no-pie): Update help text.
14204 (static-pie): New option.
14205 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
14206 -static-pie support.
14207 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
14208 (LINK_EH_SPEC): Likewise.
14209 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
14210 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14211 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
14212 * gcc.c (LINK_COMMAND_SPEC): Likewise.
14213 (init_gcc_specs): Likewise.
14214 (init_spec): Likewise.
14215 (display_help): Update help message for -pie.
14216 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
14219 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
14221 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
14222 (movdi_aarch64): Likewise.
14223 (movti_aarch64): Likewise.
14225 2017-09-12 Simon Wright <simon@pushface.org>
14228 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
14229 calculation of the minor version, always output as 0.
14231 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14234 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14235 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
14236 on it early, rather than manual conversion late. For
14237 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
14238 instead of performing manual conversion.
14240 2017-09-12 Carl Love <cel@us.ibm.com>
14242 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
14243 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
14245 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14246 VMULOSW): Add definitions.
14247 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14248 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14249 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14250 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14251 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14253 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14255 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
14257 (movti_aarch64): Likewise.
14258 (movdf_aarch64): Likewise.
14259 (movtf_aarch64): Likewise.
14260 (load_pairdi): Likewise.
14261 (store_pairdi): Likewise.
14262 (load_pairdf): Likewise.
14263 (store_pairdf): Likewise.
14264 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
14265 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
14266 (ldr_got_small_<mode>): Likewise.
14267 (ldr_got_small_28k_<mode>): Likewise.
14268 (ldr_got_tiny): Likewise.
14269 * config/aarch64/iterators.md (ldst_sz): New.
14270 (ldpstp_sz): Likewise.
14271 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
14273 (thunderx_load): Split load_8 to load_16.
14274 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
14276 (thunderx2t99_storepair_basic): Split store_8 to store_16.
14277 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
14278 (xgene1_store_pair): Split store_8 to store_16.
14279 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
14280 (falkor_st_0_st_sd): Split store_8 to store_16.
14282 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
14284 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
14285 and store1/2/3/4 to store_4/8/12/16.
14286 * config/aarch64/aarch64.md: Update for rename.
14287 * config/arm/arm.md: Likewise.: Likewise.
14288 * config/arm/arm.c: Likewise.
14289 * config/arm/thumb1.md: Likewise.
14290 * config/arm/thumb2.md: Likewise.
14291 * config/arm/vfp.md: Likewise.
14292 * config/arm/arm-generic.md: Likewise.
14293 * config/arm/arm1020e.md: Likewise.
14294 * config/arm/arm1026ejs.md: Likewise.
14295 * config/arm/arm1136jfs.md: Likewise.
14296 * config/arm/arm926ejs.md: Likewise.
14297 * config/arm/cortex-a15.md: Likewise.
14298 * config/arm/cortex-a17.md: Likewise.
14299 * config/arm/cortex-a5.md: Likewise.
14300 * config/arm/cortex-a53.md: Likewise.
14301 * config/arm/cortex-a57.md: Likewise.
14302 * config/arm/cortex-a7.md: Likewise.
14303 * config/arm/cortex-a8.md: Likewise.
14304 * config/arm/cortex-a9.md: Likewise.
14305 * config/arm/cortex-m4.md: Likewise.
14306 * config/arm/cortex-m7.md: Likewise.
14307 * config/arm/cortex-r4.md: Likewise.
14308 * config/arm/exynos-m1.md: Likewise.
14309 * config/arm/fa526.md: Likewise.
14310 * config/arm/fa606te.md: Likewise.
14311 * config/arm/fa626te.md: Likewise.
14312 * config/arm/fa726te.md: Likewise.
14313 * config/arm/fmp626.md: Likewise.
14314 * config/arm/iwmmxt.md: Likewise.
14315 * config/arm/ldmstm.md: Likewise.
14316 * config/arm/marvell-pj4.md: Likewise.
14317 * config/arm/xgene1.md: Likewise.
14318 * config/aarch64/thunderx.md: Likewise.
14319 * config/aarch64/thunderx2t99.md: Likewise.
14320 * config/aarch64/falkor.md: Likewise.
14322 2017-09-12 Martin Liska <mliska@suse.cz>
14324 * attribs.c (private_lookup_attribute): New function.
14325 * attribs.h (private_lookup_attribute): Declared here.
14326 (lookup_attribute): Called from this place.
14328 2017-09-12 Richard Biener <rguenther@suse.de>
14330 PR tree-optimization/82157
14331 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
14332 stmts with side-effects.
14334 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14335 Alan Hayward <alan.hayward@arm.com>
14336 David Sherwood <david.sherwood@arm.com>
14338 * target.def (hard_regno_nregs): New hook.
14339 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
14340 * targhooks.h (default_hard_regno_nregs): Declare.
14341 * targhooks.c (default_hard_regno_nregs): New function.
14342 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
14343 (TARGET_HARD_REGNO_NREGS): ...this hook.
14344 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
14345 (CLASS_MAX_NREGS): Likewise.
14346 * doc/tm.texi: Regenerate.
14347 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
14348 instead of HARD_REGNO_NREGS.
14349 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
14350 HARD_REGNO_NREGS in the comment.
14351 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
14352 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
14353 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
14354 Return an unsigned int.
14355 (TARGET_HARD_REGNO_NREGS): Redefine.
14356 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
14357 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
14358 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14359 (arc_hard_regno_nregs): New function.
14360 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
14361 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
14362 (arm_hard_regno_nregs): New function.
14363 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
14364 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
14365 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
14366 (TARGET_HARD_REGNO_NREGS): Redefine.
14367 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
14368 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
14369 (HARD_REGNO_NREGS): Delete.
14370 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
14371 (cr16_hard_regno_nregs): New function.
14372 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
14373 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
14374 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
14375 (cris_hard_regno_nregs): New function.
14376 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
14377 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
14378 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14379 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
14380 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
14381 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
14382 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
14383 (frv_hard_regno_nregs): Make static. Take and return an
14385 (frv_class_max_nregs): Remove outdated copy of documentation.
14386 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
14387 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
14388 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
14389 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
14390 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
14391 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
14392 (TARGET_HARD_REGNO_NREGS): Redefine.
14393 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
14394 (CLASS_MAX_NREGS): Update comment.
14395 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
14396 (ia64_hard_regno_nregs): New function.
14397 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
14398 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
14399 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
14400 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
14401 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
14403 (m32c_hard_regno_nregs): Likewise. Make static.
14404 (TARGET_HARD_REGNO_NREGS): Redefine.
14405 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
14406 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
14407 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
14408 (m68k_hard_regno_nregs): New function.
14409 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
14410 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
14411 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
14412 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
14413 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
14414 Take and return an unsigned int.
14415 (TARGET_HARD_REGNO_NREGS): Redefine.
14416 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
14417 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
14418 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
14419 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
14420 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
14421 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
14422 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
14423 (msp430_hard_regno_nregs): Make static. Take and return an
14425 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
14426 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
14427 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
14428 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
14429 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
14430 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
14431 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
14432 (TARGET_HARD_REGNO_NREGS): Redefine.
14433 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
14434 (PA_HARD_REGNO_NREGS): ...this.
14435 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
14436 (PA_HARD_REGNO_NREGS): ...this.
14437 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14438 (pa_hard_regno_nregs): New function.
14439 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
14440 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
14441 (pdp11_hard_regno_nregs): New function.
14442 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
14443 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
14444 (rs6000_hard_regno_nregs_hook): New function.
14445 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
14446 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
14447 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
14448 Take and return an unsigned int. Move earlier in file.
14449 (TARGET_HARD_REGNO_NREGS): Redefine.
14450 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
14451 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
14452 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
14453 (rl78_hard_regno_nregs): Make static. Take and return an
14455 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
14456 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
14457 (rs6000_hard_regno_nregs_hook): New function.
14458 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
14459 * config/rx/rx.c (rx_hard_regno_nregs): New function.
14460 (TARGET_HARD_REGNO_NREGS): Redefine.
14461 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
14462 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
14463 instead of HARD_REGNO_NREGS.
14464 (s390_hard_regno_nregs): New function.
14465 (s390_hard_regno_mode_ok): Add comment from s390.h.
14466 (TARGET_HARD_REGNO_NREGS): Redefine.
14467 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
14468 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
14469 (sh_hard_regno_nregs): New function.
14470 (sh_pass_in_reg_p): Use it.
14471 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
14472 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
14473 (sparc_hard_regno_nregs): New function.
14474 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
14475 * config/spu/spu.c (spu_hard_regno_nregs): New function.
14476 (spu_function_arg_advance): Use it, supplying a valid register number.
14477 (TARGET_HARD_REGNO_NREGS): Redefine.
14478 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
14479 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
14480 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
14481 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
14482 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
14483 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
14484 (CLASS_MAX_NREGS): Remove copy of old documentation.
14485 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
14486 (visium_hard_regno_nregs): New function.
14487 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
14488 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
14489 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
14490 xtensa_hard_regno_nregs): New function.
14491 * system.h (HARD_REGNO_NREGS): Poison.
14493 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14495 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
14496 hard_regno_nregs instead of HARD_REGNO_NREGS.
14497 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
14498 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
14499 (c6x_expand_epilogue): Likewise.
14500 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
14501 (frv_read_iacc_argument): Likewise.
14502 * config/sh/sh.c: Include regs.h.
14503 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14504 (regs_used): Likewise.
14505 (output_stack_adjust): Likewise.
14506 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
14507 * expmed.c: Include regs.h.
14508 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
14509 * ree.c: Include regs.h.
14510 (combine_reaching_defs): Use hard_regno_nregs instead of
14512 (add_removable_extension): Likewise.
14514 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14516 * regs.h (hard_regno_nregs): Turn into a function.
14517 (end_hard_regno): Update accordingly.
14518 * caller-save.c (setup_save_areas): Likewise.
14519 (save_call_clobbered_regs): Likewise.
14520 (replace_reg_with_saved_mem): Likewise.
14521 (insert_restore): Likewise.
14522 (insert_save): Likewise.
14523 * combine.c (can_change_dest_mode): Likewise.
14524 (move_deaths): Likewise.
14525 (distribute_notes): Likewise.
14526 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
14527 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
14528 (rs6000_split_multireg_move): Likewise.
14529 (rs6000_register_move_cost): Likewise.
14530 (rs6000_memory_move_cost): Likewise.
14531 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
14532 (rs6000_split_multireg_move): Likewise.
14533 (rs6000_register_move_cost): Likewise.
14534 (rs6000_memory_move_cost): Likewise.
14535 * cselib.c (cselib_reset_table): Likewise.
14536 (cselib_lookup_1): Likewise.
14537 * emit-rtl.c (set_mode_and_regno): Likewise.
14538 * function.c (aggregate_value_p): Likewise.
14539 * ira-color.c (setup_profitable_hard_regs): Likewise.
14540 (check_hard_reg_p): Likewise.
14541 (calculate_saved_nregs): Likewise.
14542 (assign_hard_reg): Likewise.
14543 (improve_allocation): Likewise.
14544 (calculate_spill_cost): Likewise.
14545 * ira-emit.c (modify_move_list): Likewise.
14546 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
14547 (ira_hard_reg_in_set_p): Likewise.
14548 * ira.c (setup_reg_mode_hard_regset): Likewise.
14549 (clarify_prohibited_class_mode_regs): Likewise.
14550 (check_allocation): Likewise.
14551 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14552 (lra_setup_reg_renumber): Likewise.
14553 (setup_try_hard_regno_pseudos): Likewise.
14554 (spill_for): Likewise.
14555 (assign_hard_regno): Likewise.
14556 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
14557 * lra-constraints.c (in_class_p): Likewise.
14558 (lra_constraint_offset): Likewise.
14559 (simplify_operand_subreg): Likewise.
14560 (lra_constraints): Likewise.
14561 (split_reg): Likewise.
14562 (split_if_necessary): Likewise.
14563 (invariant_p): Likewise.
14564 (inherit_in_ebb): Likewise.
14565 * lra-lives.c (process_bb_lives): Likewise.
14566 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
14567 (get_hard_regs): Likewise.
14568 (do_remat): Likewise.
14569 * lra-spills.c (assign_spill_hard_regs): Likewise.
14570 * mode-switching.c (create_pre_exit): Likewise.
14571 * postreload.c (reload_combine_recognize_pattern): Likewise.
14572 * recog.c (peep2_find_free_register): Likewise.
14573 * regcprop.c (kill_value_regno): Likewise.
14574 (set_value_regno): Likewise.
14575 (copy_value): Likewise.
14576 (maybe_mode_change): Likewise.
14577 (find_oldest_value_reg): Likewise.
14578 (copyprop_hardreg_forward_1): Likewise.
14579 * regrename.c (check_new_reg_p): Likewise.
14580 (regrename_do_replace): Likewise.
14581 * reload.c (push_reload): Likewise.
14582 (combine_reloads): Likewise.
14583 (find_dummy_reload): Likewise.
14584 (operands_match_p): Likewise.
14585 (find_reloads): Likewise.
14586 (find_equiv_reg): Likewise.
14587 (reload_adjust_reg_for_mode): Likewise.
14588 * reload1.c (count_pseudo): Likewise.
14589 (count_spilled_pseudo): Likewise.
14590 (find_reg): Likewise.
14591 (clear_reload_reg_in_use): Likewise.
14592 (free_for_value_p): Likewise.
14593 (allocate_reload_reg): Likewise.
14594 (choose_reload_regs): Likewise.
14595 (reload_adjust_reg_for_temp): Likewise.
14596 (emit_reload_insns): Likewise.
14597 (delete_output_reload): Likewise.
14598 * rtlanal.c (subreg_get_info): Likewise.
14599 * sched-deps.c (sched_analyze_reg): Likewise.
14600 * sel-sched.c (init_regs_for_mode): Likewise.
14601 (mark_unavailable_hard_regs): Likewise.
14602 (choose_best_reg_1): Likewise.
14603 (verify_target_availability): Likewise.
14604 * valtrack.c (dead_debug_insert_temp): Likewise.
14605 * var-tracking.c (track_loc_p): Likewise.
14606 (emit_note_insn_var_location): Likewise.
14607 * varasm.c (make_decl_rtl): Likewise.
14608 * reginfo.c (choose_hard_reg_mode): Likewise.
14609 (init_reg_modes_target): Refer directly to
14610 this_target_regs->x_hard_regno_nregs.
14612 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14614 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
14615 instead of hard_regno_nregs.
14617 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14619 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
14620 end_hard_regno instead of hard_regno_nregs.
14621 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
14622 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
14623 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
14624 * ira-color.c (improve_allocation): Likewise.
14625 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14626 * lra-lives.c (mark_regno_live): Likewise.
14627 (mark_regno_dead): Likewise.
14628 * lra-remat.c (operand_to_remat): Likewise.
14629 * lra.c (collect_non_operand_hard_regs): Likewise.
14630 * postreload.c (reload_combine_note_store): Likewise.
14631 (move2add_valid_value_p): Likewise.
14632 * reload.c (regno_clobbered_p): Likewise.
14634 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14636 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
14638 * config/v850/v850.c (v850_reorg): Likewise.
14639 * reload.c (refers_to_regno_for_reload_p): Likewise.
14640 (find_equiv_reg): Likewise.
14641 * reload1.c (reload_reg_reaches_end_p): Likewise.
14643 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14645 * caller-save.c (add_used_regs): Use REG_NREGS instead of
14647 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
14648 * config/arm/arm.c (output_move_neon): Likewise.
14649 (arm_attr_length_move_neon): Likewise.
14650 (neon_split_vcombine): Likewise.
14651 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
14652 (c6x_mark_reg_written): Likewise.
14653 (c6x_dwarf_register_span): Likewise.
14654 * config/i386/i386.c (ix86_save_reg): Likewise.
14655 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
14656 (rws_access_reg): Likewise.
14657 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
14658 * mode-switching.c (create_pre_exit): Likewise.
14659 * ree.c (combine_reaching_defs): Likewise.
14660 (add_removable_extension): Likewise.
14661 * regcprop.c (find_oldest_value_reg): Likewise.
14662 (copyprop_hardreg_forward_1): Likewise.
14663 * reload.c (reload_inner_reg_of_subreg): Likewise.
14664 (push_reload): Likewise.
14665 (combine_reloads): Likewise.
14666 (find_dummy_reload): Likewise.
14667 (reload_adjust_reg_for_mode): Likewise.
14668 * reload1.c (find_reload_regs): Likewise.
14669 (forget_old_reloads_1): Likewise.
14670 (reload_reg_free_for_value_p): Likewise.
14671 (reload_adjust_reg_for_temp): Likewise.
14672 (emit_reload_insns): Likewise.
14673 (delete_output_reload): Likewise.
14674 * sel-sched.c (choose_best_reg_1): Likewise.
14675 (choose_best_pseudo_reg): Likewise.
14677 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14678 Alan Hayward <alan.hayward@arm.com>
14679 David Sherwood <david.sherwood@arm.com>
14681 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
14682 * target.def (slow_unaligned_access): New hook.
14683 * targhooks.h (default_slow_unaligned_access): Declare.
14684 * targhooks.c (default_slow_unaligned_access): New function.
14685 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
14686 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
14687 * doc/tm.texi: Regenerate.
14688 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
14689 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
14690 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
14692 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
14693 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
14695 (rs6000_slow_unaligned_access): New function.
14696 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
14697 (expand_block_compare): Likewise.
14698 (expand_strn_compare): Likewise.
14699 (rs6000_rtx_costs): Likewise.
14700 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
14701 (riscv_slow_unaligned_access): Likewise.
14702 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
14703 (riscv_slow_unaligned_access_p): ...this and make static.
14704 (riscv_option_override): Update accordingly.
14705 (riscv_slow_unaligned_access): New function.
14706 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
14707 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
14708 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
14709 (rs6000_slow_unaligned_access): New function.
14710 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
14711 (rs6000_rtx_costs): Likewise.
14712 * config/rs6000/rs6000-string.c (expand_block_compare)
14713 (expand_strn_compare): Use targetm.slow_unaligned_access instead
14714 of SLOW_UNALIGNED_ACCESS.
14715 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
14716 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
14717 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
14718 of SLOW_UNALIGNED_ACCESS.
14719 * expmed.c (simple_mem_bitfield_p): Likewise.
14720 * expr.c (alignment_for_piecewise_move): Likewise.
14721 (emit_group_load_1): Likewise.
14722 (emit_group_store): Likewise.
14723 (copy_blkmode_from_reg): Likewise.
14724 (emit_push_insn): Likewise.
14725 (expand_assignment): Likewise.
14726 (store_field): Likewise.
14727 (expand_expr_real_1): Likewise.
14728 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
14729 * lra-constraints.c (simplify_operand_subreg): Likewise.
14730 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
14731 * gimple-ssa-store-merging.c: Likewise in block comment at start
14733 * tree-ssa-strlen.c: Include target.h.
14734 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
14735 of SLOW_UNALIGNED_ACCESS.
14736 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
14738 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
14740 PR rtl-optimization/82185
14741 * expmed.c (emit_store_flag_int): Only test tem if it has been
14744 2017-09-12 Richard Biener <rguenther@suse.de>
14746 PR middle-end/82149
14747 * match.pd ((FTYPE) N CMP CST): Fix typo.
14749 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
14751 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
14753 (mips_near_type_p): Add 'short_call' attribute as a synonym
14755 * doc/extend.texi (short_call): Document new function attribute.
14757 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14760 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
14761 assertion check that in the condition.
14762 (get_atomic_generic_size): Likewise. Before testing if parameter
14763 has pointer type, if it has array type, call for C++
14764 default_conversion to perform array-to-pointer conversion.
14766 2017-09-12 Richard Biener <rguenther@suse.de>
14768 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
14769 for operations we cannot scalarize.
14771 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
14773 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
14774 vectors heap vectors. Clean up comments.
14775 Make visited_bbs a reference.
14776 (profitable_jump_thread_path): Make GC
14777 vectors heap vectors. Clean up comments.
14779 (convert_and_register_jump_thread_path): Make GC vectors heap
14781 (check_subpath_and_update_thread_path): Same. Clean up comments.
14782 Make visited_bbs a reference.
14783 (handle_phi): Abstract common code to to
14784 register_jump_thread_path_if_profitable.
14785 Rename VAR_BB to DEF_BB.
14787 Make GC vectors heap vectors.
14788 Make visited_bbs a reference.
14789 (handle_assignment): Same.
14790 (register_jump_thread_path_if_profitable): New.
14791 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
14793 Make GC vectors heap vectors. Clean up comments.
14794 Make visited_bbs a reference.
14795 (find_jump_threads_backwards): Make visited_bbs live in the stack.
14796 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
14799 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
14802 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
14803 words of E_DImode object are reachable by xtensa_uimm8x4 access.
14805 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
14807 Revert r251800 and r251799.
14809 2017-09-11 Martin Jambor <mjambor@suse.cz>
14812 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
14813 arguments in advance.
14814 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
14815 use it to find predecessor edges.
14816 (naive_outof_ssa): Collect vector of predecessors.
14818 2017-09-08 Jason Merrill <jason@redhat.com>
14820 PR c++/70029 - ICE with ref-qualifier and -flto
14821 * langhooks.h (struct lang_hooks_for_types): Add
14822 copy_lang_qualifiers.
14823 * attribs.c (build_type_attribute_qual_variant): Use it.
14824 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
14826 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
14827 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
14829 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
14832 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
14833 (*mulsi3_sp64): New instruction.
14834 (mulsi3): New expander.
14836 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
14838 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
14840 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14842 * sancov.c: Include memmodel.h.
14844 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
14847 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
14850 2017-09-07 Carl Love <cel@us.ibm.com>
14852 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
14853 the sldi instruction.
14855 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
14857 * sancov.c: Include tm_p.h.
14859 2017-09-07 Jakub Jelinek <jakub@redhat.com>
14862 * output.h (switch_to_other_text_partition): New declaration.
14863 * varasm.c (switch_to_other_text_partition): New function.
14864 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
14865 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
14866 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
14867 to the other text partition before emitting LCL label and switch back
14868 after emitting the word after it.
14870 2017-09-07 Richard Biener <rguenther@suse.de>
14872 * passes.def (pass_split_crit_edges): Remove instance before PRE.
14873 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
14874 critical edges here, after loop init.
14875 (pass_data_pre): Remove PROP_no_crit_edges flags.
14876 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
14877 for valueization of call args to avoid leaking VN_TOP.
14878 (visit_use): Assert we do not visit default defs.
14879 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
14880 Use error_mark_node to more easily detect leaking VN_TOP.
14881 All default-defs are varying, not VN_TOP. Mark them visited.
14882 (run_scc_vn): Make code match comment.
14884 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14886 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
14887 OPTION_MASK_FLOAT128_KEYWORD.
14888 (POWERPC_MASKS): Likewise.
14889 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
14890 support for the -mfloat128-type option, and make -mfloat128
14891 default on PowerPC Linux systems. Define or undefine
14892 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
14893 Define __float128 to be __ieee128 if IEEE 128-bit support is
14894 enabled, or undefine it.
14895 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
14896 Delete defining __FLOAT128_TYPE__.
14897 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
14898 -mfloat128-type option and make -mfloat128 default on PowerPC
14900 (TARGET_FLOAT128_TYPE): Likewise.
14901 (-mfloat128-type): Likewise.
14902 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14903 Delete the -mfloat128-type option and make -mfloat128 default on
14904 PowerPC Linux systems. Always use __ieee128 to be the keyword for
14905 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
14906 128-bit floating point is enabled. Change tests from using
14907 -mfloat128-type to -mfloat128.
14908 (rs6000_mangle_type): Use the correct mangling for the __float128
14909 type even if normal long double is restricted to 64-bits.
14910 (floatn_mode): Enable the _Float128 type by default on VSX Linux
14912 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
14913 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
14914 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
14916 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
14917 documentation for -mfloat128.
14919 2017-09-06 Olivier Hainque <hainque@adacore.com>
14921 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
14923 2017-09-06 Wish Wu <wishwu007@gmail.com>
14924 Jakub Jelinek <jakub@redhat.com>
14926 * asan.c (initialize_sanitizer_builtins): Add
14927 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
14928 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
14929 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
14930 BT_FN_VOID_UINT64_PTR variables.
14931 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
14932 (BT_FN_VOID_UINT16_UINT16): Likewise.
14933 (BT_FN_VOID_UINT32_UINT32): Likewise.
14934 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
14935 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
14936 (BT_FN_VOID_UINT64_PTR): Likewise.
14937 * common.opt (flag_sanitize_coverage): New variable.
14938 (fsanitize-coverage=trace-pc): Remove.
14939 (fsanitize-coverage=): Add.
14940 * flag-types.h (enum sanitize_coverage_code): New enum.
14941 * fold-const.c (fold_range_test): Disable non-short-circuit
14942 optimization if flag_sanitize_coverage.
14943 (fold_truth_andor): Likewise.
14944 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14945 * opts.c (COVERAGE_SANITIZER_OPT): Define.
14946 (coverage_sanitizer_opts): New array.
14947 (get_closest_sanitizer_option): Add OPTS argument, handle also
14948 OPT_fsanitize_coverage_.
14949 (parse_sanitizer_options): Adjusted to also handle
14950 OPT_fsanitize_coverage_.
14951 (common_handle_option): Add OPT_fsanitize_coverage_.
14952 * sancov.c (instrument_comparison, instrument_switch): New function.
14953 (sancov_pass): Add trace-cmp support.
14954 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
14955 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
14956 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
14957 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
14958 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
14959 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
14960 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
14961 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
14962 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
14963 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
14965 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
14967 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14968 error. Only quit immediately if parsing is complete.
14969 (BEGIN): Initialize fatal_err and parse_done.
14970 (begin fpu, end fpu): Check number of arguments.
14971 (begin arch, end arch): Likewise.
14972 (begin cpu, end cpu): Likewise.
14973 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14974 (optalias): Likewise.
14976 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
14978 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14979 * config/arm/arm-isa.h: Delete. Move definitions to ...
14980 * arm-cpus.in: ... here. Use new feature and fgroup values.
14981 * config/arm/arm.c (arm_option_override): Use lower case for feature
14983 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14984 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14985 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14987 (print_isa_bits_for): New function.
14988 (gen_isa): New function.
14989 (gen_comm_data): Use print_isa_bits_for.
14990 (define feature): New keyword.
14991 (define fgroup): New keyword.
14992 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
14993 (arm-isa.h): Add rule to generate file.
14994 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14995 case for feature bit names.
14997 2017-09-06 Richard Biener <rguenther@suse.de>
14999 * tree-ssa-pre.c (NECESSARY): Remove.
15000 (create_expression_by_pieces): Do not touch pass-local flags.
15001 (insert_into_preds_of_block): Likewise.
15002 (do_pre_regular_insertion): Likewise.
15003 (eliminate_insert): Likewise.
15004 (eliminate_dom_walker::before_dom_children): Likewise.
15005 (fini_eliminate): Do not look at inserted_exprs.
15006 (mark_operand_necessary): Remove.
15007 (remove_dead_inserted_code): Replace with simple work-list
15008 algorithm based on inserted_exprs and SSA uses.
15009 (pass_pre::execute): Re-order fini_eliminate and
15010 remove_dead_inserted_code.
15012 2017-09-06 Olivier Hainque <hainque@adacore.com>
15014 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
15015 for VxWorks 7. Adjust surrounding comments.
15017 2017-09-06 Richard Biener <rguenther@suse.de>
15019 * gimple-ssa-strength-reduction.c
15020 (find_candidates_dom_walker::before_dom_children): Also allow
15023 2017-09-06 Richard Biener <rguenther@suse.de>
15025 PR tree-optimization/82108
15026 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
15027 for gap in the non-permutation SLP case.
15029 2017-09-06 Martin Jambor <mjambor@suse.cz>
15031 PR tree-optimization/82078
15032 * tree-sra.c (sort_and_splice_var_accesses): Move call to
15033 add_access_to_work_queue...
15034 (build_accesses_from_assign): ...here.
15035 (propagate_all_subaccesses): Make sure racc is the group
15036 representative, if there is one.
15038 2017-09-06 Jakub Jelinek <jakub@redhat.com>
15040 PR middle-end/82095
15041 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
15044 2017-09-06 Richard Biener <rguenther@suse.de>
15046 * gimple-ssa-strength-reduction.c
15047 (find_candidates_dom_walker::before_doom_children): Use a
15048 type and not a mode check.
15050 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
15053 * config/arm/predicates.md (arm_general_adddi_operand): Create new
15055 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
15057 2017-09-05 Jeff Law <law@redhat.com>
15059 PR tree-optimization/64910
15060 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
15061 cases where we have 3 or more operands.
15063 2017-09-05 Jakub Jelinek <jakub@redhat.com>
15065 PR middle-end/81768
15066 * omp-low.c (lower_omp_for): Recompute tree invariant if
15067 gimple_omp_for_initial/final is ADDR_EXPR.
15069 PR middle-end/81768
15070 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
15071 into gimple val before gimplification fo the COND_EXPR.
15073 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
15075 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
15076 REGION_COPY argument.
15077 (thread_through_all_blocks): Remove unused argument to
15078 duplicate_thread_path.
15080 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15081 Alan Hayward <alan.hayward@arm.com>
15082 David Sherwood <david.sherwood@arm.com>
15084 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
15085 Take a scalar_mode rather than a machine_mode.
15086 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15087 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
15088 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
15089 (aarch64_gen_adjusted_ldpstp): Likewise.
15090 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
15092 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15093 Alan Hayward <alan.hayward@arm.com>
15094 David Sherwood <david.sherwood@arm.com>
15096 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
15097 Take a scalar_int_mode instead of a machine_mode.
15098 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15099 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15100 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15101 (aarch64_simd_attr_length_rglist): Delete.
15102 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
15103 a scalar_int_mode instead of a machine_mode.
15104 (aarch64_add_offset): Likewise.
15105 (aarch64_internal_mov_immediate): Likewise
15106 (aarch64_add_constant_internal): Likewise.
15107 (aarch64_add_constant): Likewise.
15108 (aarch64_movw_imm): Likewise.
15109 (aarch64_rtx_arith_op_extract_p): Likewise.
15110 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
15111 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
15112 Remove assert that the mode isn't a vector.
15113 (aarch64_output_scalar_simd_mov_immediate): Likewise.
15114 (aarch64_expand_mov_immediate): Update calls after above changes.
15115 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
15116 (aarch64_and_bitmask_imm): Check for scalar integer modes.
15117 (aarch64_move_imm): Likewise.
15118 (aarch64_can_const_movi_rtx_p): Likewise.
15119 (aarch64_strip_extend): Likewise.
15120 (aarch64_extr_rtx_p): Likewise.
15121 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
15122 a CONST_INT when the mode parameter is VOIDmode.
15123 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
15125 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15127 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
15128 * stor-layout.c (bitwise_mode_for_mode): Likewise.
15129 (bitwise_type_for_mode): Update accordingly.
15131 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15133 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
15134 * stor-layout.c (mode_for_size_tree): Likewise.
15135 (mode_for_array): Update accordingly.
15136 (layout_decl): Likewise.
15137 (compute_record_mode): Likewise. Only set the mode once.
15139 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15141 * target.def (get_mask_mode): Change return type to opt_mode.
15143 * doc/tm.texi: Regenerate.
15144 * targhooks.h (default_get_mask_mode): Return an opt_mode.
15145 * targhooks.c (default_get_mask_mode): Likewise.
15146 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
15147 * optabs-query.c (can_vec_mask_load_store_p): Update use of
15148 targetm.get_mask_mode.
15149 * tree.c (build_truth_vector_type): Likewise.
15151 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15153 * machmode.h (mode_for_vector): Return an opt_mode.
15154 * stor-layout.c (mode_for_vector): Likewise.
15155 (mode_for_int_vector): Update accordingly.
15156 (layout_type): Likewise.
15157 * config/i386/i386.c (emit_memmov): Likewise.
15158 (ix86_expand_set_or_movmem): Likewise.
15159 (ix86_expand_vector_init): Likewise.
15160 (ix86_get_mask_mode): Likewise.
15161 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
15163 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
15164 * expmed.c (extract_bit_field_1): Likewise.
15165 * expr.c (expand_expr_real_2): Likewise.
15166 * optabs-query.c (can_vec_perm_p): Likewise.
15167 (can_vec_mask_load_store_p): Likewise.
15168 * optabs.c (expand_vec_perm): Likewise.
15169 * targhooks.c (default_get_mask_mode): Likewise.
15170 * tree-vect-stmts.c (vectorizable_store): Likewise.
15171 (vectorizable_load): Likewise.
15172 (get_vectype_for_scalar_type_and_size): Likewise.
15174 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15176 * machmode.h (mode_for_int_vector): New function.
15177 * stor-layout.c (mode_for_int_vector): Likewise.
15178 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
15179 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
15180 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
15181 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
15182 (s390_expand_vcond): Likewise.
15184 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15186 * machmode.h (opt_machine_mode): New type.
15187 (opt_mode<T>): Allow construction from anything that can be
15189 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
15190 (mode_for_size): Return an opt_machine_mode.
15191 * stor-layout.c (mode_for_size): Likewise.
15192 (mode_for_size_tree): Update call accordingly.
15193 (bitwise_mode_for_mode): Likewise.
15194 (make_fract_type): Likewise.
15195 (make_accum_type): Likewise.
15196 * caller-save.c (replace_reg_with_saved_mem): Update call
15198 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15199 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15200 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15201 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
15202 * expmed.c (extract_bit_field_1): Likewise.
15203 * reload.c (get_secondary_mem): Likewise.
15204 * varasm.c (assemble_integer): Likewise.
15205 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
15208 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15210 * machmode.h (decimal_float_mode_for_size): New function.
15211 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
15212 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
15213 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
15214 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
15215 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
15216 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
15218 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15220 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
15221 (get_builtin_sync_mode): Likewise.
15222 (expand_ifn_atomic_compare_exchange): Likewise.
15223 (expand_builtin_atomic_clear): Likewise.
15224 (expand_builtin_atomic_test_and_set): Likewise.
15225 (fold_builtin_atomic_always_lock_free): Likewise.
15226 * calls.c (compute_argument_addresses): Likewise.
15227 (emit_library_call_value_1): Likewise.
15228 (store_one_arg): Likewise.
15229 * combine.c (combine_instructions): Likewise.
15230 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
15231 * config/arm/arm.c (arm_function_value): Likewise.
15232 (aapcs_allocate_return_reg): Likewise.
15233 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
15234 * config/i386/i386.c (construct_container): Likewise.
15235 (ix86_gimplify_va_arg): Likewise.
15236 (ix86_expand_sse_cmp): Likewise.
15237 (emit_memmov): Likewise.
15238 (emit_memset): Likewise.
15239 (expand_small_movmem_or_setmem): Likewise.
15240 (ix86_expand_pextr): Likewise.
15241 (ix86_expand_pinsr): Likewise.
15242 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
15243 * config/microblaze/microblaze.c (microblaze_block_move_straight):
15245 * config/mips/mips.c (mips_function_value_1) Likewise.
15246 (mips_block_move_straight): Likewise.
15247 (mips_expand_ins_as_unaligned_store): Likewise.
15248 * config/powerpcspe/powerpcspe.c
15249 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15250 (rs6000_darwin64_record_arg_flush): Likewise.
15251 * config/rs6000/rs6000.c
15252 (rs6000_darwin64_record_arg_advance_flush): Likewise.
15253 (rs6000_darwin64_record_arg_flush): Likewise.
15254 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
15255 (sparc_function_value_1): Likewise.
15256 * config/spu/spu.c (adjust_operand): Likewise.
15257 (spu_emit_branch_or_set): Likewise.
15258 (arith_immediate_p): Likewise.
15259 * emit-rtl.c (gen_lowpart_common): Likewise.
15260 * expr.c (expand_expr_real_1): Likewise.
15261 * function.c (assign_parm_setup_block): Likewise.
15262 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
15263 * reload1.c (alter_reg): Likewise.
15264 * stor-layout.c (mode_for_vector): Likewise.
15265 (layout_type): Likewise.
15267 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
15269 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
15270 (spu_convert_move): Likewise.
15271 * lower-subreg.c (resolve_simple_move): Likewise.
15273 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15276 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
15277 define_insn to a define_expand.
15278 (altivec_vsum2sws_direct): New define_insn.
15279 (altivec_vsumsws): Convert from a define_insn to a define_expand.
15281 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15283 * config/arm/arm.c (arm_option_params_internal): Improve setting of
15286 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
15291 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
15292 realign stack if stack alignment needed is less than incoming
15295 2017-09-05 Marek Polacek <polacek@redhat.com>
15298 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
15301 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
15303 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
15305 2017-09-05 Richard Biener <rguenther@suse.de>
15307 PR tree-optimization/82084
15308 * fold-const.c (can_native_encode_string_p): Handle wide characters.
15310 2017-09-05 Richard Biener <rguenther@suse.de>
15312 PR tree-optimization/82102
15313 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
15315 2017-09-05 Martin Liska <mliska@suse.cz>
15317 PR tree-optimization/82032
15318 * tree-cfg.c (generate_range_test): New function.
15319 * tree-cfg.h (generate_range_test): Declared here.
15320 * tree-cfgcleanup.c (convert_single_case_switch): New function.
15321 (cleanup_control_expr_graph): Use it.
15322 * tree-switch-conversion.c (try_switch_expansion): Remove
15324 (emit_case_nodes): Use generate_range_test.
15326 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15329 * config/i386/i386.md (*<btsc><mode>_mask): Add
15330 TARGET_USE_BT to insn constraint.
15331 (*btr<mode>_mask): Ditto.
15333 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
15335 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
15336 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
15338 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
15341 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
15342 TARGET_NEON and TARGET_IWMMXT.
15343 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
15344 TARGET_NEON and TARGET_IWMMXT.
15345 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
15347 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
15349 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
15350 (ix86_rewrite_tls_address): Ditto.
15351 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
15352 (ix86_rewrite_tls_address_1): Ditto.
15353 (ix86_rewrite_tls_address): Ditto.
15354 * config/i386/predicates.md (tls_address_pattern): New predicate.
15355 * config/i386/i386.md (TLS address splitter): New splitter.
15357 2017-09-04 Richard Biener <rguenther@suse.de>
15359 PR tree-optimization/82084
15360 * fold-const.h (can_native_encode_string_p): Declare.
15361 * fold-const.c (can_native_encode_string_p): Factor out from ...
15362 (native_encode_string): ... here.
15363 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
15364 vectorizing stores from constants we later cannot handle.
15366 2017-09-04 Marek Polacek <polacek@redhat.com>
15369 * doc/invoke.texi: Update -Wtautological-compare documentation.
15371 2017-09-04 Jeff Law <law@redhat.com>
15373 PR tree-optimization/64910
15374 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
15375 swap the first and last operand if the last is a constant.
15377 2017-09-04 Marek Polacek <polacek@redhat.com>
15380 * convert.c (do_narrow): When sanitizing signed integer overflows,
15381 bail out for signed types.
15382 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
15384 2017-09-04 Richard Biener <rguenther@suse.de>
15386 PR tree-optimization/82060
15387 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15388 Move devirtualization after stmt folding and before EH/AB/noreturn
15389 cleanup to get the stmt refs canonicalized. Use a bool instead
15390 of gimple_modified_p since that doesn't work for NOPs. Schedule
15391 NOPs generated by folding for removal.
15393 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15394 Alan Hayward <alan.hayward@arm.com>
15395 David Sherwood <david.sherwood@arm.com>
15397 * coretypes.h (pad_direction): New enum.
15398 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
15399 (FUNCTION_ARG_PADDING): Likewise.
15400 * target.def (function_arg_padding): New hook.
15401 * targhooks.h (default_function_arg_padding): Declare.
15402 * targhooks.c (default_function_arg_padding): New function.
15403 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
15404 (TARGET_FUNCTION_ARG_PADDING): ...this.
15405 * doc/tm.texi: Regenerate.
15406 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
15407 instead of direction.
15408 (compute_argument_addresses): Likewise.
15409 (load_register_parameters): Likewise.
15410 (emit_library_call_value_1): Likewise.
15411 (store_one_arg): Use targetm.calls.function_arg_padding instead
15412 of FUNCTION_ARG_PADDING.
15413 (must_pass_in_stack_var_size_or_pad): Likewise.
15414 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
15415 (emit_group_store): Likewise.
15416 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
15417 instead of FUNCTION_ARG_PADDING.
15418 (emit_push_insn): Likewise, and propagate enum change throughout
15420 * function.h (direction): Delete.
15421 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
15423 * function.c (assign_parm_find_stack_rtl): Likewise.
15424 (assign_parm_setup_block_p): Likewise.
15425 (assign_parm_setup_block): Likewise.
15426 (gimplify_parameters): Likewise.
15427 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
15428 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
15430 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
15431 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15432 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
15433 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
15434 (aarch64_function_arg_padding): ...this new function.
15435 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
15436 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15437 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
15438 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15439 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
15440 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15441 (arm_pad_arg_upward): Replace with...
15442 (arm_function_arg_padding): ...this new function.
15443 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
15445 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
15446 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
15447 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15448 (ia64_hpux_function_arg_padding): Replace with...
15449 (ia64_function_arg_padding): ...this new function. Use pad_direction
15450 instead of direction. Check for TARGET_HPUX.
15451 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
15452 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15453 (iq2000_function_arg_padding): New function.
15454 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
15455 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
15456 (mips_function_arg_padding): ...this new function.
15457 (mips_pad_reg_upward): Update accordingly.
15458 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15459 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
15460 targetm.calls.function_arg_padding.
15461 (FUNCTION_ARG_PADDING): Delete.
15462 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
15463 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
15464 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15465 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
15466 (nios2_block_reg_padding): Return pad_direction instead of direction.
15467 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
15468 instead of direction.
15469 (nios2_function_arg_padding): Likewise. Make static.
15470 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15471 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
15472 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
15473 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
15474 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15475 (pa_function_arg_padding): Make static. Return pad_direction instead
15477 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
15478 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15479 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
15480 instead of direction. Use targetm.calls.function_arg_padding.
15481 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
15482 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15483 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
15484 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
15485 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
15487 (function_arg_padding): Rename to...
15488 (rs6000_function_arg_padding): ...this. Make static. Return
15489 pad_direction instead of direction.
15490 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15491 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
15492 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
15493 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
15494 instead of direction. Use targetm.calls.function_arg_padding.
15495 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
15496 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
15497 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
15498 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
15499 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15500 (function_arg_padding): Rename to...
15501 (rs6000_function_arg_padding): ...this. Make static. Return
15502 pad_direction instead of direction.
15503 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
15504 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
15505 * config/s390/s390.c (s390_function_arg_padding): New function.
15506 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15507 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
15508 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
15509 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
15510 (function_arg_padding): Rename to...
15511 (sparc_function_arg_padding): ...this. Make static. Return
15512 pad_direction instead of direction.
15513 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
15514 * config/spu/spu.c (spu_function_arg_padding): New function.
15515 (TARGET_FUNCTION_ARG_PADDING): Redefine.
15516 * system.h (FUNCTION_ARG_PADDING): Poison.
15518 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15519 Alan Hayward <alan.hayward@arm.com>
15520 David Sherwood <david.sherwood@arm.com>
15522 * target.def (modes_tieable_p): New hook.
15523 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
15524 (TARGET_MODES_TIEABLE_P): ...this.
15525 * doc/tm.texi.in: Regenerate.
15526 * hooks.h (hook_bool_mode_mode_true): Declare.
15527 * hooks.c (hook_bool_mode_mode_true): New function.
15528 * combine.c (subst): Use targetm.modes_tieable_p instead of
15530 * dse.c (find_shift_sequence): Likewise.
15531 * expmed.c (extract_low_bits): Likewise.
15532 * lower-subreg.c: Include target.h.
15533 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
15535 * rtlanal.c (rtx_cost): Likewise.
15536 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
15537 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
15538 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
15539 (TARGET_MODES_TIEABLE_P): Redefine.
15540 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
15541 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
15542 (TARGET_MODES_TIEABLE_P): Redefine.
15543 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
15544 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
15545 (arc_modes_tieable_p): New function.
15546 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
15547 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
15548 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
15549 (arm_modes_tieable_p): Make static.
15550 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
15551 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
15552 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
15553 (TARGET_MODES_TIEABLE_P): Redefine.
15554 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
15555 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
15556 (TARGET_MODES_TIEABLE_P): Redefine.
15557 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
15558 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
15559 (cr16_modes_tieable_p): New function.
15560 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
15561 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
15562 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
15563 (TRULY_NOOP_TRUNCATION): Update comment.
15564 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
15565 (TRULY_NOOP_TRUNCATION): Update comment.
15566 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
15567 (frv_modes_tieable_p): New function.
15568 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
15569 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
15570 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
15571 (TARGET_MODES_TIEABLE_P): Redefine.
15572 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
15573 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
15574 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
15575 (TARGET_MODES_TIEABLE_P): Redefine.
15576 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
15577 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
15578 (ia64_modes_tieable_p): New function.
15579 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
15580 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
15581 (iq2000_modes_tieable_p): New function.
15582 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
15583 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
15584 (lm32_modes_tieable_p): New function.
15585 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
15586 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
15587 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
15588 (TARGET_MODES_TIEABLE_P): Redefine.
15589 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
15590 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
15591 (m32r_modes_tieable_p): New function.
15592 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
15593 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
15594 (m68k_modes_tieable_p): New function.
15595 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
15596 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
15597 (mcore_modes_tieable_p): New function.
15598 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
15599 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
15601 (TARGET_MODES_TIEABLE_P): Redefine.
15602 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
15603 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
15604 * config/mips/mips.c (mips_modes_tieable_p): Make static.
15605 (TARGET_MODES_TIEABLE_P): Redefine.
15606 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
15607 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
15608 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
15609 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
15610 (mn10300_modes_tieable_p): ...this and make static.
15611 (TARGET_MODES_TIEABLE_P): Redefine.
15612 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
15613 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
15614 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
15615 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
15616 (msp430_modes_tieable_p): Make static.
15617 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
15618 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
15619 (TARGET_MODES_TIEABLE_P): Redefine.
15620 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
15621 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
15622 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
15623 (TARGET_MODES_TIEABLE_P): Redefine.
15624 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
15625 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
15626 * config/pa/pa.c (pa_modes_tieable_p): Make static.
15627 (TARGET_MODES_TIEABLE_P): Redefine.
15628 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
15629 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
15630 (pdp11_modes_tieable_p): New function.
15631 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
15632 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
15633 (rs6000_modes_tieable_p): New function.
15634 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
15635 * config/powerpcspe/powerpcspe.md: Update comment.
15636 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
15637 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
15638 (TARGET_MODES_TIEABLE_P): Redefine.
15639 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
15640 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
15641 (rl78_modes_tieable_p): New function.
15642 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
15643 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
15644 (rs6000_modes_tieable_p): New function.
15645 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
15646 * config/rs6000/rs6000.md: Update comment.
15647 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
15648 * config/rx/rx.c (rx_modes_tieable_p): New function.
15649 (TARGET_MODES_TIEABLE_P): Redefine.
15650 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
15651 * config/s390/s390.c (s390_modes_tieable_p): New function.
15652 (TARGET_MODES_TIEABLE_P): Redefine.
15653 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
15654 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
15655 (sh_modes_tieable_p): New function.
15656 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
15657 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
15658 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
15659 (sparc_modes_tieable_p): Make static.
15660 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
15661 * config/spu/spu.c (spu_modes_tieable_p): New function.
15662 (TARGET_MODES_TIEABLE_P): Redefine.
15663 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
15664 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
15665 (TARGET_MODES_TIEABLE_P): Redefine.
15666 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
15667 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
15668 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
15669 * config/v850/v850.c (v850_modes_tieable_p): New function.
15670 (TARGET_MODES_TIEABLE_P): Redefine.
15671 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
15672 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
15673 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
15674 (visium_modes_tieable_p): New function.
15675 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
15676 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
15677 (xtensa_modes_tieable_p): New function.
15678 * system.h (MODES_TIEABLE_P): Poison.
15680 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15681 Alan Hayward <alan.hayward@arm.com>
15682 David Sherwood <david.sherwood@arm.com>
15684 * target.def (hard_regno_mode_ok): New hook.
15685 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
15686 (TARGET_HARD_REGNO_MODE_OK): ...this.
15687 * doc/tm.texi.in: Regenerate.
15688 * hooks.h (hook_bool_uint_mode_true): Declare.
15689 * hooks.c (hook_bool_uint_mode_true): New function.
15690 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
15691 HARD_REGNO_MODE_OK.
15692 * genpreds.c (write_insn_preds_c): Add an include of target.h.
15693 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
15694 instead of HARD_REGNO_MODE_OK.
15695 * caller-save.c: Include target.h.
15696 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
15697 HARD_REGNO_MODE_OK.
15698 * combine.c (can_combine_p): Likewise.
15699 (combinable_i3pat): Likewise.
15700 (can_change_dest_mode): Likewise.
15701 * expr.c (init_expr_target): Likewise.
15702 (convert_move): Likewise.
15703 (convert_modes): Likewise.
15704 * ira.c (setup_prohibited_class_mode_regs): Likewise.
15705 (setup_prohibited_mode_move_regs): Likewise.
15706 * ira.h (target_ira): Likewise.
15707 * lra-assigns.c (find_hard_regno_for_1): Likewise.
15708 * lra-constraints.c (process_alt_operands): Likewise.
15709 (split_reg): Likewise.
15710 * recog.c (peep2_find_free_register): Likewise.
15711 * ree.c (combine_reaching_defs): Likewise.
15712 * regcprop.c (maybe_mode_change): Likewise.
15713 * reginfo.c (init_reg_sets_1): Likewise.
15714 (choose_hard_reg_mode): Likewise.
15715 (simplifiable_subregs): Likewise.
15716 * regrename.c (check_new_reg_p): Likewise.
15717 * reload.c (find_valid_class): Likewise.
15718 (find_valid_class_1): Likewise.
15719 (reload_inner_reg_of_subreg): Likewise.
15720 (push_reload): Likewise.
15721 (combine_reloads): Likewise.
15722 (find_dummy_reload): Likewise.
15723 (find_reloads): Likewise.
15724 * reload1.c (find_reg): Likewise.
15725 (set_reload_reg): Likewise.
15726 (allocate_reload_reg): Likewise.
15727 (choose_reload_regs): Likewise.
15728 (reload_adjust_reg_for_temp): Likewise.
15729 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
15730 (simplify_subreg_regno): Likewise.
15731 * sel-sched.c (init_regs_for_mode): Likewise.
15732 * varasm.c (make_decl_rtl): Likewise.
15733 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
15734 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
15735 HARD_REGNO_MODE_OK.
15736 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
15737 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
15738 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15739 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
15740 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
15741 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15742 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
15743 (arc_mode_class): Delete.
15744 (HARD_REGNO_MODE_OK): Delete.
15745 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15746 (arc_hard_regno_mode_ok): Rename old array to...
15747 (arc_hard_regno_mode_ok_modes): ...this.
15748 (arc_conditional_register_usage): Update accordingly.
15749 (arc_mode_class): Make static.
15750 (arc_hard_regno_mode_ok): New function.
15751 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
15752 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
15753 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15754 (arm_hard_regno_mode_ok): Make static.
15755 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
15756 HARD_REGNO_MODE_OK.
15757 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
15758 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
15759 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
15761 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15762 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
15763 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
15764 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
15765 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
15766 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15767 * config/bfin/predicates.md (valid_reg_operand): Use
15768 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
15769 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
15770 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
15771 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15772 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
15773 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
15774 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15775 (cr16_hard_regno_mode_ok): Make static and return a bool.
15776 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
15777 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15778 (cris_hard_regno_mode_ok): New function.
15779 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
15780 (epiphany_mode_class): Delete.
15781 (HARD_REGNO_MODE_OK): Delete.
15782 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
15783 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15784 (hard_regno_mode_ok): Rename to...
15785 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
15786 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
15787 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
15788 HARD_REGNO_MODE_OK.
15789 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
15790 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
15791 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15792 (frv_hard_regno_mode_ok): Make static and return a bool.
15793 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
15794 HARD_REGNO_MODE_OK.
15795 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
15796 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
15797 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
15798 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
15800 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15801 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
15802 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
15803 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
15805 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15806 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
15807 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15808 (ia64_hard_regno_mode_ok): New function.
15809 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
15810 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15811 (iq2000_hard_regno_mode_ok): New function.
15812 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
15813 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15814 (lm32_hard_regno_mode_ok): New function.
15815 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
15816 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
15817 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
15818 instead of HARD_REGNO_MODE_OK.
15819 (m32c_hard_regno_ok): Rename to...
15820 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
15821 (m32c_cannot_change_mode_class): Update accordingly.
15822 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15823 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
15824 (m32r_mode_class): Delete.
15825 (HARD_REGNO_MODE_OK): Delete.
15826 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15827 (m32r_hard_regno_mode_ok): Rename to...
15828 (m32r_hard_regno_modes): ...this.
15829 (m32r_mode_class): Make static.
15830 (m32r_hard_regno_mode_ok): New function.
15831 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
15832 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
15833 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15834 (m68k_hard_regno_mode_ok): Make static.
15835 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
15836 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15837 (mcore_hard_regno_mode_ok): New function.
15838 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
15839 (HARD_REGNO_MODE_OK): Delete.
15840 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
15842 (microblaze_hard_regno_mode_ok_p): ...this and make static.
15843 (microblaze_hard_regno_mode_ok): New function.
15844 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15845 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
15846 (mips_hard_regno_mode_ok): Delete.
15847 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
15848 (mips_hard_regno_mode_ok_p): ...this and make static.
15849 (mips_hard_regno_mode_ok_p): Rename to...
15850 (mips_hard_regno_mode_ok_uncached): ...this.
15851 (mips_hard_regno_mode_ok): New function.
15852 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
15853 of HARD_REGNO_MODE_OK.
15854 (mips_option_override): Update after above name changes.
15855 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15856 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
15857 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
15858 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
15859 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
15860 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15861 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
15862 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
15863 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
15864 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15865 (msp430_hard_regno_mode_ok): Make static and return a bool.
15866 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
15867 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
15868 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
15870 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15871 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
15872 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
15873 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
15874 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
15875 (PA_HARD_REGNO_MODE_OK): ...this
15876 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
15877 (PA_HARD_REGNO_MODE_OK): ...this.
15878 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15879 (pa_hard_regno_mode_ok): New function.
15880 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
15881 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15882 (pdp11_hard_regno_mode_ok): New function.
15883 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
15884 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
15886 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
15888 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15889 (rs6000_hard_regno_mode_ok): Rename to...
15890 (rs6000_hard_regno_mode_ok_uncached): ...this.
15891 (rs6000_init_hard_regno_mode_ok): Update accordingly.
15892 (rs6000_hard_regno_mode_ok): New function.
15893 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
15894 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
15895 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
15896 (riscv_hard_regno_mode_ok): ...this and make static.
15897 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15898 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
15899 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
15900 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15901 (rl78_hard_regno_mode_ok): Make static and return bool.
15902 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
15903 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
15905 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
15906 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15907 (rs6000_hard_regno_mode_ok): Rename to...
15908 (rs6000_hard_regno_mode_ok_uncached): ...this.
15909 (rs6000_init_hard_regno_mode_ok): Update accordingly.
15910 (rs6000_hard_regno_mode_ok): New function.
15911 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
15912 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
15913 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15914 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
15915 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
15916 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
15917 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15918 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
15919 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
15920 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15921 (sh_hard_regno_mode_ok): Make static.
15922 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
15923 instead of HARD_REGNO_MODE_OK.
15924 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
15925 (sparc_mode_class): Delete.
15926 (HARD_REGNO_MODE_OK): Delete.
15927 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15928 (hard_regno_mode_classes): Make static.
15929 (sparc_mode_class): Likewise.
15930 (sparc_hard_regno_mode_ok): New function.
15931 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
15932 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
15933 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
15935 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15936 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
15937 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
15938 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
15939 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
15940 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15941 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
15942 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
15943 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15944 (visium_hard_regno_mode_ok): New function.
15945 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
15946 instead of HARD_REGNO_MODE_OK.
15947 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
15948 (HARD_REGNO_MODE_OK): Delete.
15949 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
15950 (xtensa_hard_regno_mode_ok_p): ...this and make static.
15951 (xtensa_option_override): Update accordingly.
15952 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15953 (xtensa_hard_regno_mode_ok): New function.
15954 * system.h (HARD_REGNO_MODE_OK): Poison.
15956 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15957 Alan Hayward <alan.hayward@arm.com>
15958 David Sherwood <david.sherwood@arm.com>
15960 * target.def (hard_regno_call_part_clobbered): New hook.
15961 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
15962 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
15963 * doc/tm.texi: Regenerate.
15964 * hooks.h (hook_bool_uint_mode_false): Declare.
15965 * hooks.c (hook_bool_uint_mode_false): New function.
15966 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15967 * cselib.c (cselib_process_insn): Use
15968 targetm.hard_regno_call_part_clobbered instead of
15969 HARD_REGNO_CALL_PART_CLOBBERED.
15970 * ira-conflicts.c (ira_build_conflicts): Likewise.
15971 * ira-costs.c (ira_tune_allocno_costs): Likewise.
15972 * lra-constraints.c (need_for_call_save_p): Likewise.
15973 * lra-lives.c: Include target.h.
15974 (check_pseudos_live_through_calls): Use
15975 targetm.hard_regno_call_part_clobbered instead of
15976 HARD_REGNO_CALL_PART_CLOBBERED.
15977 * regcprop.c: Include target.h.
15978 (copyprop_hardreg_forward_1): Use
15979 targetm.hard_regno_call_part_clobbered instead of
15980 HARD_REGNO_CALL_PART_CLOBBERED.
15981 * reginfo.c (choose_hard_reg_mode): Likewise.
15982 * regrename.c (check_new_reg_p): Likewise.
15983 * reload.c (find_equiv_reg): Likewise.
15984 * reload1.c (emit_reload_insns): Likewise.
15985 * sched-deps.c (deps_analyze_insn): Likewise.
15986 * sel-sched.c (init_regs_for_mode): Likewise.
15987 (mark_unavailable_hard_regs): Likewise.
15988 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
15989 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15990 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
15992 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15993 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15994 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
15996 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
15998 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15999 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16000 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
16002 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16003 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16004 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
16006 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16007 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
16009 * config/powerpcspe/powerpcspe.c
16010 (rs6000_hard_regno_call_part_clobbered): New function.
16011 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16012 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16013 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
16015 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16016 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16017 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
16019 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
16020 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
16021 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
16023 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16024 Alan Hayward <alan.hayward@arm.com>
16025 David Sherwood <david.sherwood@arm.com>
16027 * rtl.h (subreg_memory_offset): Declare.
16028 * emit-rtl.c (subreg_memory_offset): New function.
16029 * expmed.c (store_bit_field_1): Use it.
16030 * expr.c (undefined_operand_subword_p): Likewise.
16031 * simplify-rtx.c (simplify_subreg): Likewise.
16033 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
16035 PR rtl-optimization/57448
16038 * optabs.c (expand_atomic_load): Place compiler memory barriers if
16039 using atomic_load pattern.
16040 (expand_atomic_store): Likewise.
16042 2017-09-04 Jakub Jelinek <jakub@redhat.com>
16045 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
16046 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
16047 handling. Use replace_call_with_value with NULL instead of
16048 gsi_replace, unlink_stmt_vdef and release_defs.
16050 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
16053 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
16055 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
16058 * rtl.h (emit_library_call_value_1): Declare.
16059 (emit_library_call): Replace declaration with a series of overloads.
16060 Remove the parameter count argument.
16061 (emit_library_call_value): Likewise.
16062 * calls.c (emit_library_call_value_1): Make global. Replace varargs
16063 with an "rtx_mode_t *".
16064 (emit_library_call_value): Delete.
16065 (emit_library_call): Likewise.
16066 * asan.c (asan_emit_stack_protection): Update calls accordingly.
16067 (asan_emit_allocas_unpoison): Likewise.
16068 * builtins.c (expand_builtin_powi): Likewise.
16069 (expand_asan_emit_allocas_unpoison): Likewise.
16070 * cfgexpand.c (expand_main_function): Likewise.
16071 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
16072 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
16073 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
16074 * config/arm/arm.c (arm_trampoline_init): Likewise.
16075 (arm_call_tls_get_addr): Likewise.
16076 (arm_expand_divmod_libfunc): Likewise.
16077 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
16078 (smulsi3_highpart): Likewise.
16079 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
16080 (c6x_expand_compare): Likewise.
16081 (c6x_expand_movmem): Likewise.
16082 * config/frv/frv.c (frv_trampoline_init): Likewise.
16083 * config/i386/i386.c (ix86_trampoline_init): Likewise.
16084 (ix86_expand_divmod_libfunc): Likewise.
16085 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
16086 (ia64_expand_compare): Likewise.
16087 (ia64_profile_hook): Likewise.
16088 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
16089 (nonlocal_goto): Likewise.
16090 (restore_stack_nonlocal): Likewise.
16091 * config/m32r/m32r.c (block_move_call): Likewise.
16092 (m32r_trampoline_init): Likewise.
16093 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
16094 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
16095 (m68k_call_m68k_read_tp): Likewise.
16096 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
16097 (microblaze_expand_divide): Likewise.
16098 * config/mips/mips.h (mips_args): Likewise.
16099 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
16100 (MIPS_ICACHE_SYNC): Likewise.
16101 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
16102 (nios2_trampoline_init): Likewise.
16103 * config/pa/pa.c (hppa_tls_call): Likewise.
16104 (pa_trampoline_init): Likewise.
16105 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
16106 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
16107 (expand_strn_compare): Likewise.
16108 (rs6000_generate_compare): Likewise.
16109 (rs6000_expand_float128_convert): Likewise.
16110 (output_profile_hook): Likewise.
16111 (rs6000_trampoline_init): Likewise.
16112 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
16113 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
16114 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
16115 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
16116 (rs6000_generate_compare): Likewise.
16117 (rs6000_expand_float128_convert): Likewise.
16118 (output_profile_hook): Likewise.
16119 (rs6000_trampoline_init): Likewise.
16120 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
16121 * config/sh/sh.c (sh_trampoline_init): Likewise.
16122 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
16123 (sparc_emit_float_lib_cmp): Likewise.
16124 (sparc32_initialize_trampoline): Likewise.
16125 (sparc64_initialize_trampoline): Likewise.
16126 (sparc_profile_hook): Likewise.
16127 * config/spu/spu.c (ea_load_store): Likewise.
16128 * config/spu/spu.md (floatunssidf2): Likewise.
16129 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
16130 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
16131 * config/visium/visium.c (expand_block_move_4): Likewise.
16132 (expand_block_move_2): Likewise.
16133 (expand_block_move_1): Likewise.
16134 (expand_block_set_4): Likewise.
16135 (expand_block_set_2): Likewise.
16136 (expand_block_set_1): Likewise.
16137 (visium_trampoline_init): Likewise.
16138 (visium_profile_hook): Likewise.
16139 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
16140 (xtensa_setup_frame_addresses): Likewise.
16141 (xtensa_trampoline_init): Likewise.
16142 * except.c (sjlj_emit_function_enter): Likewise.
16143 (sjlj_emit_function_exit): Likewise.
16144 * explow.c (allocate_dynamic_stack_space): Likewise.
16145 (probe_stack_range): Likewise.
16146 * expr.c (convert_mode_scalar): Likewise.
16147 * optabs.c (expand_binop): Likewise.
16148 (expand_twoval_binop_libfunc): Likewise.
16149 (expand_unop): Likewise.
16150 (prepare_cmp_insn): Likewise.
16151 (prepare_float_lib_cmp): Likewise.
16152 (expand_float): Likewise.
16153 (expand_fix): Likewise.
16154 (expand_fixed_convert): Likewise.
16155 (maybe_emit_sync_lock_test_and_set): Likewise.
16156 (expand_atomic_compare_and_swap): Likewise.
16157 (expand_mem_thread_fence): Likewise.
16158 (expand_atomic_fetch_op): Likewise.
16160 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
16162 * doc/generic.texi (OpenACC): Adjust URL.
16163 * doc/invoke.texi (C Dialect Options): Ditto.
16165 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
16167 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
16168 predicate for operand 1. Add (m,<S>) constraint.
16169 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
16170 Prevent memory operand 1 with register operand 2.
16172 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
16174 PR rtl-optimization/82024
16175 * combine.c (try_combine): If the combination result is a PARALLEL,
16176 and we only need to retain the SET in there that would be placed
16177 at I2, check that we can place that at I3 instead, before doing so.
16179 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16182 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
16184 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
16185 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
16188 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
16189 Jeff Law <law@redhat.com>
16191 * varasm.c (bss_initializer_p): Do not put constants into .bss
16192 (categorize_decl_for_section): Handle bss_initializer_p returning
16193 false when DECL_INITIAL is NULL.
16195 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16198 * config/s390/s390.c (s390_can_inline_p): New function.
16200 2017-09-01 Jeff Law <law@redhat.com>
16202 PR tree-optimization/82052
16203 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
16204 Always initialize the returned slot after a hash table miss
16205 when INSERT is true.
16207 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
16209 * config/s390/s390.md (mem_signal_fence): Remove.
16210 * doc/md.texi (mem_signal_fence): Remove.
16211 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
16213 * target-insns.def (mem_signal_fence): Remove.
16215 2017-09-01 Jakub Jelinek <jakub@redhat.com>
16218 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
16221 * asan.c (create_odr_indicator): Strip name encoding from assembler
16222 name before appending it after __odr_asan_.
16224 2017-09-01 Martin Liska <mliska@suse.cz>
16226 PR tree-optimization/82059
16227 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
16228 frequency only when an edge is redirected.
16230 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16232 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
16233 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
16234 (arc_conditional_register_usage): Remove ARC600 lp_count
16236 (arc_file_start): Emit Tag_ARC_CPU_variation.
16237 (arc_can_use_doloop_p): New conditions to use ZOLs.
16238 (hwloop_fail): New function.
16239 (hwloop_optimize): Likewise.
16240 (hwloop_pattern_reg): Likewise.
16241 (arc_doloop_hooks): New struct, to be used with reorg_loops.
16242 (arc_reorg_loops): New function, calls reorg_loops.
16243 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
16244 (arc600_corereg_hazard): Remove ZOL checking, case handled by
16246 (arc_loop_hazard): Remove function, functionality moved into
16248 (arc_hazard): Remove arc_loop_hazard call.
16249 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
16250 into hwloop_optimize.
16251 (arc_label_align): Remove ZOL handling.
16252 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
16253 * config/arc/arc.md (doloop_begin): Remove pattern.
16254 (doloop_begin_i): Likewise.
16255 (doloop_end_i): Likewise.
16256 (doloop_fallback): Likewise.
16257 (doloop_fallback_m): Likewise.
16258 (doloop_end): Reimplement expand.
16259 (arc_lp): New pattern for LP instruction.
16260 (loop_end): New pattern.
16261 (loop_fail): Likewise.
16262 (decrement_and_branch_until_zero): Likewise.
16263 * config/arc/arc.opt (mlpc-width): New option.
16264 * doc/invoke.texi (mlpc-width): Document option.
16266 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16268 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
16269 (arc_ccfsm_advance): Fix checking for delay slots.
16270 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
16272 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16274 * config/arc/arc.md (movqi_insn): Add stores to save constant long
16276 (movhi_insn): Update store instruction constraint which are saving
16277 6-bit short immediates.
16278 (movsi_insn): Consider also short scaled load operations.
16279 (zero_extendhisi2_i): Use Usd constraint instead of T.
16280 (extendhisi2_i): Add q constraint.
16281 (arc_clzsi2): Add type and length attributes.
16282 (arc_ctzsi2): Likewise.
16283 * config/arc/constraints.md (Usc): Update constraint, the
16284 assembler can parse two relocations for a single instruction.
16286 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
16288 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
16289 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
16291 2017-08-31 Olivier Hainque <hainque@adacore.com>
16293 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
16294 match as powerpc-wrs-vxworks*.
16296 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
16298 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
16299 register constraint for by-element operand.
16300 (aarch64_mls_elt_merge<mode>): Likewise.
16302 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16304 * config/arc/arc.c (arc_can_follow_jump): Check for short
16307 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16309 * config.gcc: Use g.opt for arc.
16310 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
16311 functionality moved to ...
16312 (legitimate_scaled_address_p): New function, ...here.
16313 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
16314 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
16315 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
16317 (arc_override_options): Handle G option.
16318 (arc_output_pic_addr_const): Correct function definition.
16319 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
16320 (arc_decl_anon_ns_mem_p): Delete.
16321 (arc_in_small_data_p): Overhaul this function to take into
16322 consideration the value given via G option.
16323 (arc_rewrite_small_data_1): Renamed and corrected old
16324 arc_rewrite_small_data function.
16325 (arc_rewrite_small_data): New function.
16326 (small_data_pattern): Don't use pic_offset_table_rtx.
16327 * config/arc/arc.h (CC1_SPEC): Recognize G option.
16328 * config/arc/simdext.md (movmisalignv2hi): Use
16329 prepare_move_operands function.
16331 (movmisalign*): Likewise.
16332 * doc/invoke.texi (ARC options): Document -G option.
16334 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
16336 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
16338 * config/arc/arc.c (arc_print_operand): Output scalled address for
16339 sdata whenever is possible.
16340 (arc_in_small_data_p): Allow sdata for 64bit datum when double
16341 load/stores are available.
16342 (compact_sda_memory_operand): Check for the alignment required by
16343 code density instructions.
16344 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
16346 * config/arc/constraints.md (Usd): Update constraint.
16347 (Us0): New constraint.
16348 (Usc): Update constraint.
16350 2017-08-31 Richard Biener <rguenther@suse.de>
16352 PR middle-end/82054
16353 * dwarf2out.c (dwarf2out_early_global_decl): Process each
16354 function only once.
16356 2017-08-31 Tamar Christina <tamar.christina@arm.com>
16358 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
16359 Resize type_signature.
16361 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16362 Alan Hayward <alan.hayward@arm.com>
16363 David Sherwood <david.sherwood@arm.com>
16365 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
16366 subregs whose inner modes can be stored in GPRs.
16367 (aarch64_classify_index): Likewise.
16369 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16370 Alan Hayward <alan.hayward@arm.com>
16371 David Sherwood <david.sherwood@arm.com>
16373 * config/aarch64/iterators.md (V_cmp_result): Rename to...
16374 (V_INT_EQUIV): ...this.
16375 (v_cmp_result): Rename to...
16376 (v_int_equiv): ...this.
16377 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
16378 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
16379 (copysign<mode>3): Likewise.
16380 (aarch64_simd_bsl<mode>_internal): Likewise.
16381 (aarch64_simd_bsl<mode>): Likewise.
16382 (vec_cmp<mode><mode>): Likewise.
16383 (vcond<mode><mode>): Likewise.
16384 (vcond<v_cmp_mixed><mode>): Likewise.
16385 (vcondu<mode><v_cmp_mixed>): Likewise.
16386 (aarch64_cm<optab><mode>): Likewise.
16387 (aarch64_cmtst<mode>): Likewise.
16388 (aarch64_fac<optab><mode>): Likewise.
16389 (vec_perm_const<mode>): Likewise.
16390 (vcond_mask_<mode><v_cmp_result>): Rename to...
16391 (vcond_mask_<mode><v_int_equiv>): ...this.
16392 (vec_cmp<mode><v_cmp_result>): Rename to...
16393 (vec_cmp<mode><v_int_equiv>): ...this.
16395 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
16396 Alan Hayward <alan.hayward@arm.com>
16397 David Sherwood <david.sherwood@arm.com>
16399 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
16401 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
16402 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
16403 (UNSPEC_LD4_DREG): New unspecs.
16404 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
16405 (aarch64_ld2<mode>_dreg_be): Replace with...
16406 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
16408 (aarch64_ld3<mode>_dreg_le)
16409 (aarch64_ld3<mode>_dreg_be): Replace with...
16410 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
16412 (aarch64_ld4<mode>_dreg_le)
16413 (aarch64_ld4<mode>_dreg_be): Replace with...
16414 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
16417 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16419 PR tree-optimization/81987
16420 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
16421 insert an initializer in a location not dominated by the stride
16424 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16426 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
16427 on the entire header of the finally block in the fallthru case.
16429 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
16431 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
16433 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
16435 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
16436 rs6000_emit_move_from_cr and call renamed function.
16437 (rs6000_emit_prologue): Call renamed functions.
16438 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
16439 movesi_from_cr, remove volatile CRs.
16441 2017-08-30 Jon Beniston <jon@beniston.com>
16442 Richard Biener <rguenther@suse.de>
16444 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
16445 of VECTOR_MODE_P check.
16446 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
16447 element vector types.
16449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16451 * df.h (df_read_modify_subreg_p): Remove in favor of...
16452 * rtl.h (read_modify_subreg_p): ...this new function. Take a
16453 const_rtx instead of an rtx.
16454 * cprop.c (local_cprop_find_used_regs): Update accordingly.
16455 * df-problems.c (df_word_lr_mark_ref): Likewise.
16456 * ira-lives.c (mark_pseudo_reg_live): Likewise.
16457 (mark_pseudo_reg_dead): Likewise.
16458 (mark_ref_dead): Likewise.
16459 * reginfo.c (init_subregs_of_mode): Likewise.
16460 * sched-deps.c (sched_analyze_1): Likewise.
16461 * df-scan.c (df_def_record_1): Likewise.
16462 (df_uses_record): Likewise.
16463 (df_read_modify_subreg_p): Remove in favor of...
16464 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
16465 const_rtx instead of an rtx.
16467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16468 Alan Hayward <alan.hayward@arm.com>
16469 David Sherwood <david.sherwood@arm.com>
16471 * rtl.h (partial_subreg_p): New function.
16472 * caller-save.c (save_call_clobbered_regs): Use it.
16473 * calls.c (expand_call): Likewise.
16474 * combine.c (combinable_i3pat): Likewise.
16475 (simplify_set): Likewise.
16476 (make_extraction): Likewise.
16477 (make_compound_operation_int): Likewise.
16478 (gen_lowpart_or_truncate): Likewise.
16479 (force_to_mode): Likewise.
16480 (make_field_assignment): Likewise.
16481 (reg_truncated_to_mode): Likewise.
16482 (record_truncated_value): Likewise.
16483 (move_deaths): Likewise.
16484 * cse.c (record_jump_cond): Likewise.
16485 (cse_insn): Likewise.
16486 * cselib.c (cselib_lookup_1): Likewise.
16487 * expmed.c (extract_bit_field_using_extv): Likewise.
16488 * function.c (assign_parm_setup_reg): Likewise.
16489 * ifcvt.c (noce_convert_multiple_sets): Likewise.
16490 * ira-build.c (create_insn_allocnos): Likewise.
16491 * lra-coalesce.c (merge_pseudos): Likewise.
16492 * lra-constraints.c (match_reload): Likewise.
16493 (simplify_operand_subreg): Likewise.
16494 (curr_insn_transform): Likewise.
16495 * lra-lives.c (process_bb_lives): Likewise.
16496 * lra.c (new_insn_reg): Likewise.
16497 (lra_substitute_pseudo): Likewise.
16498 * regcprop.c (mode_change_ok): Likewise.
16499 (maybe_mode_change): Likewise.
16500 (copyprop_hardreg_forward_1): Likewise.
16501 * reload.c (push_reload): Likewise.
16502 (find_reloads): Likewise.
16503 (find_reloads_subreg_address): Likewise.
16504 * reload1.c (alter_reg): Likewise.
16505 (eliminate_regs_1): Likewise.
16506 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16508 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
16510 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
16511 back to if statements, including unpack.
16513 2017-08-30 Martin Liska <mliska@suse.cz>
16515 PR inline-asm/82001
16516 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
16518 (func_checker::compare_asm_inputs_outputs): ... this function.
16519 (func_checker::compare_gimple_asm): Use the function to compare
16520 also ASM constrains.
16521 * ipa-icf-gimple.h: Rename the function.
16523 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16524 Alan Hayward <alan.hayward@arm.com>
16525 David Sherwood <david.sherwood@arm.com>
16527 * coretypes.h (complex_mode): New type.
16528 * gdbhooks.py (build_pretty_printer): Handle it.
16529 * machmode.h (complex_mode): New class.
16530 (complex_mode::includes_p): New function.
16531 (is_complex_int_mode): Likewise.
16532 (is_complex_float_mode): Likewise.
16533 * genmodes.c (get_mode_class): Handle complex mode classes.
16534 * function.c (expand_function_end): Use is_complex_int_mode.
16536 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16537 Alan Hayward <alan.hayward@arm.com>
16538 David Sherwood <david.sherwood@arm.com>
16540 * coretypes.h (scalar_mode_pod): New typedef.
16541 * gdbhooks.py (build_pretty_printer): Handle it.
16542 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
16543 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
16544 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
16545 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
16546 as_a <scalar_mode>.
16548 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16549 Alan Hayward <alan.hayward@arm.com>
16550 David Sherwood <david.sherwood@arm.com>
16552 * machmode.h (mode_for_vector): Take a scalar_mode instead
16554 * stor-layout.c (mode_for_vector): Likewise.
16555 * explow.c (promote_mode): Use as_a <scalar_mode>.
16556 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
16558 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16559 Alan Hayward <alan.hayward@arm.com>
16560 David Sherwood <david.sherwood@arm.com>
16562 * target.def (preferred_simd_mode): Take a scalar_mode
16563 instead of a machine_mode.
16564 * targhooks.h (default_preferred_simd_mode): Likewise.
16565 * targhooks.c (default_preferred_simd_mode): Likewise.
16566 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
16567 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
16568 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
16569 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
16570 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
16571 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
16572 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
16573 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
16575 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
16576 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
16577 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
16578 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
16579 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
16580 * doc/tm.texi: Regenerate.
16581 * optabs-query.c (can_vec_mask_load_store_p): Return false for
16584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16585 Alan Hayward <alan.hayward@arm.com>
16586 David Sherwood <david.sherwood@arm.com>
16588 * target.def (scalar_mode_supported_p): Take a scalar_mode
16589 instead of a machine_mode.
16590 * targhooks.h (default_scalar_mode_supported_p): Likewise.
16591 * targhooks.c (default_scalar_mode_supported_p): Likewise.
16592 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
16593 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
16594 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
16595 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
16596 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
16597 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
16598 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
16599 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
16600 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
16601 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
16602 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
16603 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
16605 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
16606 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
16607 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
16608 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
16609 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
16611 * doc/tm.texi: Regenerate.
16613 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16614 Alan Hayward <alan.hayward@arm.com>
16615 David Sherwood <david.sherwood@arm.com>
16617 * coretypes.h (opt_scalar_mode): New typedef.
16618 * gdbhooks.py (build_pretty_printers): Handle it.
16619 * machmode.h (mode_iterator::get_2xwider): Add overload for
16621 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
16623 * expr.c (convert_mode_scalar): Likewise.
16624 * omp-low.c (omp_clause_aligned_alignment): Likewise.
16625 * optabs.c (expand_float): Likewise.
16626 (expand_fix): Likewise.
16627 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16630 Alan Hayward <alan.hayward@arm.com>
16631 David Sherwood <david.sherwood@arm.com>
16633 * optabs.c (expand_float): Explicitly check for scalars before
16634 using a branching expansion.
16635 (expand_fix): Likewise.
16637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16638 Alan Hayward <alan.hayward@arm.com>
16639 David Sherwood <david.sherwood@arm.com>
16641 * expr.c (convert_mode): Split scalar handling out into...
16642 (convert_mode_scalar): ...this new function. Treat the modes
16645 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16646 Alan Hayward <alan.hayward@arm.com>
16647 David Sherwood <david.sherwood@arm.com>
16649 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
16651 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
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 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
16658 rather than a machine_mode.
16659 (fixed_from_string): Likewise.
16660 (fixed_convert): Likewise.
16661 (fixed_convert_from_int): Likewise.
16662 (fixed_convert_from_real): Likewise.
16663 (real_convert_from_fixed): Likewise.
16664 * fixed-value.c (fixed_from_double_int): Likewise.
16665 (fixed_from_string): Likewise.
16666 (fixed_convert): Likewise.
16667 (fixed_convert_from_int): Likewise.
16668 (fixed_convert_from_real): Likewise.
16669 (real_convert_from_fixed): Likewise.
16670 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
16672 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16673 Alan Hayward <alan.hayward@arm.com>
16674 David Sherwood <david.sherwood@arm.com>
16676 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
16677 of separate mode class checks. Do not allow vector modes here.
16678 (immed_wide_int_const): Use as_a <scalar_mode>.
16679 * explow.c (trunc_int_for_mode): Likewise.
16680 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
16681 (wi::shwi): Likewise.
16682 (wi::min_value): Likewise.
16683 (wi::max_value): Likewise.
16684 * dwarf2out.c (loc_descriptor): Likewise.
16685 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
16686 for CONST_WIDE_INT.
16688 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16689 Alan Hayward <alan.hayward@arm.com>
16690 David Sherwood <david.sherwood@arm.com>
16692 * tree.h (SCALAR_TYPE_MODE): New macro.
16693 * expr.c (expand_expr_addr_expr_1): Use it.
16694 (expand_expr_real_2): Likewise.
16695 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
16696 (fold_convert_const_fixed_from_int): Likewise.
16697 (fold_convert_const_fixed_from_real): Likewise.
16698 (native_encode_fixed): Likewise
16699 (native_encode_complex): Likewise
16700 (native_encode_vector): Likewise.
16701 (native_interpret_fixed): Likewise.
16702 (native_interpret_real): Likewise.
16703 (native_interpret_complex): Likewise.
16704 (native_interpret_vector): Likewise.
16705 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
16706 (simd_clone_adjust_argument_types): Likewise.
16707 (simd_clone_init_simd_arrays): Likewise.
16708 (simd_clone_adjust): Likewise.
16709 * stor-layout.c (layout_type): Likewise.
16710 * tree.c (build_minus_one_cst): Likewise.
16711 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16712 * tree-inline.c (estimate_move_cost): Likewise.
16713 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
16714 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
16715 (vectorizable_reduction): Likewise.
16716 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
16717 (vect_recog_mixed_size_cond_pattern): Likewise.
16718 (check_bool_pattern): Likewise.
16719 (adjust_bool_pattern): Likewise.
16720 (search_type_for_mask_1): Likewise.
16721 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
16722 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
16723 (vectorizable_load): Likewise.
16724 (vectorizable_store): Likewise.
16725 * ubsan.c (ubsan_encode_value): Likewise.
16726 * varasm.c (output_constant): Likewise.
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 * coretypes.h (scalar_mode): New class.
16733 * machmode.h (scalar_mode): Likewise.
16734 (scalar_mode::includes_p): New function.
16735 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
16736 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
16737 * genmodes.c (get_mode_class): Handle remaining scalar modes.
16738 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
16739 * expmed.c (store_bit_field_1): Likewise.
16740 (extract_bit_field_1): Likewise.
16741 * expr.c (write_complex_part): Likewise.
16742 (read_complex_part): Likewise.
16743 (emit_move_complex_push): Likewise.
16744 (expand_expr_real_2): Likewise.
16745 * function.c (assign_parm_setup_reg): Likewise.
16746 (assign_parms_unsplit_complex): Likewise.
16747 * optabs.c (expand_binop): Likewise.
16748 * rtlanal.c (subreg_get_info): Likewise.
16749 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16750 * varasm.c (output_constant_pool_2): Likewise.
16752 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16753 Alan Hayward <alan.hayward@arm.com>
16754 David Sherwood <david.sherwood@arm.com>
16756 * expmed.c (extract_high_half): Use scalar_int_mode and remove
16758 (expmed_mult_highpart_optab): Likewise.
16759 (expmed_mult_highpart): Likewise.
16761 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16762 Alan Hayward <alan.hayward@arm.com>
16763 David Sherwood <david.sherwood@arm.com>
16765 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
16766 instead of a machine_mode.
16767 (builtin_memset_read_str): Likewise.
16768 * builtins.c (c_readstr): Likewise.
16769 (builtin_memcpy_read_str): Likewise.
16770 (builtin_strncpy_read_str): Likewise.
16771 (builtin_memset_read_str): Likewise.
16772 (builtin_memset_gen_str): Likewise.
16773 (expand_builtin_signbit): Use scalar_int_mode for local variables.
16774 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
16775 instead of a machine_mode.
16776 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
16778 (make_extraction): Likewise.
16779 (try_widen_shift_mode): Take and return scalar_int_modes instead
16781 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
16782 a scalar_int_mode instead of a machine_mode.
16783 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
16784 (avr_addr_space_pointer_mode): Likewise.
16785 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
16786 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
16787 (msp430_unwind_word_mode): Likewise.
16788 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
16789 (spu_addr_space_pointer_mode): Likewise.
16790 (spu_addr_space_address_mode): Likewise.
16791 (spu_libgcc_cmp_return_mode): Likewise.
16792 (spu_libgcc_shift_count_mode): Likewise.
16793 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
16794 (rl78_addr_space_pointer_mode): Likewise.
16795 (fl78_unwind_word_mode): Likewise.
16796 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
16798 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
16799 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
16800 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
16801 (mips_valid_pointer_mode): Likewise.
16802 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
16803 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
16804 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
16806 (ft32_addr_space_address_mode): Likewise.
16807 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
16808 scalar_int_mode instead of a machine_mode.
16809 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
16811 (m32c_addr_space_address_mode): Likewise.
16812 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
16813 (rs6000_eh_return_filter_mode): Likewise.
16814 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
16815 (rs6000_eh_return_filter_mode): Likewise.
16816 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
16817 (s390_libgcc_shift_count_mode): Likewise.
16818 (s390_unwind_word_mode): Likewise.
16819 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
16821 * target.def (mode_rep_extended): Likewise.
16822 (valid_pointer_mode): Likewise.
16823 (addr_space.valid_pointer_mode): Likewise.
16824 (eh_return_filter_mode): Return a scalar_int_mode rather than
16826 (libgcc_cmp_return_mode): Likewise.
16827 (libgcc_shift_count_mode): Likewise.
16828 (unwind_word_mode): Likewise.
16829 (addr_space.pointer_mode): Likewise.
16830 (addr_space.address_mode): Likewise.
16831 * doc/tm.texi: Regenerate.
16832 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
16834 (do_jump): Use scalar_int_mode for local variables.
16835 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
16836 rather than a machine_mode.
16837 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
16838 (scompare_loc_descriptor_wide): Likewise.
16839 (scompare_loc_descriptor_narrow): Likewise.
16840 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
16842 * except.c (sjlj_emit_dispatch_table): Likewise.
16843 (expand_builtin_eh_copy_values): Likewise.
16844 * explow.c (convert_memory_address_addr_space_1): Likewise.
16845 Take a scalar_int_mode rather than a machine_mode.
16846 (convert_memory_address_addr_space): Take a scalar_int_mode rather
16847 than a machine_mode.
16848 (memory_address_addr_space): Use scalar_int_mode for local variables.
16849 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
16850 rather than a machine_mode.
16851 * expmed.c (mask_rtx): Likewise.
16852 (init_expmed_one_conv): Likewise.
16853 (expand_mult_highpart_adjust): Likewise.
16854 (extract_high_half): Likewise.
16855 (expmed_mult_highpart_optab): Likewise.
16856 (expmed_mult_highpart): Likewise.
16857 (expand_smod_pow2): Likewise.
16858 (expand_sdiv_pow2): Likewise.
16859 (emit_store_flag_int): Likewise.
16860 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
16862 (extract_low_bits): Likewise.
16863 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
16865 * expr.c (pieces_addr::adjust): Likewise.
16866 (can_store_by_pieces): Likewise.
16867 (store_by_pieces): Likewise.
16868 (clear_by_pieces_1): Likewise.
16869 (expand_expr_addr_expr_1): Likewise.
16870 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
16871 (expand_expr_real_1): Likewise.
16872 (try_casesi): Likewise.
16873 * final.c (shorten_branches): Likewise.
16874 * fold-const.c (fold_convert_const_int_from_fixed): Change the
16875 type of "mode" to machine_mode.
16876 * internal-fn.c (expand_arith_overflow_result_store): Take a
16877 scalar_int_mode rather than a machine_mode.
16878 (expand_mul_overflow): Use scalar_int_mode for local variables.
16879 * loop-doloop.c (doloop_modify): Likewise.
16880 (doloop_optimize): Likewise.
16881 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
16882 than a machine_mode.
16883 (expand_doubleword_shift_condmove): Likewise.
16884 (expand_doubleword_shift): Likewise.
16885 (expand_doubleword_clz): Likewise.
16886 (expand_doubleword_popcount): Likewise.
16887 (expand_doubleword_parity): Likewise.
16888 (expand_absneg_bit): Use scalar_int_mode for local variables.
16889 (prepare_float_lib_cmp): Likewise.
16890 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
16891 rather than a machine_mode.
16892 (convert_memory_address_addr_space): Likewise.
16893 (get_mode_bounds): Likewise.
16894 (get_address_mode): Return a scalar_int_mode rather than a
16896 * rtlanal.c (get_address_mode): Likewise.
16897 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
16898 than a machine_mode.
16899 * targhooks.c (default_mode_rep_extended): Likewise.
16900 (default_valid_pointer_mode): Likewise.
16901 (default_addr_space_valid_pointer_mode): Likewise.
16902 (default_eh_return_filter_mode): Return a scalar_int_mode rather
16903 than a machine_mode.
16904 (default_libgcc_cmp_return_mode): Likewise.
16905 (default_libgcc_shift_count_mode): Likewise.
16906 (default_unwind_word_mode): Likewise.
16907 (default_addr_space_pointer_mode): Likewise.
16908 (default_addr_space_address_mode): Likewise.
16909 * targhooks.h (default_eh_return_filter_mode): Likewise.
16910 (default_libgcc_cmp_return_mode): Likewise.
16911 (default_libgcc_shift_count_mode): Likewise.
16912 (default_unwind_word_mode): Likewise.
16913 (default_addr_space_pointer_mode): Likewise.
16914 (default_addr_space_address_mode): Likewise.
16915 (default_mode_rep_extended): Take a scalar_int_mode rather than
16917 (default_valid_pointer_mode): Likewise.
16918 (default_addr_space_valid_pointer_mode): Likewise.
16919 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
16921 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
16922 rather than a machine_mode.
16923 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
16924 for local variables.
16925 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16926 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
16927 than a machine_mode.
16929 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16930 Alan Hayward <alan.hayward@arm.com>
16931 David Sherwood <david.sherwood@arm.com>
16933 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
16934 the mode argument to scalar_int_mode.
16935 (do_jump_by_parts_zero_rtx): Likewise.
16936 (do_jump_by_parts_equality_rtx): Likewise.
16937 (do_jump_by_parts_greater): Take a mode argument.
16938 (do_jump_by_parts_equality): Likewise.
16939 (do_jump_1): Update calls accordingly.
16941 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16942 Alan Hayward <alan.hayward@arm.com>
16943 David Sherwood <david.sherwood@arm.com>
16945 * is-a.h (safe_dyn_cast): New function.
16946 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
16947 (jump_table_for_label): Likewise.
16948 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
16949 instead of an rtx_insn *.
16950 (shorten_branches): Use dyn_cast instead of LABEL_P and
16951 JUMP_TABLE_DATA_P. Use jump_table_for_label and
16952 rtx_jump_table_data::get_data_mode.
16953 (final_scan_insn): Likewise.
16955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16956 Alan Hayward <alan.hayward@arm.com>
16957 David Sherwood <david.sherwood@arm.com>
16959 * combine.c (try_combine): Use is_a <scalar_int_mode> when
16960 trying to combine a full-register integer set with a subreg
16963 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16964 Alan Hayward <alan.hayward@arm.com>
16965 David Sherwood <david.sherwood@arm.com>
16967 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
16968 that is always either address_mode or pointer_mode.
16970 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16971 Alan Hayward <alan.hayward@arm.com>
16972 David Sherwood <david.sherwood@arm.com>
16974 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
16975 when the two are known to be equal.
16977 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16978 Alan Hayward <alan.hayward@arm.com>
16979 David Sherwood <david.sherwood@arm.com>
16981 * simplify-rtx.c (simplify_const_unary_operation): Use
16982 is_a <scalar_int_mode> instead of checking for a nonzero
16983 precision. Forcibly convert op_mode to a scalar_int_mode
16984 in that case. More clearly differentiate the operand and
16985 result modes and use the former when deciding what the value
16986 of a count-bits operation should be. Use is_int_mode instead
16987 of checking for a MODE_INT. Remove redundant check for whether
16988 this mode has a zero precision.
16990 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16991 Alan Hayward <alan.hayward@arm.com>
16992 David Sherwood <david.sherwood@arm.com>
16994 * optabs.c (widen_leading): Change the type of the mode argument
16995 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
16996 (widen_bswap): Likewise.
16997 (expand_parity): Likewise.
16998 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
16999 (expand_ffs): Likewise.
17000 (epand_unop): Check for scalar integer modes before calling the
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 (const_scalar_mask_from_tree): Add a mode argument.
17009 (expand_expr_real_1): Update call accordingly.
17011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17012 Alan Hayward <alan.hayward@arm.com>
17013 David Sherwood <david.sherwood@arm.com>
17015 * expmed.c (store_bit_field_using_insv): Add op0_mode and
17016 value_mode arguments. Use scalar_int_mode internally.
17017 (store_bit_field_1): Rename the new integer mode from imode
17018 to op0_mode and use it instead of GET_MODE (op0). Update calls
17019 to store_split_bit_field, store_bit_field_using_insv and
17020 store_fixed_bit_field.
17021 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
17022 Use scalar_int_mode internally. Use a bit count rather than a mode
17023 when calculating the largest bit size for get_best_mode.
17024 Update calls to store_split_bit_field and store_fixed_bit_field_1.
17025 (store_fixed_bit_field_1): Add mode and value_mode arguments.
17026 Remove assertion that OP0 has a scalar integer mode.
17027 (store_split_bit_field): Add op0_mode and value_mode arguments.
17028 Update calls to extract_fixed_bit_field.
17029 (extract_bit_field_using_extv): Add an op0_mode argument.
17030 Use scalar_int_mode internally.
17031 (extract_bit_field_1): Rename the new integer mode from imode to
17032 op0_mode and use it instead of GET_MODE (op0). Update calls to
17033 extract_split_bit_field, extract_bit_field_using_extv and
17034 extract_fixed_bit_field.
17035 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
17036 to extract_split_bit_field and extract_fixed_bit_field_1.
17037 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
17038 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
17039 on the target mode.
17040 (extract_split_bit_field): Add an op0_mode argument. Update call
17041 to extract_fixed_bit_field.
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 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
17048 * explow.c (hard_function_value): Likewise.
17049 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
17050 convert_to_mode call outside the loop.
17051 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
17052 for the mode iterator. Require the mode specified by max_pieces
17054 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
17056 (copy_blkmode_to_reg): Likewise.
17057 (set_storage_via_setmem): Likewise.
17058 * optabs.c (prepare_cmp_insn): Likewise.
17059 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17060 * stor-layout.c (finish_bitfield_representative): Likewise.
17062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17063 Alan Hayward <alan.hayward@arm.com>
17064 David Sherwood <david.sherwood@arm.com>
17066 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
17067 * expr.c (convert_move): Use them.
17068 (convert_modes): Likewise.
17069 (store_expr_with_bounds): Likewise.
17071 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17072 Alan Hayward <alan.hayward@arm.com>
17073 David Sherwood <david.sherwood@arm.com>
17075 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
17076 parameter for the mode of "x". Remove the "known_x", "known_mode"
17077 and "known_ret" arguments. Change the type of the mode argument
17078 to scalar_int_mode.
17079 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
17080 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
17081 (reg_num_sign_bit_copies_for_combine): Likewise.
17082 * rtlanal.c (nonzero_bits1): Likewise.
17083 (num_sign_bit_copies1): Likewise.
17084 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
17085 (reg_num_sign_bit_copies_general): Likewise.
17086 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
17087 (reg_nonzero_bits_general): Likewise.
17089 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17090 Alan Hayward <alan.hayward@arm.com>
17091 David Sherwood <david.sherwood@arm.com>
17093 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
17094 than in subroutines. Return 1 for non-integer modes.
17095 (cached_num_sign_bit_copies): Change the type of the mode parameter
17096 to scalar_int_mode.
17097 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
17098 classes. Handle CONST_INT_P first and then check whether X also
17099 has a scalar integer mode. Check the same thing for inner registers
17100 of a SUBREG and for values that are being extended or truncated.
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 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
17107 in subroutines. Return the mode mask for non-integer modes.
17108 (cached_nonzero_bits): Change the type of the mode parameter
17109 to scalar_int_mode.
17110 (nonzero_bits1): Likewise. Remove early exit for other mode
17111 classes. Handle CONST_INT_P first and then check whether X
17112 also has a scalar integer mode.
17114 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17115 Alan Hayward <alan.hayward@arm.com>
17116 David Sherwood <david.sherwood@arm.com>
17118 * expr.c (widest_int_mode_for_size): Make the comment match the code.
17119 Return a scalar_int_mode and assert that the size is greater than
17121 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
17122 (op_by_pieces_d::op_by_pieces_d): Likewise.
17123 (op_by_pieces_d::run): Likewise.
17124 (can_store_by_pieces): Likewise.
17126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17127 Alan Hayward <alan.hayward@arm.com>
17128 David Sherwood <david.sherwood@arm.com>
17130 * combine.c (extract_left_shift): Add a mode argument and update
17132 (make_compound_operation_int): Change the type of the mode parameter
17133 to scalar_int_mode and update the call to extract_left_shift.
17135 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17136 Alan Hayward <alan.hayward@arm.com>
17137 David Sherwood <david.sherwood@arm.com>
17139 * combine.c (simplify_and_const_int): Change the type of the mode
17140 parameter to scalar_int_mode.
17141 (simplify_and_const_int_1): Likewise. Update recursive call.
17143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17144 Alan Hayward <alan.hayward@arm.com>
17145 David Sherwood <david.sherwood@arm.com>
17147 * combine.c (simplify_compare_const): Check that the mode is a
17148 scalar_int_mode (rather than VOIDmode) before testing its
17150 (simplify_comparison): Move COMPARISON_P handling out of the
17151 loop and restrict the latter part of the loop to scalar_int_modes.
17152 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
17153 and when considering SUBREG_REGs. Use is_int_mode instead of
17154 checking GET_MODE_CLASS against MODE_INT.
17156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17157 Alan Hayward <alan.hayward@arm.com>
17158 David Sherwood <david.sherwood@arm.com>
17160 * combine.c (try_widen_shift_mode): Move check for equal modes to...
17161 (simplify_shift_const_1): ...here. Use scalar_int_mode for
17162 shift_unit_mode and for modes involved in scalar shifts.
17164 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17165 Alan Hayward <alan.hayward@arm.com>
17166 David Sherwood <david.sherwood@arm.com>
17168 * combine.c (force_int_to_mode): New function, split out from...
17169 (force_to_mode): ...here. Keep xmode up-to-date and use it
17170 instead of GET_MODE (x).
17172 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17173 Alan Hayward <alan.hayward@arm.com>
17174 David Sherwood <david.sherwood@arm.com>
17176 * optabs-query.h (extraction_insn::struct_mode): Change type to
17177 opt_scalar_int_mode and update comment.
17178 (extraction_insn::field_mode): Change type to scalar_int_mode.
17179 (extraction_insn::pos_mode): Likewise.
17180 * combine.c (make_extraction): Update accordingly.
17181 * optabs-query.c (get_traditional_extraction_insn): Likewise.
17182 (get_optab_extraction_insn): Likewise.
17183 * recog.c (simplify_while_replacing): Likewise.
17184 * expmed.c (narrow_bit_field_mem): Change the type of the mode
17185 parameter to opt_scalar_int_mode.
17187 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17188 Alan Hayward <alan.hayward@arm.com>
17189 David Sherwood <david.sherwood@arm.com>
17191 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
17192 to a scalar_int_mode instead of a machine_mode.
17193 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
17194 (get_best_mode): Return a boolean and use a pointer argument to store
17195 the selected mode. Replace the limit mode parameter with a bit limit.
17196 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
17197 for the values returned by bit_field_mode_iterator::next_mode.
17198 (store_bit_field): Update call to get_best_mode.
17199 (store_fixed_bit_field): Likewise.
17200 (extract_fixed_bit_field): Likewise.
17201 * expr.c (optimize_bitfield_assignment_op): Likewise.
17202 * fold-const.c (optimize_bit_field_compare): Likewise.
17203 (fold_truth_andor_1): Likewise.
17204 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
17205 Update for new type of m_mode.
17206 (get_best_mode): As above.
17208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17209 Alan Hayward <alan.hayward@arm.com>
17210 David Sherwood <david.sherwood@arm.com>
17212 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
17213 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
17214 (store_bit_field): Check is_a <scalar_int_mode> before calling
17215 strict_volatile_bitfield_p.
17216 (extract_bit_field): Likewise.
17218 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17219 Alan Hayward <alan.hayward@arm.com>
17220 David Sherwood <david.sherwood@arm.com>
17222 * target.def (cstore_mode): Return a scalar_int_mode.
17223 * doc/tm.texi: Regenerate.
17224 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
17225 * targhooks.h (default_cstore_mode): Likewise.
17226 * targhooks.c (default_cstore_mode): Likewise, using a forced
17228 * expmed.c (emit_cstore): Expect the target of the cstore to be
17231 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17232 Alan Hayward <alan.hayward@arm.com>
17233 David Sherwood <david.sherwood@arm.com>
17235 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
17237 (niter_desc): Likewise mode.
17238 (iv_analyze): Add a mode parameter.
17240 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
17241 and change its type to scalar_int_mode.
17242 * loop-iv.c: Update commentary at head of file.
17243 (iv_constant): Pass the mode paraeter before the rtx it describes
17244 and change its type to scalar_int_mode. Remove VOIDmode handling.
17245 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
17246 (iv_extend): Likewise.
17247 (shorten_into_mode): Likewise.
17248 (iv_add): Use scalar_int_mode.
17249 (iv_mult): Likewise.
17250 (iv_shift): Likewise.
17251 (canonicalize_iv_subregs): Likewise.
17252 (get_biv_step_1): Pass the outer_mode parameter before the rtx
17253 it describes and change its mode to scalar_int_mode. Also change
17254 the type of the returned inner_mode to scalar_int_mode.
17255 (get_biv_step): Likewise, turning outer_mode from a pointer
17256 into a direct parameter. Update call to get_biv_step_1.
17257 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
17258 iv_constant and get_biv_step.
17259 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
17260 and change its type to scalar_int_mode. Don't initialise iv->mode
17261 to VOIDmode and remove later checks for its still being VOIDmode.
17262 Update calls to iv_analyze_op and iv_analyze_expr. Check
17263 is_a <scalar_int_mode> when changing the mode under consideration.
17264 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
17265 Update call to iv_analyze_expr.
17266 (iv_analyze_op): Add a mode parameter. Reject subregs whose
17267 inner register is not also a scalar_int_mode. Update call to
17269 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
17270 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
17271 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
17272 separate mode class checks. Update calls to iv_analyze. Remove
17273 fix-up of VOIDmodes after iv_analyze_biv.
17274 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
17275 don't have a scalar_int_mode. Update call to biv_p.
17277 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17278 Alan Hayward <alan.hayward@arm.com>
17279 David Sherwood <david.sherwood@arm.com>
17281 * cfgexpand.c (convert_debug_memory_address): Use
17282 as_a <scalar_int_mode>.
17283 * combine.c (expand_compound_operation): Likewise.
17284 (make_extraction): Likewise.
17285 (change_zero_ext): Likewise.
17286 (simplify_comparison): Likewise.
17287 * cse.c (cse_insn): Likewise.
17288 * dwarf2out.c (minmax_loc_descriptor): Likewise.
17289 (mem_loc_descriptor): Likewise.
17290 (loc_descriptor): Likewise.
17291 * expmed.c (init_expmed_one_mode): Likewise.
17292 (synth_mult): Likewise.
17293 (emit_store_flag_1): Likewise.
17294 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
17295 of a comparison with size.
17296 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
17297 (reduce_to_bit_field_precision): Likewise.
17298 * function.c (expand_function_end): Likewise.
17299 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
17300 * loop-doloop.c (doloop_modify): Likewise.
17301 * optabs.c (expand_binop): Likewise.
17302 (expand_unop): Likewise.
17303 (expand_copysign_absneg): Likewise.
17304 (prepare_cmp_insn): Likewise.
17305 (maybe_legitimize_operand): Likewise.
17306 * recog.c (const_scalar_int_operand): Likewise.
17307 * rtlanal.c (get_address_mode): Likewise.
17308 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17309 (simplify_cond_clz_ctz): Likewise.
17310 * tree-nested.c (get_nl_goto_field): Likewise.
17311 * tree.c (build_vector_type_for_mode): Likewise.
17312 * var-tracking.c (use_narrower_mode): Likewise.
17314 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17315 Alan Hayward <alan.hayward@arm.com>
17316 David Sherwood <david.sherwood@arm.com>
17318 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
17319 * builtins.c (expand_builtin_signbit): Use it.
17320 * cfgexpand.c (expand_debug_expr): Likewise.
17321 * dojump.c (do_jump): Likewise.
17322 (do_compare_and_jump): Likewise.
17323 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
17324 * expmed.c (make_tree): Likewise.
17325 * expr.c (expand_expr_real_2): Likewise.
17326 (expand_expr_real_1): Likewise.
17327 (try_casesi): Likewise.
17328 * fold-const-call.c (fold_const_call_ss): Likewise.
17329 * fold-const.c (unextend): Likewise.
17330 (extract_muldiv_1): Likewise.
17331 (fold_single_bit_test): Likewise.
17332 (native_encode_int): Likewise.
17333 (native_encode_string): Likewise.
17334 (native_interpret_int): Likewise.
17335 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
17336 * internal-fn.c (expand_addsub_overflow): Likewise.
17337 (expand_neg_overflow): Likewise.
17338 (expand_mul_overflow): Likewise.
17339 (expand_arith_overflow): Likewise.
17340 * match.pd: Likewise.
17341 * stor-layout.c (layout_type): Likewise.
17342 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
17343 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
17344 * tree-ssanames.c (get_range_info): Likewise.
17345 * tree-switch-conversion.c (array_value_type) Likewise.
17346 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
17347 (vect_recog_divmod_pattern): Likewise.
17348 (vect_recog_mixed_size_cond_pattern): Likewise.
17349 * tree-vrp.c (extract_range_basic): Likewise.
17350 (simplify_float_conversion_using_ranges): Likewise.
17351 * tree.c (int_fits_type_p): Likewise.
17352 * ubsan.c (instrument_bool_enum_load): Likewise.
17353 * varasm.c (mergeable_string_section): Likewise.
17354 (narrowing_initializer_constant_valid_p): Likewise.
17355 (output_constant): Likewise.
17357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17358 Alan Hayward <alan.hayward@arm.com>
17359 David Sherwood <david.sherwood@arm.com>
17361 * machmode.h (NARROWEST_INT_MODE): New macro.
17362 * expr.c (alignment_for_piecewise_move): Use it instead of
17363 GET_CLASS_NARROWEST_MODE (MODE_INT).
17364 (push_block): Likewise.
17365 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
17367 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17370 Alan Hayward <alan.hayward@arm.com>
17371 David Sherwood <david.sherwood@arm.com>
17373 * postreload.c (move2add_valid_value_p): Change the type of the
17374 mode parameter to scalar_int_mode.
17375 (move2add_use_add2_insn): Add a mode parameter and use it instead
17377 (move2add_use_add3_insn): Likewise.
17378 (reload_cse_move2add): Update accordingly.
17380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17381 Alan Hayward <alan.hayward@arm.com>
17382 David Sherwood <david.sherwood@arm.com>
17384 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
17386 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
17387 * optabs.c (expand_unop): Likewise.
17389 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17390 Alan Hayward <alan.hayward@arm.com>
17391 David Sherwood <david.sherwood@arm.com>
17393 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
17394 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
17395 (popcount_loc_descriptor): Likewise.
17396 (bswap_loc_descriptor): Likewise.
17397 (rotate_loc_descriptor): Likewise.
17398 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
17399 calling the functions above.
17401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17402 Alan Hayward <alan.hayward@arm.com>
17403 David Sherwood <david.sherwood@arm.com>
17405 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
17407 (try_combine): Likewise.
17408 (simplify_if_then_else): Likewise.
17409 * cse.c (cse_insn): Likewise.
17410 * dwarf2out.c (mem_loc_descriptor): Likewise.
17411 * emit-rtl.c (gen_lowpart_common): Likewise.
17412 * simplify-rtx.c (simplify_truncation): Likewise.
17413 (simplify_binary_operation_1): Likewise.
17414 (simplify_const_relational_operation): Likewise.
17415 (simplify_ternary_operation): Likewise.
17416 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
17418 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17419 Alan Hayward <alan.hayward@arm.com>
17420 David Sherwood <david.sherwood@arm.com>
17422 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
17423 * reload.c (push_reload): Likewise.
17424 (find_reloads): Likewise.
17426 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17427 Alan Hayward <alan.hayward@arm.com>
17428 David Sherwood <david.sherwood@arm.com>
17430 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
17431 (make_compound_operation_int): Likewise.
17432 (change_zero_ext): Likewise.
17433 * expr.c (convert_move): Likewise.
17434 (convert_modes): Likewise.
17435 * fwprop.c (forward_propagate_subreg): Likewise.
17436 * loop-iv.c (get_biv_step_1): Likewise.
17437 * optabs.c (widen_operand): Likewise.
17438 * postreload.c (move2add_valid_value_p): Likewise.
17439 * recog.c (simplify_while_replacing): Likewise.
17440 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17441 (simplify_binary_operation_1): Likewise. Remove redundant
17442 mode equality check.
17444 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17445 Alan Hayward <alan.hayward@arm.com>
17446 David Sherwood <david.sherwood@arm.com>
17448 * combine.c (combine_simplify_rtx): Add checks for
17449 is_a <scalar_int_mode>.
17450 (simplify_if_then_else): Likewise.
17451 (make_field_assignment): Likewise.
17452 (simplify_comparison): Likewise.
17453 * ifcvt.c (noce_try_bitop): Likewise.
17454 * loop-invariant.c (canonicalize_address_mult): Likewise.
17455 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17458 Alan Hayward <alan.hayward@arm.com>
17459 David Sherwood <david.sherwood@arm.com>
17461 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
17462 is_a <scalar_int_mode> instead of != BLKmode.
17464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17465 Alan Hayward <alan.hayward@arm.com>
17466 David Sherwood <david.sherwood@arm.com>
17468 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
17469 instead of != VOIDmode.
17470 * combine.c (if_then_else_cond): Likewise.
17471 (change_zero_ext): Likewise.
17472 * dwarf2out.c (mem_loc_descriptor): Likewise.
17473 (loc_descriptor): Likewise.
17474 * rtlanal.c (canonicalize_condition): Likewise.
17475 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
17477 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17478 Alan Hayward <alan.hayward@arm.com>
17479 David Sherwood <david.sherwood@arm.com>
17481 * simplify-rtx.c (simplify_binary_operation_1): Use
17482 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
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 * wide-int.h (int_traits<unsigned char>) New class.
17489 (int_traits<unsigned short>) Likewise.
17490 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
17491 Use GET_MODE_UNIT_PRECISION and remove redundant test for
17493 * combine.c (set_nonzero_bits_and_sign_copies): Use
17494 is_a <scalar_int_mode>.
17495 (find_split_point): Likewise.
17496 (combine_simplify_rtx): Likewise.
17497 (simplify_logical): Likewise.
17498 (expand_compound_operation): Likewise.
17499 (expand_field_assignment): Likewise.
17500 (make_compound_operation): Likewise.
17501 (extended_count): Likewise.
17502 (change_zero_ext): Likewise.
17503 (simplify_comparison): Likewise.
17504 * dwarf2out.c (scompare_loc_descriptor): Likewise.
17505 (ucompare_loc_descriptor): Likewise.
17506 (minmax_loc_descriptor): Likewise.
17507 (mem_loc_descriptor): Likewise.
17508 (loc_descriptor): Likewise.
17509 * expmed.c (init_expmed_one_mode): Likewise.
17510 * lra-constraints.c (lra_constraint_offset): Likewise.
17511 * optabs.c (prepare_libcall_arg): Likewise.
17512 * postreload.c (move2add_note_store): Likewise.
17513 * reload.c (operands_match_p): Likewise.
17514 * rtl.h (load_extend_op): Likewise.
17515 * rtlhooks.c (gen_lowpart_general): Likewise.
17516 * simplify-rtx.c (simplify_truncation): Likewise.
17517 (simplify_unary_operation_1): Likewise.
17518 (simplify_binary_operation_1): Likewise.
17519 (simplify_const_binary_operation): Likewise.
17520 (simplify_const_relational_operation): Likewise.
17521 (simplify_subreg): Likewise.
17522 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17523 * var-tracking.c (adjust_mems): Likewise.
17524 (prepare_call_arguments): Likewise.
17526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17527 Alan Hayward <alan.hayward@arm.com>
17528 David Sherwood <david.sherwood@arm.com>
17530 * machmode.h (is_int_mode): New fuction.
17531 * combine.c (find_split_point): Use it.
17532 (combine_simplify_rtx): Likewise.
17533 (simplify_if_then_else): Likewise.
17534 (simplify_set): Likewise.
17535 (simplify_shift_const_1): Likewise.
17536 (simplify_comparison): Likewise.
17537 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
17538 * cse.c (notreg_cost): Likewise.
17539 (cse_insn): Likewise.
17540 * cselib.c (cselib_lookup_1): Likewise.
17541 * dojump.c (do_jump_1): Likewise.
17542 (do_compare_rtx_and_jump): Likewise.
17543 * dse.c (get_call_args): Likewise.
17544 * dwarf2out.c (rtl_for_decl_init): Likewise.
17545 (native_encode_initializer): Likewise.
17546 * expmed.c (emit_store_flag_1): Likewise.
17547 (emit_store_flag): Likewise.
17548 * expr.c (convert_modes): Likewise.
17549 (store_field): Likewise.
17550 (expand_expr_real_1): Likewise.
17551 * fold-const.c (fold_read_from_constant_string): Likewise.
17552 * gimple-ssa-sprintf.c (get_format_string): Likewise.
17553 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
17554 * optabs.c (expand_binop): Likewise.
17555 (expand_unop): Likewise.
17556 (expand_abs_nojump): Likewise.
17557 (expand_one_cmpl_abs_nojump): Likewise.
17558 * simplify-rtx.c (mode_signbit_p): Likewise.
17559 (val_signbit_p): Likewise.
17560 (val_signbit_known_set_p): Likewise.
17561 (val_signbit_known_clear_p): Likewise.
17562 (simplify_relational_operation_1): Likewise.
17563 * tree.c (vector_type_mode): Likewise.
17565 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17566 Alan Hayward <alan.hayward@arm.com>
17567 David Sherwood <david.sherwood@arm.com>
17569 * machmode.h (smallest_mode_for_size): Fix formatting.
17570 (smallest_int_mode_for_size): New function.
17571 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
17572 instead of smallest_mode_for_size.
17573 * combine.c (make_extraction): Likewise.
17574 * config/arc/arc.c (arc_expand_movmem): Likewise.
17575 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
17576 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17577 * config/s390/s390.c (s390_expand_insv): Likewise.
17578 * config/sparc/sparc.c (assign_int_registers): Likewise.
17579 * config/spu/spu.c (spu_function_value): Likewise.
17580 (spu_function_arg): Likewise.
17581 * coverage.c (get_gcov_type): Likewise.
17582 (get_gcov_unsigned_t): Likewise.
17583 * dse.c (find_shift_sequence): Likewise.
17584 * expmed.c (store_bit_field_1): Likewise.
17585 * expr.c (convert_move): Likewise.
17586 (store_field): Likewise.
17587 * internal-fn.c (expand_arith_overflow): Likewise.
17588 * optabs-query.c (get_best_extraction_insn): Likewise.
17589 * optabs.c (expand_twoval_binop_libfunc): Likewise.
17590 * stor-layout.c (layout_type): Likewise.
17591 (initialize_sizetypes): Likewise.
17592 * targhooks.c (default_get_mask_mode): Likewise.
17593 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
17595 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17596 Alan Hayward <alan.hayward@arm.com>
17597 David Sherwood <david.sherwood@arm.com>
17599 * machmode.h (opt_mode::else_blk): New function.
17600 (int_mode_for_mode): Declare.
17601 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
17602 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
17604 * cfgexpand.c (expand_debug_expr): Likewise.
17605 * combine.c (gen_lowpart_or_truncate): Likewise.
17606 (gen_lowpart_for_combine): Likewise.
17607 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
17608 * config/avr/avr.c (avr_to_int_mode): Likewise.
17609 (avr_out_plus_1): Likewise.
17610 (avr_out_plus): Likewise.
17611 (avr_out_round): Likewise.
17612 * config/i386/i386.c (ix86_split_to_parts): Likewise.
17613 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17614 (s390_expand_vcond): Likewise.
17615 * config/spu/spu.c (spu_split_immediate): Likewise.
17616 (spu_expand_mov): Likewise.
17617 * dse.c (get_stored_val): Likewise.
17618 * expmed.c (store_bit_field_1): Likewise.
17619 (convert_extracted_bit_field): Use int_mode_for_mode instead of
17621 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
17622 (extract_low_bits): Likewise.
17623 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
17624 handling rather than repeating the check.
17625 (emit_group_store): Likewise.
17626 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
17627 * optabs.c (expand_absneg_bit): Likewise.
17628 (expand_copysign_absneg): Likewise.
17629 (expand_copysign_bit): Likewise.
17630 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
17631 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
17632 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
17633 * var-tracking.c (prepare_call_arguments): Likewise.
17634 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
17635 int_mode_for_mode instead of mode_for_size.
17636 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17638 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17639 Alan Hayward <alan.hayward@arm.com>
17640 David Sherwood <david.sherwood@arm.com>
17642 * machmode.h (int_mode_for_size): New function.
17643 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
17644 instead of mode_for_size.
17645 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
17647 * combine.c (expand_field_assignment): Use int_mode_for_size
17648 instead of mode_for_size.
17649 (make_extraction): Likewise.
17650 (simplify_shift_const_1): Likewise.
17651 (simplify_comparison): Likewise.
17652 * dojump.c (do_jump): Likewise.
17653 * dwarf2out.c (mem_loc_descriptor): Likewise.
17654 * emit-rtl.c (init_derived_machine_modes): Likewise.
17655 * expmed.c (flip_storage_order): Likewise.
17656 (convert_extracted_bit_field): Likewise.
17657 * expr.c (copy_blkmode_from_reg): Likewise.
17658 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
17659 * internal-fn.c (expand_mul_overflow): Likewise.
17660 * lower-subreg.c (simple_move): Likewise.
17661 * optabs-libfuncs.c (init_optabs): Likewise.
17662 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
17663 * tree.c (vector_type_mode): Likewise.
17664 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
17665 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
17666 * tree-vect-generic.c (expand_vector_parallel): Likewise.
17667 * tree-vect-stmts.c (vectorizable_load): Likewise.
17668 (vectorizable_store): Likewise.
17670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17671 Alan Hayward <alan.hayward@arm.com>
17672 David Sherwood <david.sherwood@arm.com>
17674 * coretypes.h (pod_mode): New type.
17675 (scalar_int_mode_pod): New typedef.
17676 * machmode.h (pod_mode): New class.
17677 (int_n_data_t::m): Change type to scalar_int_mode_pod.
17678 * genmodes.c (emit_mode_int_n): Update accordingly.
17679 * lower-subreg.h (target_lower_subreg): Change type to
17680 scalar_int_mode_pod.
17681 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
17682 scalar_int_mode_pod.
17684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17685 Alan Hayward <alan.hayward@arm.com>
17686 David Sherwood <david.sherwood@arm.com>
17688 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
17689 machine_mode to scalar_int_mode.
17690 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
17691 (rs6000_option_override_internal): Remove cast to int.
17692 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
17693 machine_mode to scalar_int_mode.
17694 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
17695 (rs6000_option_override_internal): Remove cast to int.
17696 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
17697 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
17699 * config/s390/s390.c (s390_expand_builtin): Likewise.
17700 * coretypes.h (scalar_int_mode): New type.
17701 (opt_scalar_int_mode): New typedef.
17702 * machmode.h (scalar_int_mode): New class.
17703 (scalar_int_mode::includes_p): New function.
17704 (byte_mode): Change type to scalar_int_mode.
17705 (word_mode): Likewise.
17706 (ptr_mode): Likewise.
17707 * emit-rtl.c (byte_mode): Likewise.
17708 (word_mode): Likewise.
17709 (ptr_mode): Likewise.
17710 (init_derived_machine_modes): Update accordingly.
17711 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
17712 and MODE_PARTIAL_INT.
17713 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
17714 opt_scalar_int_mode.
17716 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17717 Alan Hayward <alan.hayward@arm.com>
17718 David Sherwood <david.sherwood@arm.com>
17720 * target.def (libgcc_floating_mode_supported_p): Take a
17722 * doc/tm.texi: Regenerate.
17723 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
17725 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
17726 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
17729 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17730 Alan Hayward <alan.hayward@arm.com>
17731 David Sherwood <david.sherwood@arm.com>
17733 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
17734 * doc/tm.texi: Regenerate.
17735 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
17736 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
17737 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
17738 * targhooks.h (default_floatn_mode): Likewise.
17739 * targhooks.c (default_floatn_mode): Likewise.
17740 * tree.c (build_common_tree_nodes): Update accordingly.
17742 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17743 Alan Hayward <alan.hayward@arm.com>
17744 David Sherwood <david.sherwood@arm.com>
17746 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
17747 (mode_iterator::iterate_p): Likewise.
17748 (mode_iterator::get_wider): Likewise.
17749 * expr.c (init_expr_target): Use opt_scalar_float_mode.
17751 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17752 Alan Hayward <alan.hayward@arm.com>
17753 David Sherwood <david.sherwood@arm.com>
17755 * coretypes.h (opt_scalar_float_mode): New typedef.
17756 * machmode.h (float_mode_for_size): New function.
17757 * emit-rtl.c (double_mode): Delete.
17758 (init_emit_once): Use float_mode_for_size.
17759 * stor-layout.c (layout_type): Likewise.
17760 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
17762 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17763 Alan Hayward <alan.hayward@arm.com>
17764 David Sherwood <david.sherwood@arm.com>
17766 * output.h (assemble_real): Take a scalar_float_mode.
17767 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
17768 * config/arm/arm.md (consttable_4): Likewise.
17769 (consttable_8): Likewise.
17770 (consttable_16): Likewise.
17771 * config/mips/mips.md (consttable_float): Likewise.
17772 * config/s390/s390.c (s390_output_pool_entry): Likewise.
17773 * varasm.c (assemble_real): Take a scalar_float_mode.
17774 (output_constant_pool_2): Update accordingly.
17775 (output_constant): Likewise.
17777 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17778 Alan Hayward <alan.hayward@arm.com>
17779 David Sherwood <david.sherwood@arm.com>
17781 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
17782 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
17783 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
17784 (native_encode_real): Likewise.
17785 (native_interpret_real): Likewise.
17786 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
17787 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17789 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17790 Alan Hayward <alan.hayward@arm.com>
17791 David Sherwood <david.sherwood@arm.com>
17793 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
17794 <scalar_float_mode>. Simplify.
17795 (gen_extend_conv_libfunc): Likewise.
17797 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17798 Alan Hayward <alan.hayward@arm.com>
17799 David Sherwood <david.sherwood@arm.com>
17801 * coretypes.h (scalar_float_mode): New type.
17802 * machmode.h (mode_traits::from_int): Use machine_mode if
17803 USE_ENUM_MODES is defined.
17804 (is_a): New function.
17806 (dyn_cast): Likewise.
17807 (scalar_float_mode): New class.
17808 (scalar_float_mode::includes_p): New function.
17809 (is_float_mode): Likewise.
17810 * gdbhooks.py (MachineModePrinter): New class.
17811 (build_pretty_printer): Use it for scalar_float_mode.
17812 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
17813 (format_helper::format_helper): Turn into a template.
17814 * genmodes.c (get_mode_class): New function.
17815 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
17816 or machine_mode if none.
17817 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
17818 as_a <scalar_float_mode>.
17819 * dwarf2out.c (mem_loc_descriptor): Likewise.
17820 (insert_float): Likewise.
17821 (add_const_value_attribute): Likewise.
17822 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17823 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
17824 (expand_unop): Update accordingly.
17825 (expand_abs_nojump): Likewise.
17826 (expand_copysign_absneg): Take a scalar_float_mode.
17827 (expand_copysign_bit): Likewise.
17828 (expand_copysign): Update accordingly.
17830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17831 Alan Hayward <alan.hayward@arm.com>
17832 David Sherwood <david.sherwood@arm.com>
17834 * coretypes.h (opt_mode): New class.
17835 * machmode.h (opt_mode): Likewise.
17836 (opt_mode::else_void): New function.
17837 (opt_mode::require): Likewise.
17838 (opt_mode::exists): Likewise.
17839 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
17840 (GET_MODE_2XWIDER_MODE): Likewise.
17841 (mode_iterator::get_wider): Update accordingly.
17842 (mode_iterator::get_2xwider): Likewise.
17843 (mode_iterator::get_known_wider): Likewise, turning into a template.
17844 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
17845 forcing a wider mode to exist.
17846 * config/cr16/cr16.h (LONG_REG_P): Likewise.
17847 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17848 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
17849 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
17850 * lower-subreg.c (init_lower_subreg): Likewise.
17851 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
17852 on the final iteration.
17853 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
17854 a wider mode exists before asking for a move pattern.
17855 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
17856 forcing a wider mode to exist.
17857 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
17858 returning false if no such mode exists.
17859 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
17860 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
17861 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
17862 Avoid checking for a MODE_INT if we already know the mode is not a
17864 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
17865 forcing a wider mode to exist.
17866 (expmed_mult_highpart_optab): Likewise.
17867 (expmed_mult_highpart): Likewise.
17868 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
17870 * lto-streamer-in.c (lto_input_mode_table): Likewise.
17871 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
17872 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
17873 * internal-fn.c (expand_mul_overflow): Update use of
17874 GET_MODE_2XWIDER_MODE.
17875 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17876 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
17877 GET_MODE_WIDER_MODE.
17878 (convert_plusminus_to_widen): Likewise.
17879 * tree-switch-conversion.c (array_value_type): Likewise.
17880 * var-tracking.c (emit_note_insn_var_location): Likewise.
17881 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17882 Return false inside rather than outside the loop if no wider mode
17884 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
17885 and GET_MODE_2XWIDER_MODE
17886 (can_compare_p): Use else_void.
17887 * gdbhooks.py (OptMachineModePrinter): New class.
17888 (build_pretty_printer): Use it for opt_mode.
17890 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17891 Alan Hayward <alan.hayward@arm.com>
17892 David Sherwood <david.sherwood@arm.com>
17894 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
17895 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
17897 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17898 Alan Hayward <alan.hayward@arm.com>
17899 David Sherwood <david.sherwood@arm.com>
17901 * machmode.h (mode_traits): New structure.
17902 (get_narrowest_mode): New function.
17903 (mode_iterator::start): Likewise.
17904 (mode_iterator::iterate_p): Likewise.
17905 (mode_iterator::get_wider): Likewise.
17906 (mode_iterator::get_known_wider): Likewise.
17907 (mode_iterator::get_2xwider): Likewise.
17908 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
17909 (FOR_EACH_MODE): Likewise.
17910 (FOR_EACH_MODE_FROM): Likewise.
17911 (FOR_EACH_MODE_UNTIL): Likewise.
17912 (FOR_EACH_WIDER_MODE): Likewise.
17913 (FOR_EACH_2XWIDER_MODE): Likewise.
17914 * builtins.c (expand_builtin_strlen): Use new mode iterators.
17915 * combine.c (simplify_comparison): Likewise
17916 * config/i386/i386.c (type_natural_mode): Likewise.
17917 * cse.c (cse_insn): Likewise.
17918 * dse.c (find_shift_sequence): Likewise.
17919 * emit-rtl.c (init_derived_machine_modes): Likewise.
17920 (init_emit_once): Likewise.
17921 * explow.c (hard_function_value): Likewise.
17922 * expmed.c (extract_fixed_bit_field_1): Likewise.
17923 (extract_bit_field_1): Likewise.
17924 (expand_divmod): Likewise.
17925 (emit_store_flag_1): Likewise.
17926 * expr.c (init_expr_target): Likewise.
17927 (convert_move): Likewise.
17928 (alignment_for_piecewise_move): Likewise.
17929 (widest_int_mode_for_size): Likewise.
17930 (emit_block_move_via_movmem): Likewise.
17931 (copy_blkmode_to_reg): Likewise.
17932 (set_storage_via_setmem): Likewise.
17933 (compress_float_constant): Likewise.
17934 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17935 * optabs-query.c (get_best_extraction_insn): Likewise.
17936 * optabs.c (expand_binop): Likewise.
17937 (expand_twoval_unop): Likewise.
17938 (expand_twoval_binop): Likewise.
17939 (widen_leading): Likewise.
17940 (widen_bswap): Likewise.
17941 (expand_parity): Likewise.
17942 (expand_unop): Likewise.
17943 (prepare_cmp_insn): Likewise.
17944 (prepare_float_lib_cmp): Likewise.
17945 (expand_float): Likewise.
17946 (expand_fix): Likewise.
17947 (expand_sfix_optab): Likewise.
17948 * postreload.c (move2add_use_add2_insn): Likewise.
17949 * reg-stack.c (reg_to_stack): Likewise.
17950 * reginfo.c (choose_hard_reg_mode): Likewise.
17951 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17952 * stor-layout.c (mode_for_size): Likewise.
17953 (smallest_mode_for_size): Likewise.
17954 (mode_for_vector): Likewise.
17955 (finish_bitfield_representative): Likewise.
17956 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
17957 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
17958 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17959 * var-tracking.c (prepare_call_arguments): Likewise.
17961 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17962 Alan Hayward <alan.hayward@arm.com>
17963 David Sherwood <david.sherwood@arm.com>
17965 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
17966 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
17967 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
17968 * machmode.h (mode_size): Move earlier in file.
17969 (mode_precision): Likewise.
17970 (mode_inner): Likewise.
17971 (mode_nunits): Likewise.
17972 (mode_unit_size): Likewise.
17973 (unit_unit_precision): Likewise.
17974 (mode_wider): Likewise.
17975 (mode_2xwider): Likewise.
17976 (machine_mode): New class.
17977 (mode_to_bytes): New function.
17978 (mode_to_bits): Likewise.
17979 (mode_to_precision): Likewise.
17980 (mode_to_inner): Likewise.
17981 (mode_to_unit_size): Likewise.
17982 (mode_to_unit_precision): Likewise.
17983 (mode_to_nunits): Likewise.
17984 (GET_MODE_SIZE): Use mode_to_bytes.
17985 (GET_MODE_BITSIZE): Use mode_to_bits.
17986 (GET_MODE_PRECISION): Use mode_to_precision.
17987 (GET_MODE_INNER): Use mode_to_inner.
17988 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
17989 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
17990 (GET_MODE_NUNITS): Use mode_to_nunits.
17991 * system.h (ALWAYS_INLINE): New macro.
17992 * config/powerpcspe/powerpcspe-c.c
17993 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
17994 int for arg1_mode and arg2_mode.
17996 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17997 Alan Hayward <alan.hayward@arm.com>
17998 David Sherwood <david.sherwood@arm.com>
18000 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18001 Prefix mode names with E_ in case statements.
18002 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18003 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
18004 (aarch64_split_simd_move): Likewise.
18005 (aarch64_gen_storewb_pair): Likewise.
18006 (aarch64_gen_loadwb_pair): Likewise.
18007 (aarch64_gen_store_pair): Likewise.
18008 (aarch64_gen_load_pair): Likewise.
18009 (aarch64_get_condition_code_1): Likewise.
18010 (aarch64_constant_pool_reload_icode): Likewise.
18011 (get_rsqrte_type): Likewise.
18012 (get_rsqrts_type): Likewise.
18013 (get_recpe_type): Likewise.
18014 (get_recps_type): Likewise.
18015 (aarch64_gimplify_va_arg_expr): Likewise.
18016 (aarch64_simd_container_mode): Likewise.
18017 (aarch64_emit_load_exclusive): Likewise.
18018 (aarch64_emit_store_exclusive): Likewise.
18019 (aarch64_expand_compare_and_swap): Likewise.
18020 (aarch64_gen_atomic_cas): Likewise.
18021 (aarch64_emit_bic): Likewise.
18022 (aarch64_emit_atomic_swap): Likewise.
18023 (aarch64_emit_atomic_load_op): Likewise.
18024 (aarch64_evpc_trn): Likewise.
18025 (aarch64_evpc_uzp): Likewise.
18026 (aarch64_evpc_zip): Likewise.
18027 (aarch64_evpc_ext): Likewise.
18028 (aarch64_evpc_rev): Likewise.
18029 (aarch64_evpc_dup): Likewise.
18030 (aarch64_gen_ccmp_first): Likewise.
18031 (aarch64_gen_ccmp_next): Likewise.
18032 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18033 (alpha_emit_xfloating_libcall): Likewise.
18034 (emit_insxl): Likewise.
18035 (alpha_arg_type): Likewise.
18036 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
18037 (arc_preferred_simd_mode): Likewise.
18038 (arc_secondary_reload): Likewise.
18039 (get_arc_condition_code): Likewise.
18040 (arc_print_operand): Likewise.
18041 (arc_legitimate_constant_p): Likewise.
18042 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18043 * config/arc/arc.md (casesi_load): Likewise.
18044 (casesi_compact_jump): Likewise.
18045 * config/arc/predicates.md (proper_comparison_operator): Likewise.
18046 (cc_use_register): Likewise.
18047 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18048 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18049 (arm_init_iwmmxt_builtins): Likewise.
18050 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
18051 (neon_expand_vector_init): Likewise.
18052 (arm_attr_length_move_neon): Likewise.
18053 (maybe_get_arm_condition_code): Likewise.
18054 (arm_emit_vector_const): Likewise.
18055 (arm_preferred_simd_mode): Likewise.
18056 (arm_output_iwmmxt_tinsr): Likewise.
18057 (thumb1_output_casesi): Likewise.
18058 (thumb2_output_casesi): Likewise.
18059 (arm_emit_load_exclusive): Likewise.
18060 (arm_emit_store_exclusive): Likewise.
18061 (arm_expand_compare_and_swap): Likewise.
18062 (arm_evpc_neon_vuzp): Likewise.
18063 (arm_evpc_neon_vzip): Likewise.
18064 (arm_evpc_neon_vrev): Likewise.
18065 (arm_evpc_neon_vtrn): Likewise.
18066 (arm_evpc_neon_vext): Likewise.
18067 (arm_validize_comparison): Likewise.
18068 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
18069 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
18070 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
18071 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
18072 (c6x_preferred_simd_mode): Likewise.
18073 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
18074 (epiphany_rtx_costs): Likewise.
18075 * config/epiphany/predicates.md (proper_comparison_operator):
18077 * config/frv/frv.c (condexec_memory_operand): Likewise.
18078 (frv_emit_move): Likewise.
18079 (output_move_single): Likewise.
18080 (output_condmove_single): Likewise.
18081 (frv_hard_regno_mode_ok): Likewise.
18082 (frv_matching_accg_mode): Likewise.
18083 * config/h8300/h8300.c (split_adds_subs): Likewise.
18084 (h8300_rtx_costs): Likewise.
18085 (h8300_print_operand): Likewise.
18086 (compute_mov_length): Likewise.
18087 (output_logical_op): Likewise.
18088 (compute_logical_op_length): Likewise.
18089 (compute_logical_op_cc): Likewise.
18090 (h8300_shift_needs_scratch_p): Likewise.
18091 (output_a_shift): Likewise.
18092 (compute_a_shift_length): Likewise.
18093 (compute_a_shift_cc): Likewise.
18094 (expand_a_rotate): Likewise.
18095 (output_a_rotate): Likewise.
18096 * config/i386/i386.c (classify_argument): Likewise.
18097 (function_arg_advance_32): Likewise.
18098 (function_arg_32): Likewise.
18099 (function_arg_64): Likewise.
18100 (function_value_64): Likewise.
18101 (ix86_gimplify_va_arg): Likewise.
18102 (ix86_legitimate_constant_p): Likewise.
18103 (put_condition_code): Likewise.
18104 (split_double_mode): Likewise.
18105 (ix86_avx256_split_vector_move_misalign): Likewise.
18106 (ix86_expand_vector_logical_operator): Likewise.
18107 (ix86_split_idivmod): Likewise.
18108 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
18109 (ix86_build_const_vector): Likewise.
18110 (ix86_build_signbit_mask): Likewise.
18111 (ix86_match_ccmode): Likewise.
18112 (ix86_cc_modes_compatible): Likewise.
18113 (ix86_expand_branch): Likewise.
18114 (ix86_expand_sse_cmp): Likewise.
18115 (ix86_expand_sse_movcc): Likewise.
18116 (ix86_expand_int_sse_cmp): Likewise.
18117 (ix86_expand_vec_perm_vpermi2): Likewise.
18118 (ix86_expand_vec_perm): Likewise.
18119 (ix86_expand_sse_unpack): Likewise.
18120 (ix86_expand_int_addcc): Likewise.
18121 (ix86_split_to_parts): Likewise.
18122 (ix86_vectorize_builtin_gather): Likewise.
18123 (ix86_vectorize_builtin_scatter): Likewise.
18124 (avx_vpermilp_parallel): Likewise.
18125 (inline_memory_move_cost): Likewise.
18126 (ix86_tieable_integer_mode_p): Likewise.
18127 (x86_maybe_negate_const_int): Likewise.
18128 (ix86_expand_vector_init_duplicate): Likewise.
18129 (ix86_expand_vector_init_one_nonzero): Likewise.
18130 (ix86_expand_vector_init_one_var): Likewise.
18131 (ix86_expand_vector_init_concat): Likewise.
18132 (ix86_expand_vector_init_interleave): Likewise.
18133 (ix86_expand_vector_init_general): Likewise.
18134 (ix86_expand_vector_set): Likewise.
18135 (ix86_expand_vector_extract): Likewise.
18136 (emit_reduc_half): Likewise.
18137 (ix86_emit_i387_round): Likewise.
18138 (ix86_mangle_type): Likewise.
18139 (ix86_expand_round_sse4): Likewise.
18140 (expand_vec_perm_blend): Likewise.
18141 (canonicalize_vector_int_perm): Likewise.
18142 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
18143 (expand_vec_perm_1): Likewise.
18144 (expand_vec_perm_interleave3): Likewise.
18145 (expand_vec_perm_even_odd_pack): Likewise.
18146 (expand_vec_perm_even_odd_1): Likewise.
18147 (expand_vec_perm_broadcast_1): Likewise.
18148 (ix86_vectorize_vec_perm_const_ok): Likewise.
18149 (ix86_expand_vecop_qihi): Likewise.
18150 (ix86_expand_mul_widen_hilo): Likewise.
18151 (ix86_expand_sse2_abs): Likewise.
18152 (ix86_expand_pextr): Likewise.
18153 (ix86_expand_pinsr): Likewise.
18154 (ix86_preferred_simd_mode): Likewise.
18155 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
18156 * config/i386/sse.md (*andnot<mode>3): Likewise.
18157 (<mask_codefor><code><mode>3<mask_name>): Likewise.
18158 (*<code><mode>3): Likewise.
18159 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
18160 (ia64_expand_atomic_op): Likewise.
18161 (ia64_arg_type): Likewise.
18162 (ia64_mode_to_int): Likewise.
18163 (ia64_scalar_mode_supported_p): Likewise.
18164 (ia64_vector_mode_supported_p): Likewise.
18165 (expand_vec_perm_broadcast): Likewise.
18166 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
18167 (iq2000_function_arg_advance): Likewise.
18168 (iq2000_function_arg): Likewise.
18169 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
18170 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
18171 (m68k_libcall_value): Likewise.
18172 (m68k_function_value): Likewise.
18173 (sched_attr_op_type): Likewise.
18174 * config/mcore/mcore.c (mcore_output_move): Likewise.
18175 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
18177 (microblaze_function_arg): Likewise.
18178 * config/mips/mips.c (mips16_build_call_stub): Likewise.
18179 (mips_print_operand): Likewise.
18180 (mips_mode_ok_for_mov_fmt_p): Likewise.
18181 (mips_vector_mode_supported_p): Likewise.
18182 (mips_preferred_simd_mode): Likewise.
18183 (mips_expand_vpc_loongson_even_odd): Likewise.
18184 (mips_expand_vec_unpack): Likewise.
18185 (mips_expand_vi_broadcast): Likewise.
18186 (mips_expand_vector_init): Likewise.
18187 (mips_expand_vec_reduc): Likewise.
18188 (mips_expand_msa_cmp): Likewise.
18189 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
18190 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
18191 (cc_flags_for_mode): Likewise.
18192 * config/msp430/msp430.c (msp430_print_operand): Likewise.
18193 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
18194 (nds32_output_casesi_pc_relative): Likewise.
18195 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18196 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
18197 (nvptx_gen_unpack): Likewise.
18198 (nvptx_gen_pack): Likewise.
18199 (nvptx_gen_shuffle): Likewise.
18200 (nvptx_gen_wcast): Likewise.
18201 (nvptx_preferred_simd_mode): Likewise.
18202 * config/pa/pa.c (pa_secondary_reload): Likewise.
18203 * config/pa/predicates.md (base14_operand): Likewise.
18204 * config/powerpcspe/powerpcspe-c.c
18205 (altivec_resolve_overloaded_builtin): Likewise.
18206 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
18208 (rs6000_preferred_simd_mode): Likewise.
18209 (output_vec_const_move): Likewise.
18210 (rs6000_expand_vector_extract): Likewise.
18211 (rs6000_split_vec_extract_var): Likewise.
18212 (reg_offset_addressing_ok_p): Likewise.
18213 (rs6000_legitimate_offset_address_p): Likewise.
18214 (rs6000_legitimize_address): Likewise.
18215 (rs6000_emit_set_const): Likewise.
18216 (rs6000_const_vec): Likewise.
18217 (rs6000_emit_move): Likewise.
18218 (spe_build_register_parallel): Likewise.
18219 (rs6000_darwin64_record_arg_recurse): Likewise.
18220 (swap_selector_for_mode): Likewise.
18221 (spe_init_builtins): Likewise.
18222 (paired_init_builtins): Likewise.
18223 (altivec_init_builtins): Likewise.
18224 (do_load_for_compare): Likewise.
18225 (rs6000_generate_compare): Likewise.
18226 (rs6000_expand_float128_convert): Likewise.
18227 (emit_load_locked): Likewise.
18228 (emit_store_conditional): Likewise.
18229 (rs6000_output_function_epilogue): Likewise.
18230 (rs6000_handle_altivec_attribute): Likewise.
18231 (rs6000_function_value): Likewise.
18232 (emit_fusion_gpr_load): Likewise.
18233 (emit_fusion_p9_load): Likewise.
18234 (emit_fusion_p9_store): Likewise.
18235 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
18236 (fusion_gpr_mem_load): Likewise.
18237 (fusion_addis_mem_combo_load): Likewise.
18238 (fusion_addis_mem_combo_store): Likewise.
18239 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
18240 (fusion_gpr_mem_load): Likewise.
18241 (fusion_addis_mem_combo_load): Likewise.
18242 (fusion_addis_mem_combo_store): Likewise.
18243 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18245 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
18246 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
18247 (rs6000_preferred_simd_mode): Likewise.
18248 (output_vec_const_move): Likewise.
18249 (rs6000_expand_vector_extract): Likewise.
18250 (rs6000_split_vec_extract_var): Likewise.
18251 (reg_offset_addressing_ok_p): Likewise.
18252 (rs6000_legitimate_offset_address_p): Likewise.
18253 (rs6000_legitimize_address): Likewise.
18254 (rs6000_emit_set_const): Likewise.
18255 (rs6000_const_vec): Likewise.
18256 (rs6000_emit_move): Likewise.
18257 (rs6000_darwin64_record_arg_recurse): Likewise.
18258 (swap_selector_for_mode): Likewise.
18259 (paired_init_builtins): Likewise.
18260 (altivec_init_builtins): Likewise.
18261 (rs6000_expand_float128_convert): Likewise.
18262 (emit_load_locked): Likewise.
18263 (emit_store_conditional): Likewise.
18264 (rs6000_output_function_epilogue): Likewise.
18265 (rs6000_handle_altivec_attribute): Likewise.
18266 (rs6000_function_value): Likewise.
18267 (emit_fusion_gpr_load): Likewise.
18268 (emit_fusion_p9_load): Likewise.
18269 (emit_fusion_p9_store): Likewise.
18270 * config/rx/rx.c (rx_gen_move_template): Likewise.
18271 (flags_from_mode): Likewise.
18272 * config/s390/predicates.md (s390_alc_comparison): Likewise.
18273 (s390_slb_comparison): Likewise.
18274 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
18275 (s390_vector_mode_supported_p): Likewise.
18276 (s390_cc_modes_compatible): Likewise.
18277 (s390_match_ccmode_set): Likewise.
18278 (s390_canonicalize_comparison): Likewise.
18279 (s390_emit_compare_and_swap): Likewise.
18280 (s390_branch_condition_mask): Likewise.
18281 (s390_rtx_costs): Likewise.
18282 (s390_secondary_reload): Likewise.
18283 (__SECONDARY_RELOAD_CASE): Likewise.
18284 (s390_expand_cs): Likewise.
18285 (s390_preferred_simd_mode): Likewise.
18286 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
18287 * config/sh/sh.c (sh_print_operand): Likewise.
18288 (dump_table): Likewise.
18289 (sh_secondary_reload): Likewise.
18290 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18291 * config/sh/sh.md (casesi_worker_1): Likewise.
18292 (casesi_worker_2): Likewise.
18293 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
18294 (fcc_comparison_operator): Likewise.
18295 * config/sparc/sparc.c (sparc_expand_move): Likewise.
18296 (emit_soft_tfmode_cvt): Likewise.
18297 (sparc_preferred_simd_mode): Likewise.
18298 (output_cbranch): Likewise.
18299 (sparc_print_operand): Likewise.
18300 (sparc_expand_vec_perm_bmask): Likewise.
18301 (vector_init_bshuffle): Likewise.
18302 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18303 (spu_vector_mode_supported_p): Likewise.
18304 (spu_expand_insv): Likewise.
18305 (spu_emit_branch_or_set): Likewise.
18306 (spu_handle_vector_attribute): Likewise.
18307 (spu_builtin_splats): Likewise.
18308 (spu_builtin_extract): Likewise.
18309 (spu_builtin_promote): Likewise.
18310 (spu_expand_sign_extend): Likewise.
18311 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18312 (tilegx_simd_int): Likewise.
18313 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
18314 (tilepro_simd_int): Likewise.
18315 * config/v850/v850.c (const_double_split): Likewise.
18316 (v850_print_operand): Likewise.
18317 (ep_memory_offset): Likewise.
18318 * config/vax/vax.c (vax_rtx_costs): Likewise.
18319 (vax_output_int_move): Likewise.
18320 (vax_output_int_add): Likewise.
18321 (vax_output_int_subtract): Likewise.
18322 * config/visium/predicates.md (visium_branch_operator): Likewise.
18323 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
18324 (visium_print_operand_address): Likewise.
18325 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
18326 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
18327 (xtensa_expand_conditional_branch): Likewise.
18328 (xtensa_copy_incoming_a7): Likewise.
18329 (xtensa_output_literal): Likewise.
18330 * dfp.c (decimal_real_maxval): Likewise.
18331 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
18333 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18334 Alan Hayward <alan.hayward@arm.com>
18335 David Sherwood <david.sherwood@arm.com>
18337 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
18338 (mode_nunits_inline): Likewise.
18339 (mode_inner_inline): Likewise.
18340 (mode_unit_size_inline): Likewise.
18341 (mode_unit_precision_inline): Likewise.
18342 (emit_insn_modes_h): Likewise. Also emit a #define of the
18344 (emit_mode_wider): Add an E_ prefix to mode names.
18345 (emit_mode_complex): Likewise.
18346 (emit_mode_inner): Likewise.
18347 (emit_mode_adjustments): Likewise.
18348 (emit_mode_int_n): Likewise.
18349 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
18350 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
18351 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
18352 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
18353 (CRC32_BUILTIN, ENTRY): Likewise.
18354 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
18355 (aarch64_pop_regs): Likewise.
18356 (aarch64_process_components): Likewise.
18357 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
18358 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
18359 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
18360 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
18361 * config/arm/arm.c (arm_init_libfuncs): Likewise.
18362 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
18364 * config/i386/i386-builtin.def (pcmpestr): Likewise.
18365 (pcmpistr): Likewise.
18366 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
18367 * config/mmix/mmix.c (mmix_output_condition): Likewise.
18368 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18370 * config/rl78/rl78.c (mduc_regs): Likewise.
18371 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18372 (htm_expand_builtin): Likewise.
18373 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
18374 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
18375 * config/xtensa/xtensa.c (print_operand): Likewise.
18376 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
18377 (NUM_MODE_VECTOR_INT): Likewise.
18378 * genoutput.c (null_operand): Likewise.
18379 (output_operand_data): Likewise.
18380 * genrecog.c (print_parameter_value): Likewise.
18381 * lra.c (debug_operand_data): Likewise.
18383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18385 * dwarf2out.c (scompare_loc_descriptor_wide)
18386 (scompare_loc_descriptor_narrow): New functions, split out from...
18387 (scompare_loc_descriptor): ...here.
18388 * expmed.c (emit_store_flag_int): New function, split out from...
18389 (emit_store_flag): ...here.
18391 2017-08-30 Richard Biener <rguenther@suse.de>
18393 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
18394 (dwarf2out_early_finish): Move setting of AT_pubnames from
18395 early debug output to early finish.
18397 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18399 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
18400 and -mdata-region to the assembler.
18402 2017-08-30 Richard Biener <rguenther@suse.de>
18404 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
18406 (gen_subprogram_die): Add DW_AT_object_pointer only early.
18407 (dwarf2out_early_global_decl): Only generate a DIE for the
18408 abstract origin if it doesn't already exist or is a declaration DIE.
18409 (resolve_addr): Do not add the linkage name twice when
18410 generating a stub DIE for the DW_TAG_GNU_call_site target.
18412 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18414 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18415 Use machine_mode rather than int for arg1_mode.
18417 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
18420 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
18421 that the second argument of the built-in functions to unpack
18422 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
18423 switch statement instead a lot of if statements.
18424 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
18425 Allow 64-bit values to be in Altivec registers as well as
18426 traditional floating point registers.
18427 (pack<mode>, FMOVE128_VSX iterator): Likewise.
18429 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
18431 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
18432 MAX_REGS_PER_ADDRESS == 1.
18434 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
18436 * config/i386/i386.opt (flag_fentry): Do not init to -1.
18437 * config/i386/i386.c (ix86_option_override_internal): Simplify
18438 setting of opts->x_flag_entry.
18440 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18441 Jakub Jelinek <jakub@redhat.com>
18442 Richard Biener <rguenther@suse.de>
18444 PR tree-optimization/81503
18445 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
18446 folded constant fits in the target type; reorder tests for clarity.
18448 2017-08-29 Martin Liska <mliska@suse.cz>
18450 * passes.def: Include pass_lower_switch.
18451 * stmt.c (dump_case_nodes): Remove and move to
18452 tree-switch-conversion.
18453 (case_values_threshold): Likewise.
18454 (expand_switch_as_decision_tree_p): Likewise.
18455 (emit_case_decision_tree): Likewise.
18456 (expand_case): Likewise.
18457 (balance_case_nodes): Likewise.
18458 (node_has_low_bound): Likewise.
18459 (node_has_high_bound): Likewise.
18460 (node_is_bounded): Likewise.
18461 (emit_case_nodes): Likewise.
18462 (struct simple_case_node): New struct.
18463 (add_case_node): Remove.
18464 (emit_case_dispatch_table): Use vector instead of case_list.
18465 (reset_out_edges_aux): Remove.
18466 (compute_cases_per_edge): Likewise.
18467 (expand_case): Build list of simple_case_node.
18468 (expand_sjlj_dispatch_table): Use it.
18469 * tree-switch-conversion.c (struct case_node): Moved from
18470 stmt.c and adjusted.
18471 (emit_case_nodes): Likewise.
18472 (node_has_low_bound): Likewise.
18473 (node_has_high_bound): Likewise.
18474 (node_is_bounded): Likewise.
18475 (case_values_threshold): Likewise.
18476 (reset_out_edges_aux): Likewise.
18477 (compute_cases_per_edge): Likewise.
18478 (add_case_node): Likewise.
18479 (dump_case_nodes): Likewise.
18480 (balance_case_nodes): Likewise.
18481 (expand_switch_as_decision_tree_p): Likewise.
18482 (emit_jump): Likewise.
18483 (emit_case_decision_tree): Likewise.
18484 (try_switch_expansion): Likewise.
18485 (do_jump_if_equal): Likewise.
18486 (emit_cmp_and_jump_insns): Likewise.
18487 (fix_phi_operands_for_edge): New function.
18488 (record_phi_operand_mapping): Likewise.
18489 (class pass_lower_switch): New pass.
18490 (pass_lower_switch::execute): New function.
18491 (make_pass_lower_switch): Likewise.
18492 (conditional_probability):
18493 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
18494 * tree-pass.h: Add make_pass_lower_switch.
18496 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18499 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
18502 2017-08-29 Richard Biener <rguenther@suse.de>
18504 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
18505 we do not add a DW_AT_inline attribute twice.
18506 (gen_subprogram_die): Remove code setting DW_AT_inline on
18507 DECL_ABSTRACT_P nodes.
18509 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
18511 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
18512 calls to internal functions.
18513 (gimplify_modify_expr): Likewise.
18514 * tree-call-cdce.c (use_internal_fn): Likewise.
18515 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
18516 (convert_to_divmod): Set the nothrow flag.
18517 * tree-if-conv.c (predicate_mem_writes): Likewise.
18518 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
18519 (vectorizable_call): Likewise.
18520 (vectorizable_store): Likewise.
18521 (vectorizable_load): Likewise.
18522 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
18523 (vect_recog_mask_conversion_pattern): Likewise.
18525 2017-08-29 Martin Liska <mliska@suse.cz>
18528 * gcc.c (driver_handle_option): Add new argument.
18529 * opts-common.c (handle_option): Pass
18530 target_option_override_hook.
18531 * opts-global.c (lang_handle_option): Add new option.
18532 (set_default_handlers): Add new argument.
18533 (decode_options): Likewise.
18534 * opts.c (target_handle_option): Likewise.
18535 (common_handle_option): Call target_option_override_hook.
18536 * opts.h (struct cl_option_handler_func): Add hook for
18537 target option override.
18538 (struct cl_option_handlers): Likewise.
18539 (set_default_handlers): Add new argument.
18540 (decode_options): Likewise.
18541 (common_handle_option): Likewise.
18542 (target_handle_option): Likewise.
18543 * toplev.c (toplev::main): Pass targetm.target_option.override
18546 2017-08-29 Richard Biener <rguenther@suse.de>
18547 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
18549 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
18550 life to the active subtree.
18552 2017-08-28 Jeff Law <law@redhat.com>
18554 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
18555 derive_equivalences.
18556 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
18557 Code moved into....
18558 (edge_info::derive_equivalences): New private member function
18560 * tree-ssa-dom.c (class edge_info): Changed from a struct
18561 to a class. Add ctor/dtor, methods and data members.
18562 (edge_info::edge_info): Renamed from allocate_edge_info.
18563 Initialize additional members.
18564 (edge_info::~edge_info): New.
18565 (free_dom_edge_info): Delete the edge info.
18566 (record_edge_info): Use new class & associated member functions.
18567 Tighten forms for testing for edge equivalences.
18568 (record_temporary_equivalences): Iterate over the simple
18569 equivalences rather than assuming there's only one per edge.
18570 (cprop_into_successor_phis): Iterate over the simple
18571 equivalences rather than assuming there's only one per edge.
18572 (optimize_stmt): Use operand_equal_p rather than pointer
18573 equality for mini-DSE code.
18575 2017-08-28 Nathan Sidwell <nathan@acm.org>
18577 * gcc.c (execute): Fold SIGPIPE handling into switch
18578 statement. Adjust internal error message.
18580 2017-08-28 Richard Biener <rguenther@suse.de>
18583 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
18584 Do nothing for removed DIEs.
18586 2017-08-28 Richard Biener <rguenther@suse.de>
18588 PR tree-optimization/81977
18589 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
18592 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
18595 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
18596 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
18597 using targetm.gen_mem_thread_fence.
18599 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
18602 * config/i386/i386.md (*<btsc><mode>): Change operand 2
18603 predicate to register_operand. Reorder operands.
18604 (*btr<mode>): Ditto.
18605 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
18606 (*btr<mode>_mask): Ditto.
18608 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
18610 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
18611 * config/rs6000/xmmintrin.h: New file.
18612 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
18614 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18617 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
18618 parameter and_insn and return it.
18619 (recombine_lvx_pattern): Insert a copy to ensure availability of
18620 the base register of the copied masking operation at the point of
18621 the instruction replacement.
18622 (recombine_stvx_pattern): Likewise.
18624 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
18626 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
18627 undocumented switches.
18628 (-mpower9-dform-vector): Likewise.
18629 (-mpower9-dform): Likewise.
18630 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
18631 comments to delete references to -mpower9-dform* switches.
18632 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
18633 Delete reference to -mpower9-dform* switches, test for
18634 -mpower9-vector instead.
18635 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
18636 (OTHER_P9_VECTOR_MASKS): Likewise.
18637 (POWERPC_MASKS): Likewise.
18638 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
18639 tests against -mpower9-dform* to -mpower9-vector. Delete code
18640 that checked for -mpower9-dform* consistancy with other options.
18641 Add test for -mpower9-misc to enable other power9 switches.
18642 (rs6000_init_hard_regno_mode_ok): Likewise.
18643 (rs6000_option_override_internal): Likewise.
18644 (rs6000_emit_prologue): Likewise.
18645 (rs6000_emit_epilogue): Likewise.
18646 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
18647 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
18648 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
18650 (emit_fusion_p9_store): Likewise.
18651 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
18652 resetting these macros if the assembler does not support ISA 3.0
18654 (TARGET_P9_DFORM_VECTOR): Likewise.
18655 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
18656 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
18658 2017-08-25 Alan Modra <amodra@gmail.com>
18660 PR rtl-optimization/81747
18661 * cse.c (cse_extended_basic_block): Don't attempt to record
18662 equivalences for degenerate conditional branches that branch
18663 to their fall-through.
18665 2017-08-24 Martin Sebor <msebor@redhat.com>
18667 PR middle-end/81908
18668 * gimple-fold.c (size_must_be_zero_p): New function.
18669 (gimple_fold_builtin_memory_op): Call it.
18671 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
18673 * config/rs6000/mm_malloc.h: New file.
18675 2017-08-24 Bin Cheng <bin.cheng@arm.com>
18677 PR tree-optimization/81913
18678 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
18679 analysis when either IVs in condition can wrap.
18681 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
18683 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
18684 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
18686 2017-08-24 Richard Biener <rguenther@suse.de>
18689 * targhooks.c (default_target_can_inline_p): Properly
18690 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
18691 is present and always compare.
18692 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
18693 infer -mfpmath=sse from TARGET_SSE_P.
18694 (ix86_can_inline_p): Properly use target_option_default_node when
18695 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
18697 2017-08-24 Richard Biener <rguenther@suse.de>
18700 * dwarf2out.c (output_die): Handle flag_generate_offload like
18702 (output_comp_unit): Likewise.
18703 (gen_array_type_die): Likewise.
18704 (dwarf2out_early_finish): Likewise.
18705 (note_variable_value_in_expr): Likewise.
18706 (dwarf2out_finish): Likewise. Adjust assert.
18707 * cgraphunit.c (symbol_table::compile): Move setting of
18708 flag_generate_offload earlier ...
18709 (symbol_table::finalize_compilation_unit): ... here, before
18710 early debug finalization.
18712 2017-08-24 Richard Biener <rguenther@suse.de>
18714 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
18715 and ipa-fnsummary.h.
18716 (ix86_can_inline_p): When ix86_fpmath flags do not match
18717 check whether the callee uses FP math at all.
18719 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
18721 PR middle-end/81931
18722 * tree-ssanames.c (get_nonzero_bits): Use element_precision
18723 instead of TYPE_PRECISION.
18725 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
18726 Alan Hayward <alan.hayward@arm.com>
18727 David Sherwood <david.sherwood@arm.com>
18729 * combine.c (make_extraction): Use subreg_offset_from_lsb.
18731 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
18733 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
18735 (ix86_frame::stack_realign_allocate): New field.
18736 (struct machine_frame_state): Modify comments.
18737 (machine_frame_state::sp_realigned_fp_end): New field.
18738 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
18739 layout calculation.
18740 (sp_valid_at): Add assertion to assure no attempt to access invalid
18741 offset of a realigned stack.
18742 (fp_valid_at): Likewise.
18743 (choose_baseaddr): Modify comments.
18744 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
18745 ix86_expand_prologue.
18746 (ix86_expand_prologue): Modify stack realignment and allocation.
18747 (ix86_expand_epilogue): Modify comments.
18748 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
18749 avx2_runtime, avx512f, and avx512f_runtime.
18751 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
18753 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
18754 (mstackrealign): Do not init to -1.
18755 * config/i386/i386.c (ix86_option_override_internal):
18756 Check opts_set, not opts when setting default value of
18757 opts->x_ix86_force_align_arg_pointer.
18759 2017-08-23 Richard Biener <rguenther@suse.de>
18761 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
18762 lang_hooks.decl_printable_name.
18763 * print-rtl-function.c (print_rtx_function): Likewise.
18764 * tree-pretty-print.c (dump_function_header): Likewise.
18766 2017-08-23 Richard Biener <rguenther@suse.de>
18769 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
18770 -g0 at compile-time.
18772 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18774 PR middle-end/19706
18775 * doc/sourcebuild.texi (Other hardware attributes):
18778 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18780 PR middle-end/19706
18781 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
18782 Fix single-use check.
18784 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
18786 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
18788 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18790 * doc/install.texi: Modify to add more details on running selected
18793 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18795 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
18796 is combined with -mabi=ms.
18797 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
18800 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18802 PR tree-optimization/81488
18803 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
18804 and cached_basis fields.
18805 (MAX_SPREAD): New constant.
18806 (alloc_cand_and_find_basis): Initialize new fields.
18807 (clear_visited): New function.
18808 (create_phi_basis_1): Rename from create_phi_basis, set visited
18809 and cached_basis fields.
18810 (create_phi_basis): New wrapper function.
18811 (phi_add_costs_1): Rename from phi_add_costs, add spread
18812 parameter, set visited field, short-circuit when limits reached.
18813 (phi_add_costs): New wrapper function.
18814 (record_phi_increments_1): Rename from record_phi_increments, set
18816 (record_phi_increments): New wrapper function.
18817 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
18818 (phi_incr_cost): New wrapper function.
18819 (all_phi_incrs_profitable_1): Rename from
18820 all_phi_incrs_profitable, set visited field.
18821 (all_phi_incrs_profitable): New wrapper function.
18823 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18824 Alan Hayward <alan.hayward@arm.com>
18825 David Sherwood <david.sherwood@arm.com>
18827 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
18828 that takes the outer and inner modes.
18829 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
18830 comparison as the canonical test for a paradoxical subreg.
18831 * combine.c (simplify_set): Use paradoxical_subreg_p.
18832 (make_extraction): Likewise.
18833 (force_to_mode): Likewise.
18834 (rtx_equal_for_field_assignment_p): Likewise.
18835 (gen_lowpart_for_combine): Likewise.
18836 (simplify_comparison): Likewise.
18837 * cse.c (equiv_constant): Likewise.
18838 * expmed.c (store_bit_field_1): Likewise.
18839 * final.c (alter_subreg): Likewise.
18840 * fwprop.c (propagate_rtx): Likewise.
18841 (forward_propagate_subreg): Likewise.
18842 * ira-conflicts.c (ira_build_conflicts): Likewise.
18843 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
18844 * lra-constraints.c (curr_insn_transform): Likewise.
18845 (split_reg): Likewise.
18846 * lra-eliminations.c (move_plus_up): Likewise.
18847 (lra_eliminate_regs_1): Likewise.
18848 * recog.c (general_operand): Likewise.
18849 * ree.c (combine_reaching_defs): Likewise.
18850 * reload.c (push_reload): Likewise.
18851 (find_reloads): Likewise.
18852 * reload1.c (elimination_effects): Likewise.
18853 (compute_reload_subreg_offset): Likewise.
18854 (choose_reload_regs): Likewise.
18855 * rtlanal.c (subreg_lsb_1): Likewise.
18856 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18857 (simplify_subreg): Likewise.
18858 * var-tracking.c (track_loc_p): Likewise.
18859 * emit-rtl.c (byte_lowpart_offset): Likewise.
18860 (paradoxical_subreg_p): Delete out-of-line definition.
18862 2017-08-22 Jeff Law <law@redhat.com>
18864 PR tree-optimization/81741
18865 PR tree-optimization/71947
18866 * tree-ssa-dom.c: Include tree-inline.h.
18867 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
18868 equivalences if one is more expensive to compute than the other.
18869 * tree-ssa-scopedtables.h (class const_or_copies): Make
18870 record_const_or_copy_raw method private.
18871 (class avail_exprs_stack): New method simplify_binary_operation.
18872 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
18873 avail_exprs_stack::simplify_binary_operation as needed.
18874 (avail_exprs_stack::simplify_binary_operation): New function.
18876 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
18878 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
18879 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
18880 (DOT_SYMBOLS): Likewise.
18881 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
18882 (RELOCATABLE_NEEDS_FIXUP): Likewise.
18883 (RS6000_ABI_NAME): Likewise.
18884 (TARGET_CMODEL): Likewise.
18885 (TOC_SECTION_ASM_OP): Likewise.
18886 (SET_CMODEL): New macro.
18887 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
18889 2017-08-22 Richard Biener <rguenther@suse.de>
18891 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
18892 to allow for free-lang-data replacements similar to verify_type_variant.
18894 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18895 Alan Hayward <alan.hayward@arm.com>
18896 David Sherwood <david.sherwood@arm.com>
18898 * config/aarch64/aarch64.md (casesi): Use DImode rather than
18899 VOIDmode for the LABEL_REF.
18901 2017-08-22 Richard Biener <rguenther@suse.de>
18903 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
18905 2017-08-22 Richard Biener <rguenther@suse.de>
18907 * common.opt (feliminate-dwarf2-dups): Ignore.
18908 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
18909 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
18910 same_die_p_wrap, compute_section_prefix,
18911 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
18912 (comdat_symbol_id, comdat_symbol_number): Likewise.
18913 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
18915 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
18916 (output_die): Mark unreachable path unreachable.
18917 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
18918 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
18919 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
18920 (dwarf2out_early_finish): Likewise.
18922 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
18924 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
18926 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
18929 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
18930 not VAR_P. Filter attribute warnings with OPT_Wattributes.
18931 (avr_attribute_table) <io, io_low, address>: Initialize
18932 .decl_required with true.
18934 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
18936 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
18937 undocumented debugging options.
18938 (-mvsx-scalar-double): Likewise.
18939 (-mallow-df-permute): Likewise.
18940 (-mvectorize-builtins): Likewise.
18941 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18942 (rs6000_builtin_vectorized_function): Likewise.
18943 (rs6000_builtin_md_vectorized_function): Likewise.
18944 (rs6000_opt_vars): Likewise.
18946 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
18949 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
18950 (*btrq_imm): Rename from *btrq.
18951 (*btcq_imm): Rename from *btcq.
18952 (btsc): New code attribute.
18953 (*<btsc><mode>): New insn pattern.
18954 (*btr<mode>): Ditto.
18955 (*<btsc><mode>_mask): New insn_and_split pattern.
18956 (*btr<mode>_mask): Ditto.
18958 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18959 Alan Hayward <alan.hayward@arm.com>
18960 David Sherwood <david.sherwood@arm.com>
18962 * function.c (pad_below): Simplify padding calculation.
18964 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18965 Alan Hayward <alan.hayward@arm.com>
18966 David Sherwood <david.sherwood@arm.com>
18968 * target.def (function_prologue): Remove frame size argument.
18969 (function_epilogue): Likewise.
18970 * doc/tm.texi: Regenerate.
18971 * final.c (final_start_function): Update call to function_prologue.
18972 (final_end_function): Update call to function_epilogue.
18973 (default_function_pro_epilogue): Remove frame size argument.
18974 * output.h (default_function_pro_epilogue): Likewise.
18975 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
18976 (arm_output_function_prologue): Likewise.
18977 * config/frv/frv.c (frv_function_prologue): Likewise.
18978 (frv_function_epilogue): Likewise.
18979 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
18980 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
18981 (ia64_output_function_epilogue): Likewise.
18982 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
18983 (m32r_output_function_epilogue): Likewise.
18984 * config/microblaze/microblaze.c (microblaze_function_prologue)
18985 (microblaze_function_epilogue): Likewise.
18986 * config/mips/mips.c (mips_output_function_prologue): Likewise.
18987 (mips_output_function_epilogue): Likewise.
18988 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
18989 (mmix_target_asm_function_epilogue): Likewise.
18990 * config/msp430/msp430.c (msp430_start_function): Likewise.
18991 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
18992 (nds32_asm_function_epilogue): Likewise.
18993 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
18994 * config/pa/pa.c (pa_output_function_prologue): Likewise.
18995 (pa_output_function_epilogue): Likewise.
18996 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
18997 (rs6000_output_function_epilogue): Likewise.
18998 * config/rl78/rl78.c (rl78_start_function): Likewise.
18999 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
19000 (rs6000_output_function_epilogue): Likewise.
19001 * config/rx/rx.c (rx_output_function_prologue): Likewise.
19002 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
19003 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
19004 (sparc_asm_function_epilogue): Likewise.
19006 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19008 * tree.h (type_has_mode_precision_p): New function.
19009 * convert.c (convert_to_integer_1): Use it.
19010 * expr.c (expand_expr_real_2): Likewise.
19011 (expand_expr_real_1): Likewise.
19012 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
19013 * match.pd: Likewise.
19014 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
19015 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
19016 * tree-tailcall.c (process_assignment): Likewise.
19017 * tree-vect-loop.c (vectorizable_reduction): Likewise.
19018 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
19019 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
19020 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19021 (vectorizable_assignment): Likewise.
19022 (vectorizable_shift): Likewise.
19023 (vectorizable_operation): Likewise.
19024 * tree-vrp.c (register_edge_assert_for_2): Likewise.
19026 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
19028 * match.pd: Add pow (C, x) simplification.
19030 2017-08-21 Richard Biener <rguenther@suse.de>
19032 PR tree-optimization/81900
19033 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
19034 for blocks with abnormal predecessors.
19035 (compute_antic): Do not set visited flag prematurely.
19037 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
19040 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
19042 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19044 * stor-layout.h (vector_type_mode): Move to...
19045 * tree.h (vector_type_mode): ...here.
19046 * stor-layout.c (vector_type_mode): Move to...
19047 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
19049 2017-08-21 Richard Biener <rguenther@suse.de>
19051 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
19052 register_external_die hooks.
19053 (debug_false_tree_charstarstar_uhwistar): Declare.
19054 (debug_nothing_tree_charstar_uhwi): Likewise.
19055 * debug.c (do_nothing_debug_hooks): Adjust.
19056 (debug_false_tree_charstarstar_uhwistar): New do nothing.
19057 (debug_nothing_tree_charstar_uhwi): Likewise.
19058 * dbxout.c (dbx_debug_hooks): Adjust.
19059 (xcoff_debug_hooks): Likewise.
19060 * sdbout.c (sdb_debug_hooks): Likewise.
19061 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
19062 * dwarf2out.c (macinfo_label_base): New global.
19063 (dwarf2out_register_external_die): New function for the
19064 register_external_die hook.
19065 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
19066 (dwarf2_debug_hooks): Use them.
19067 (dwarf2_lineno_debug_hooks): Adjust.
19068 (struct die_struct): Add with_offset flag.
19069 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
19070 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
19071 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
19072 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
19073 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
19074 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
19075 defining section names for the early LTO debug variants.
19076 (reset_indirect_string): New helper.
19077 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
19078 (print_dw_val): Add support for offsetted symbol references.
19079 (get_ultimate_context): Split out from is_cxx.
19080 (is_cxx): Use get_ultimate_context.
19081 (is_fortran): Add decl overload.
19082 (compute_comp_unit_symbol): Split out worker from
19083 compute_section_prefix.
19084 (compute_section_prefix): Call compute_comp_unit_symbol and
19085 set comdat_type_p here.
19086 (output_die): Skip DIE symbol output for the LTO added one.
19087 Handle DIE symbol references with offset.
19088 (output_comp_unit): Guard section name mangling properly.
19089 For LTO debug sections emit a symbol at the section beginning
19090 which we use to refer to its DIEs.
19091 (add_abstract_origin_attribute): For DIEs registered via
19092 dwarf2out_register_external_die directly refer to the early
19093 DIE rather than indirectly through the shadow one we created.
19094 Remove obsolete call to dwarf2out_abstract_function for
19095 non-function/block origins.
19096 (gen_array_type_die): When generating early LTO debug do
19097 not emit DW_AT_string_length.
19098 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
19099 late when in LTO. As suggested place a gcc_unreachable for
19100 the DECL_ABSTRACT_P case.
19101 (gen_subprogram_die): Avoid another specification DIE
19102 for early built declarations/definitions for the late LTO case.
19103 (gen_variable_die): Add type references for late duplicated VLA dies
19105 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
19106 we have the abstract instance already.
19107 (process_scope_var): Adjust decl DIE contexts in LTO which
19108 first puts them in limbo.
19109 (gen_decl_die): Do not generate type DIEs late apart from
19110 types for VLAs or for decls we do not yet have a DIE. Do not
19111 call dwarf2out_abstract_function late.
19112 (dwarf2out_early_global_decl): Make sure to create DIEs
19113 for abstract instances of a decl first.
19114 (dwarf2out_late_global_decl): Adjust comment.
19115 (output_macinfo_op): With multiple macro sections use
19116 macinfo_label_base to distinguish labels.
19117 (output_macinfo): Likewise. Update macinfo_label_base.
19118 Pass in the line info label.
19119 (note_variable_value_in_expr): When generating LTO resolve
19120 all variable values here by generating DIEs as needed.
19121 (init_sections_and_labels): Add early LTO debug flag parameter
19122 and generate different sections and names if set. Add generation
19123 counter for the labels so we can have multiple of them.
19124 (reset_dies): Helper to allow DIEs to be output multiple times.
19125 (dwarf2out_finish): When outputting DIEs to the fat part of an
19126 LTO object first reset DIEs.
19127 (dwarf2out_early_finish): Output early DIEs when generating LTO.
19128 (modified_type_die): Check for decl_ultimate_origin being self
19130 (gen_type_die_with_usage): Likewise.
19131 (gen_typedef_die): Allow decl_ultimate_origin being self.
19132 (set_decl_abstract_flags): Remove.
19133 (set_block_abstract_flags): Likewise.
19134 (dwarf2out_abstract_function): Treat the early generated DIEs
19135 as the abstract copy and only add DW_AT_inline and
19136 DW_AT_artificial here and call set_decl_origin_self.
19137 If the DIE has an abstract origin don't do anything.
19138 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
19139 if we have none yet (Go fails to build one, PR78628).
19140 (variably_modified_type_p): Prevent endless recursion for Ada
19141 cyclic pointer types.
19142 * lto-streamer-in.c: Include debug.h.
19143 (dref_queue): New global.
19144 (lto_read_tree_1): Stream in DIE references.
19145 (lto_input_tree): Register DIE references.
19146 (input_function): Stream DECL_DEBUG_ARGS.
19147 * lto-streamer-out.c: Include debug.h.
19148 (lto_write_tree_1): Output DIE references.
19149 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
19150 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
19151 (output_function): Stream DECL_DEBUG_ARGS.
19152 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
19153 Stream DECL_ABSTRACT_ORIGIN.
19154 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
19155 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
19156 DECL_CONTEXT for file-scope decls.
19157 * lto-streamer.h (struct dref_entry): Declare.
19158 (dref_queue): Likewise.
19159 * cfgexpand.c (pass_expand::execute): Do not call the
19160 outlining_inline_function hook here.
19161 * lto-wrapper.c (debug_obj): New global.
19162 (tool_cleanup): Unlink it if required.
19163 (debug_objcopy): New function.
19164 (run_gcc): Handle early debug sections in the IL files by
19165 extracting them to separate files, partially linkin them and
19166 feeding the result back as result to the linker.
19167 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
19168 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
19169 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
19170 sections into a separate segment.
19171 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
19173 (darwin_asm_dwarf_section): Likewise.
19174 (darwin_asm_output_dwarf_offset): Likewise.
19175 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
19177 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19178 Alan Hayward <alan.hayward@arm.com>
19179 David Sherwood <david.sherwood@arm.com>
19181 * read-md.h (md_reader::record_potential_iterator_use): Replace
19182 pointer argument with an rtx and an index.
19183 * read-rtl.c (iterator_group::apply_iterator): Likewise.
19184 (apply_mode_iterator): Likewise.
19185 (apply_code_iterator): Likewise.
19186 (apply_int_iterator): Likewise.
19187 (apply_subst_iterator): Likewise.
19188 (record_iterator_use): Likewise.
19189 (record_attribute_use): Likewise.
19190 (md_reader::record_potential_iterator_use): Likewise. Update calls
19191 to record_iterator_use and apply_iterator.
19192 (iterator_use): Replace ptr with x and index.
19193 (attribute_use): Likewise.
19194 (apply_attribute_uses): Update calls to apply_iterator.
19195 (apply_iterators): Likewise. Update initialization of iterator_use.
19196 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
19197 and record_potential_iterator_use.
19198 (rtx_reader::read_rtx_operand): Likewise.
19200 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
19201 Alan Hayward <alan.hayward@arm.com>
19202 David Sherwood <david.sherwood@arm.com>
19204 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
19207 2017-08-21 Richard Biener <rguenther@suse.de>
19209 PR middle-end/81884
19210 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
19211 at struct end conservatively when comparing common bases.
19213 2017-08-21 Richard Biener <rguenther@suse.de>
19215 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
19216 (mem_ref_in_stmt): Remove.
19217 (determine_max_movement): Use ref index to get at the reference.
19218 (invariantness_dom_walker::before_dom_children): Deal with
19219 lim data already initialized.
19220 (gather_mem_refs_stmt): Initialize lim data and record ref index.
19222 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19224 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
19225 (TARGET_ISA_ROUND): Ditto.
19226 (TARGET_ROUND): Ditto.
19227 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
19228 * config/i386/i386.md: Ditto.
19229 * config/i386/sse.md: Ditto.
19230 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
19231 with OPTION_MASK_ISA_SSE4_1.
19233 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
19236 * doc/extend.texi (x86 Built-in Functions): Correct the name of
19237 __builtin_ia32_lzcnt_u16.
19239 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
19242 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
19243 (rs6000_set_current_function): Rewrite function to use it.
19245 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
19248 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
19249 and TYPE_WARN_IF_NOT_ALIGN.
19250 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
19251 (handle_warn_if_not_align): New.
19252 (place_union_field): Call handle_warn_if_not_align.
19253 (place_field): Call handle_warn_if_not_align.
19254 Copy TYPE_WARN_IF_NOT_ALIGN.
19255 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
19256 (layout_type): Likewise.
19257 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
19259 (tree_decl_common): Add warn_if_not_align.
19260 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
19261 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
19262 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
19263 (DECL_WARN_IF_NOT_ALIGN): Likewise.
19264 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
19265 * doc/extend.texi: Document warn_if_not_aligned attribute.
19266 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
19268 2017-08-17 Martin Liska <mliska@suse.cz>
19271 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
19272 (get_data_dependence): Use it as pointer type.
19273 (distribute_loop): Likewise.
19275 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19277 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
19278 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
19279 (p8_vmrgow_<mode>_direct): New define_insn.
19280 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
19281 handle endianness for vmrgew and vmrgow permute patterns.
19283 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19285 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
19286 * config/rs6000/rs6000-cpus.def: Remove comment.
19287 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
19288 (POWERPC_MASKS): Likewise.
19289 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
19290 use of TARGET_VSX_TIMODE.
19291 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
19292 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
19293 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
19294 (rs6000_option_override_internal): Remove dead code.
19295 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
19296 (rs6000_legitimize_reload_address): Likewise.
19297 (rs6000_legitimate_address_p): Likewise.
19298 (rs6000_opt_masks): Delete "vsx-timode".
19299 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
19300 from function comment.
19301 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
19302 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
19303 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
19305 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
19306 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
19307 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
19308 (VSX_TI): Likewise.
19310 (define_peephole2): Likewise.
19312 2017-08-17 Martin Sebor <msebor@redhat.com>
19315 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
19316 past the end of an array.
19317 (test_pp_format): Add test cases.
19319 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
19321 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
19322 missing ECF_NOTHROW flags.
19324 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
19327 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
19328 operands residing in integer registers.
19329 (*vsx_le_perm_load_<mode>): Likewise.
19330 (*vsx_le_perm_store_<mode>): Likewise.
19331 (define_peephole2): Add peepholes to optimize the above.
19333 2017-08-17 Marek Polacek <polacek@redhat.com>
19335 PR middle-end/81814
19336 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
19337 to mimic what shorten_compare did. Change the return type to bool.
19338 (fold_cond_expr_with_comparison): Update call to
19339 operand_equal_for_comparison_p.
19340 (fold_ternary_loc): Likewise.
19342 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
19344 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
19346 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
19348 2017-08-17 Richard Biener <rguenther@suse.de>
19350 * tree-ssa-structalias.c (solve_graph): When propagating
19351 to successors update the graphs succ edges and avoid duplicate work.
19353 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
19356 * config/i386/i386.c (ix86_option_override_internal): Save target
19357 specific options after ix86_stack_protector_guard_reg was changed.
19359 2017-08-17 Richard Biener <rguenther@suse.de>
19361 PR tree-optimization/81827
19362 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
19363 (new_var_info): Initialize it conservatively.
19364 (get_call_vi): Mark register vars.
19365 (new_scalar_tmp_constraint_exp): Likewise.
19366 (handle_rhs_call): Likewise.
19367 (handle_const_call): Likewise.
19368 (create_function_info_for): Likewise.
19369 (solve_constraints): Sort varinfos to separate register from
19370 non-register vars to pack points-to solution bitmaps during
19373 2017-08-17 Marek Polacek <polacek@redhat.com>
19375 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
19377 2017-08-17 Richard Biener <rguenther@suse.de>
19379 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
19380 to true when overflow is undefined and we saturated the result.
19382 2017-08-17 Alan Modra <amodra@gmail.com>
19385 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
19386 Don't use store multiple if only one reg needs saving.
19387 (interesting_frame_related_regno): New function.
19388 (rs6000_frame_related): Don't emit frame info for regs that
19390 (rs6000_emit_epilogue): Likewise.
19392 2017-08-16 Nathan Sidwell <nathan@acm.org>
19394 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
19395 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
19396 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
19397 * tree.c (free_lang_data_in_type): Use else-if chain. Always
19398 clear TYPE_LANG_1. Remove obsolete member-function stripping.
19399 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
19400 (verify_type): Adjust for TYPE_BINFO move.
19401 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
19402 process TYPE_BINFO directly.
19403 (hash_tree): Likewise.
19404 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
19406 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
19409 2017-08-16 David Malcolm <dmalcolm@redhat.com>
19411 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
19413 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
19416 * config/i386/i386.md (*anddi_1_btr): Change predicates of
19417 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
19418 Add ix86_binary_operator_ok to insn constraint.
19419 (*iordi_1_bts): Ditto.
19420 (*xordi_1_btc): Ditto.
19421 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
19422 Update corresponding peephole2 pattern.
19426 2017-08-16 Bin Cheng <bin.cheng@arm.com>
19428 PR tree-optimization/81832
19429 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
19430 copy loop header which has IFN_LOOP_DIST_ALIAS call.
19432 2017-08-16 Marek Polacek <polacek@redhat.com>
19435 * fold-const.c (fold_indirect_ref_1): Restore original behavior
19436 regarding size_zero_node.
19438 2017-08-16 Martin Liska <mliska@suse.cz>
19441 * config.gcc: Respect previously set extra_objs in case
19444 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
19446 PR tree-optimization/81835
19447 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
19448 the phi SSA_NAME. Check that the condition in a COND_EXPR does
19449 not depend on the phi.
19451 2017-08-16 Alan Modra <amodra@gmail.com>
19453 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
19456 2017-08-16 Alan Modra <amodra@gmail.com>
19458 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
19459 (save_reg_p): ..into this. Update all callers.
19460 (first_reg_to_save): Simplify.
19462 2017-08-16 Alan Modra <amodra@gmail.com>
19464 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
19467 2017-08-15 Joseph Myers <joseph@codesourcery.com>
19471 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
19472 constant count if that count is less than 32.
19474 2017-08-15 Nathan Sidwell <nathan@acm.org>
19476 * gcc.c (execute): Emit friendlier message if inferior is killed
19477 by an external cause.
19479 2017-08-15 Richard Biener <rguenther@suse.de>
19481 PR tree-optimization/81790
19482 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
19483 CONSTRUCTORs from simplifying and VN.
19485 2017-08-14 Martin Sebor <msebor@redhat.com>
19487 * builtin-attrs.def: Add comments.
19489 2017-08-14 Martin Sebor <msebor@redhat.com>
19492 * doc/extend.texi (attribute nonstring): Document new attribute.
19494 2017-08-14 Martin Sebor <msebor@redhat.com>
19497 * tree-diagnostic.c (default_tree_printer): Handle %G.
19498 * gimple-pretty-print.h (percent_G_format): Declare new function.
19499 * gimple-pretty-print.c (percent_G_format): Define.
19500 * tree-pretty-print.c (percent_K_format): Add argument.
19502 2017-08-14 Martin Sebor <msebor@redhat.com>
19504 PR translation/79998
19505 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
19506 Remove a stray space.
19508 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
19511 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
19512 (*iordi_1_bts): Ditto.
19513 (*xordi_1_btc): Ditto.
19515 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19518 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
19519 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19521 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
19522 quoted strings, and make more translator-friendly.
19523 (darwin_rs6000_override_options): Likewise.
19524 (rs6000_option_override_internal): Likewise.
19525 (rs6000_return_in_memory): Fix overlong line.
19526 (init_cmulative_args): Use quoted strings, and make more
19527 translator-friendly.
19528 (rs6000_pass_by_reference): Fix overlong line.
19529 (def_builtin): Use quoted strings.
19530 (altivec_expand_predicate_builtin): Use quoted strings, and make
19531 more translator-friendly.
19532 (htm_expand_builtin): Use quoted strings.
19533 (cpu_expand_builtin): Use quoted strings, and make more
19534 translator-friendly.
19535 (altivec_expand_builtin): Likewise.
19536 (paired_expand_predicate_builtin): Likewise.
19537 (rs6000_invalid_builtin): Likewise.
19538 (builtin_function_type): Use quoted strings.
19539 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
19540 more translator-friendly.
19541 (rs6000_trampoline_init): Likewise.
19542 (rs6000_handle_altivec_attribute): Likewise.
19543 (rs6000_inner_target_options): Use quoted strings.
19544 (rs6000_disable_incompatible_switches): Likewise.
19545 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
19546 strings, and make more translator-friendly.
19547 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
19549 2017-08-14 Bin Cheng <bin.cheng@arm.com>
19551 PR tree-optimization/81799
19552 * tree-loop-distribution.c (version_loop_by_alias_check): Force
19553 cond_expr to simple gimple operand.
19555 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
19557 PR middle-end/46932
19558 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
19560 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
19564 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
19565 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
19566 TARGET_GASISR_PROLOGUES.
19567 * config/avr/avr.c (avr_option_override): Same.
19568 (avr_pass_pre_proep::execute): Same.
19570 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
19573 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
19574 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
19576 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
19578 * config/i386/i386.md (*load_tp_<mode>): Redefine as
19579 define_insn_and_split. Split to a memory load from 0 in
19580 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
19581 using PTR mode iterator.
19582 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
19583 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
19584 (*add_tp_<mode>): Redefine as define_insn_and_split.
19585 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
19586 address space. Merge with *add_tp_x32 using PTR mode iterator.
19587 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
19588 Split to an add with a memory load from 0 in
19589 DEFAULT_TLS_SEG_REG address space.
19591 2017-08-12 Andrew Pinski <apinski@cavium.com>
19593 * config/aarch64/aarch64-option-extensions.def (rdma):
19594 Fix feature string to what Linux prints out in /proc/cpuinfo.
19596 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
19599 * dwarf2out.c (modified_type_die): For C typedef types that have
19600 an ultimate origin, process the ultimate origin instead of the
19602 (gen_typedef_die): Assert that input DECLs have no ultimate
19604 (gen_type_die_with_usage): For typedef variants that have an
19605 ultimate origin, just call gen_decl_die on the original DECL.
19606 (process_scope_var): Avoid creating DIEs for local typedefs and
19607 concrete static variables.
19609 2017-08-12 Alan Modra <amodra@gmail.com>
19613 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
19614 match gnu-user.h startfile.
19615 (ENDFILE_LINUX_SPEC): Similarly.
19617 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
19620 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
19622 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
19624 2017-08-11 Tamar Christina <tamar.christina@arm.com>
19625 * config/aarch64/aarch64.md (mov<mode>): Change.
19626 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
19627 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
19628 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
19630 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
19632 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
19633 for storage order barriers.
19635 2017-08-11 Martin Liska <mliska@suse.cz>
19638 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
19639 variables of void type.
19641 2017-08-11 Martin Liska <mliska@suse.cz>
19643 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
19644 TARGET_SUPPORTS_ALIASES.
19645 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
19646 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
19647 (optimize_weakref): Likewise.
19648 * symtab.c (symtab_node::noninterposable_alias): Likewise.
19649 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
19650 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
19652 2017-08-11 Martin Liska <mliska@suse.cz>
19655 * config/i386/i386.c (make_resolver_func): Do complete
19656 refactoring of the function.
19658 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
19661 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
19662 * config/i386/i386.c (ix86_stack_protect_guard): Use
19663 ix86_stack_protect_guard_symbol_str to generate varible declaration.
19664 * doc/invoke.texi (x86 Options): Document
19665 -mstack-protector-guard-symbol= option.
19667 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
19669 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
19670 * config/i386/i386.c (ix86_split_stack_guard): New function.
19671 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
19672 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
19673 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
19674 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
19675 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
19676 (split_stack_space_check): Call ix86_split_stack_guard.
19678 2017-08-10 Martin Sebor <msebor@redhat.com>
19680 * print-tree.c (print_node): Print location using the established
19682 Replace spaces with colons.
19683 (debug_raw, debug): Ditto.
19685 2017-08-10 Martin Sebor <msebor@redhat.com>
19688 * pretty-print.c (pp_format): Correct the handling of %s precision.
19690 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
19693 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
19695 (ix86_finalize_stack_frame_flags): This. Also clear
19696 frame_pointer_needed if -fno-omit-frame-pointer is used without
19698 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
19699 with ix86_finalize_stack_frame_flags.
19700 (ix86_expand_epilogue): Likewise.
19701 (ix86_expand_split_stack_prologue): Likewise.
19702 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
19704 2017-08-10 Martin Liska <mliska@suse.cz>
19707 * c-attribs.c (handle_target_attribute):
19708 Report warning for an empty string argument of target attribute.
19710 2017-08-09 Jakub Jelinek <jakub@redhat.com>
19713 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
19715 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
19716 or DECL_NONLOCAL labels.
19717 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
19718 or DECL_NONLOCAL labels here.
19720 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
19722 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
19723 to indicate when early gimple folding has been disabled.
19724 (rs6000_gimple_fold_builtin): Add debug content.
19725 (rs6000_invalid_builtin): Fix whitespace.
19726 (rs6000_expand_builtin): Fix whitespace.
19727 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
19729 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
19732 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
19733 SAVE_MULTIPLE if not all the registers that saves, should be saved.
19735 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
19737 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
19738 (qdf24xx): Likewise.
19739 * config/aarch64/aarch64.md: Include falkor.md.
19740 * config/aarch64/falkor.md: New.
19742 2017-08-09 Marek Polacek <polacek@redhat.com>
19745 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
19746 * diagnostic.c (emit_diagnostic): Add a comment.
19747 (emit_diagnostic_valist): New function.
19749 2017-08-09 Marek Polacek <polacek@redhat.com>
19752 * input.c (make_location): New overload.
19753 * input.h (make_location): Declare.
19755 2017-08-08 Alan Modra <amodra@gmail.com>
19756 H.J. Lu <hongjiu.lu@intel.com>
19759 * gcc.c (NO_PIE_SPEC): Delete.
19760 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
19762 (LINK_PIE_SPEC): ..to here.
19763 (LINK_COMMAND_SPEC): Support -no-pie.
19764 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
19765 chain of crtbegin*.o selection, update for PIE_SPEC changes and
19767 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
19768 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
19769 (ENDFILE_CRTEND_SPEC): Similarly.
19771 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
19774 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
19775 (mstack-protector-guard-offset=): Ditto.
19776 * config/i386/i386.c (ix86_option_override): Handle
19777 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
19779 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
19780 ix86_stack_protect_guard_offset variables.
19781 (TARGET_STACK_PROTECT_GUARD): Always define.
19782 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
19783 and -mstack-protector-guard-offset= options.
19785 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19787 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
19788 boundary case for the last candidate.
19790 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19792 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
19793 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
19795 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19797 PR middle-end/19706
19798 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
19799 * config/aarch64/aarch64-builtins.c
19800 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
19801 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
19802 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
19804 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19805 Andrew Pinski <pinskia@gmail.com>
19807 PR middle-end/19706
19808 * internal-fn.def (XORSIGN): New.
19809 * optabs.def (xorsign_optab): New.
19810 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
19811 (convert_expand_mult_copysign): New.
19812 (pass_optimize_widening_mul::execute): Call
19813 convert_expand_mult_copysign.
19815 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19817 PR tree-optimization/81354
19818 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
19819 Insert on edges rather than explicitly creating landing pads.
19820 (analyze_candidates_and_replace): Commit edge inserts.
19822 2017-08-08 Richard Biener <rguenther@suse.de>
19824 PR middle-end/81719
19825 * tree-ssa-loop-niter.c: Include tree-dfa.h.
19826 (expand_simple_operations): Also look through ADDR_EXPRs with
19827 MEM_REF bases treating them as POINTER_PLUS_EXPR.
19829 2017-08-08 Richard Biener <rguenther@suse.de>
19831 PR tree-optimization/81723
19832 * tree-vect-slp.c (struct bst_traits): New hash traits.
19833 (bst_fail): New global.
19834 (vect_build_slp_tree_2): New worker, split out from ...
19835 (vect_build_slp_tree): ... this now wrapping it with using
19836 bst_fail set to cache SLP tree build fails. Properly handle
19838 (vect_analyze_slp_instance): Allocate and free bst_fail.
19840 2017-08-08 Martin Liska <mliska@suse.cz>
19843 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
19844 LABEL_DECLs that can be from a different function.
19846 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19848 PR tree-optimization/81744
19849 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
19850 loop's number of iterations.
19852 2017-08-08 Martin Liska <mliska@suse.cz>
19854 * asan.c: Include header files.
19855 * attribs.c (build_decl_attribute_variant): New function moved
19857 (build_type_attribute_qual_variant): Likewise.
19858 (cmp_attrib_identifiers): Likewise.
19859 (simple_cst_list_equal): Likewise.
19860 (omp_declare_simd_clauses_equal): Likewise.
19861 (attribute_value_equal): Likewise.
19862 (comp_type_attributes): Likewise.
19863 (build_type_attribute_variant): Likewise.
19864 (lookup_ident_attribute): Likewise.
19865 (remove_attribute): Likewise.
19866 (merge_attributes): Likewise.
19867 (merge_type_attributes): Likewise.
19868 (merge_decl_attributes): Likewise.
19869 (merge_dllimport_decl_attributes): Likewise.
19870 (handle_dll_attribute): Likewise.
19871 (attribute_list_equal): Likewise.
19872 (attribute_list_contained): Likewise.
19873 * attribs.h (lookup_attribute): New function moved from tree.[ch].
19874 (lookup_attribute_by_prefix): Likewise.
19875 * bb-reorder.c: Include header files.
19876 * builtins.c: Likewise.
19877 * calls.c: Likewise.
19878 * cfgexpand.c: Likewise.
19879 * cgraph.c: Likewise.
19880 * cgraphunit.c: Likewise.
19881 * convert.c: Likewise.
19882 * dwarf2out.c: Likewise.
19883 * final.c: Likewise.
19884 * fold-const.c: Likewise.
19885 * function.c: Likewise.
19886 * gimple-expr.c: Likewise.
19887 * gimple-fold.c: Likewise.
19888 * gimple-pretty-print.c: Likewise.
19889 * gimple.c: Likewise.
19890 * gimplify.c: Likewise.
19891 * hsa-common.c: Likewise.
19892 * hsa-gen.c: Likewise.
19893 * internal-fn.c: Likewise.
19894 * ipa-chkp.c: Likewise.
19895 * ipa-cp.c: Likewise.
19896 * ipa-devirt.c: Likewise.
19897 * ipa-fnsummary.c: Likewise.
19898 * ipa-inline.c: Likewise.
19899 * ipa-visibility.c: Likewise.
19901 * lto-cgraph.c: Likewise.
19902 * omp-expand.c: Likewise.
19903 * omp-general.c: Likewise.
19904 * omp-low.c: Likewise.
19905 * omp-offload.c: Likewise.
19906 * omp-simd-clone.c: Likewise.
19907 * opts-global.c: Likewise.
19908 * passes.c: Likewise.
19909 * predict.c: Likewise.
19910 * sancov.c: Likewise.
19911 * sanopt.c: Likewise.
19912 * symtab.c: Likewise.
19913 * toplev.c: Likewise.
19914 * trans-mem.c: Likewise.
19915 * tree-chkp.c: Likewise.
19916 * tree-eh.c: Likewise.
19917 * tree-into-ssa.c: Likewise.
19918 * tree-object-size.c: Likewise.
19919 * tree-parloops.c: Likewise.
19920 * tree-profile.c: Likewise.
19921 * tree-ssa-ccp.c: Likewise.
19922 * tree-ssa-live.c: Likewise.
19923 * tree-ssa-loop.c: Likewise.
19924 * tree-ssa-sccvn.c: Likewise.
19925 * tree-ssa-structalias.c: Likewise.
19926 * tree-ssa.c: Likewise.
19927 * tree-streamer-in.c: Likewise.
19928 * tree-vectorizer.c: Likewise.
19929 * tree-vrp.c: Likewise.
19930 * tsan.c: Likewise.
19931 * ubsan.c: Likewise.
19932 * varasm.c: Likewise.
19933 * varpool.c: Likewise.
19934 * tree.c: Remove functions moved to attribs.[ch].
19935 * tree.h: Likewise.
19936 * config/aarch64/aarch64.c: Add attrs.h header file.
19937 * config/alpha/alpha.c: Likewise.
19938 * config/arc/arc.c: Likewise.
19939 * config/arm/arm.c: Likewise.
19940 * config/avr/avr.c: Likewise.
19941 * config/bfin/bfin.c: Likewise.
19942 * config/c6x/c6x.c: Likewise.
19943 * config/cr16/cr16.c: Likewise.
19944 * config/cris/cris.c: Likewise.
19945 * config/darwin.c: Likewise.
19946 * config/epiphany/epiphany.c: Likewise.
19947 * config/fr30/fr30.c: Likewise.
19948 * config/frv/frv.c: Likewise.
19949 * config/ft32/ft32.c: Likewise.
19950 * config/h8300/h8300.c: Likewise.
19951 * config/i386/winnt.c: Likewise.
19952 * config/ia64/ia64.c: Likewise.
19953 * config/iq2000/iq2000.c: Likewise.
19954 * config/lm32/lm32.c: Likewise.
19955 * config/m32c/m32c.c: Likewise.
19956 * config/m32r/m32r.c: Likewise.
19957 * config/m68k/m68k.c: Likewise.
19958 * config/mcore/mcore.c: Likewise.
19959 * config/microblaze/microblaze.c: Likewise.
19960 * config/mips/mips.c: Likewise.
19961 * config/mmix/mmix.c: Likewise.
19962 * config/mn10300/mn10300.c: Likewise.
19963 * config/moxie/moxie.c: Likewise.
19964 * config/msp430/msp430.c: Likewise.
19965 * config/nds32/nds32-isr.c: Likewise.
19966 * config/nds32/nds32.c: Likewise.
19967 * config/nios2/nios2.c: Likewise.
19968 * config/nvptx/nvptx.c: Likewise.
19969 * config/pa/pa.c: Likewise.
19970 * config/pdp11/pdp11.c: Likewise.
19971 * config/powerpcspe/powerpcspe.c: Likewise.
19972 * config/riscv/riscv.c: Likewise.
19973 * config/rl78/rl78.c: Likewise.
19974 * config/rx/rx.c: Likewise.
19975 * config/s390/s390.c: Likewise.
19976 * config/sh/sh.c: Likewise.
19977 * config/sol2.c: Likewise.
19978 * config/sparc/sparc.c: Likewise.
19979 * config/spu/spu.c: Likewise.
19980 * config/stormy16/stormy16.c: Likewise.
19981 * config/tilegx/tilegx.c: Likewise.
19982 * config/tilepro/tilepro.c: Likewise.
19983 * config/v850/v850.c: Likewise.
19984 * config/vax/vax.c: Likewise.
19985 * config/visium/visium.c: Likewise.
19986 * config/xtensa/xtensa.c: Likewise.
19988 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19991 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
19992 constraints since the -mupper-regs-* switches have been
19994 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
19995 into a vector from a double word element that was extracted from
19996 another vector, and eliminate extra XXPERMDI instructions.
19997 (vsx_concat_<mode>_2): Likewise.
19998 (vsx_concat_<mode>_3): Likewise.
19999 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
20000 concat to allow optimizing inserts from previous extracts.
20002 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
20004 * config/i386/i386.c (ix86_stack_protect_guard): Generate
20005 memory reference to a SSP offset in TLS address space.
20006 (ix86_print_operand) <case '@'>: Remove.
20007 (ix86_print_operand_punct_valid_p): Remove '@' code.
20008 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
20009 UNSPEC_SP_TLS_TEST.
20010 (stack_tls_protect_set_<mode>): Remove.
20011 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
20012 (stack_tls_protect_test_<mode>): Remove.
20013 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
20015 2017-08-07 Olivier Hainque <hainque@adacore.com>
20018 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
20020 2017-08-07 Douglas Rupp <rupp@adacore.com>
20022 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
20023 variable was referenced as multidir in command.
20025 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20028 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
20031 2017-08-07 Martin Liska <mliska@suse.cz>
20033 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
20034 * config/rl78/rl78.c: Add include of attribs.h.
20035 * config/sh/sh.c: Likewise.
20036 * config/v850/v850.c: Likewise.
20038 2017-08-07 Tom de Vries <tom@codesourcery.com>
20040 PR middle-end/78266
20041 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
20043 2017-08-07 Martin Liska <mliska@suse.cz>
20045 * config/mips/mips.c: Include attribs.h.
20047 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
20050 * doc/invoke.texi: Document change in behvaior for -Ofast for
20053 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
20055 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
20057 (aarch64_pop_regs): Likewise.
20058 (aarch64_gen_load_pair): Likewise.
20059 (aarch64_save_callee_saves): Likewise.
20060 (aarch64_restore_callee_saves): Likewise.
20062 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20064 * config/i386/i386.c: Revert the last change.
20066 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20069 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
20071 (ix86_finalize_stack_frame_flags): This. Also clear
20072 frame_pointer_needed if -fno-omit-frame-pointer is used without
20074 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
20075 with ix86_finalize_stack_frame_flags.
20076 (ix86_expand_epilogue): Likewise.
20077 (ix86_expand_split_stack_prologue): Likewise.
20079 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
20082 * config/i386/i386.c (get_builtin_code_for_version): Set priority
20083 to P_AES for Westmere.
20085 2017-08-07 Jonathan Yong <10walls@gmail.com>
20087 * config/i386/mingw.opt (fset-stack-executable): Removed.
20088 * config/i386/cygming.opt (fset-stack-executable): Moved
20090 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
20092 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
20094 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
20096 2017-08-07 Marek Polacek <polacek@redhat.com>
20098 PR middle-end/81737
20099 * fold-const.c (fold_indirect_ref_1): Check type_domain.
20101 2017-08-07 Martin Liska <mliska@suse.cz>
20103 * attribs.h (canonicalize_attr_name): New function.
20104 (cmp_attribs): Move from c-format.c and adjusted.
20105 (is_attribute_p): Moved from tree.h.
20106 * tree-inline.c: Add new includes.
20107 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
20108 (private_is_attribute_p): Remove.
20109 (private_lookup_attribute): Likewise.
20110 (private_lookup_attribute_by_prefix): Simplify.
20111 (remove_attribute): Use is_attribute_p.
20112 * tree.h: Remove removed declarations.
20114 2017-08-07 Jakub Jelinek <jakub@redhat.com>
20116 PR middle-end/81698
20117 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
20118 instead of computing it in the function. Formatting fix.
20119 (expand_case): Don't rely on default_edge being the first edge,
20120 clear it if removing it, pass default_edge to
20121 emit_case_dispatch_table.
20122 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
20125 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
20127 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
20128 insn in the function, emit NOP after the insn.
20130 2017-08-06 Tom de Vries <tom@codesourcery.com>
20132 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
20135 2017-08-06 Tom de Vries <tom@codesourcery.com>
20137 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
20140 2017-08-04 Yury Gribov <tetra2005@gmail.com>
20142 PR tree-optimization/57371
20143 * match.pd: New pattern.
20145 2017-08-04 Marek Polacek <polacek@redhat.com>
20147 PR middle-end/81695
20148 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
20149 perform the computation in offset_int.
20151 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20153 PR tree-optimization/81136
20154 * tree-vectorizer.h: Include tree-hash-traits.h.
20155 (vec_base_alignments): New typedef.
20156 (vec_info): Add a base_alignments field.
20157 (vect_record_base_alignments): Declare.
20158 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
20160 (DR_IS_CONDITIONAL_IN_STMT): New macro.
20161 (create_data_ref): Add an is_conditional_in_stmt argument.
20162 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
20163 the is_conditional_in_stmt field.
20164 (data_ref_loc): Add an is_conditional_in_stmt field.
20165 (get_references_in_stmt): Set the is_conditional_in_stmt field.
20166 (find_data_references_in_stmt): Update call to create_data_ref.
20167 (graphite_find_data_references_in_stmt): Likewise.
20168 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
20169 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
20170 (vect_record_base_alignment): New function.
20171 (vect_record_base_alignments): Likewise.
20172 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
20173 for nested statements even if we fail to compute a misalignment.
20174 Use pooled base alignments for unconditional references.
20175 (vect_find_same_alignment_drs): Compare base addresses instead
20177 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
20178 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
20180 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20182 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
20183 Add an explicit name for the enum. Use auto_vec for slp_instances
20184 and grouped_stores.
20185 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
20187 (_bb_vec_info): Add a constructor and destructor.
20188 (vinfo_for_stmt): Return NULL for uids of -1 as well.
20189 (destroy_loop_vec_info): Delete.
20190 (vect_destroy_datarefs): Likewise.
20191 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
20192 (vec_info::vec_info): New function.
20193 (vec_info::~vec_info): Likewise.
20194 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
20195 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
20196 destroy_loop_vec_info.
20197 * tree-vect-loop.c (new_loop_vec_info): Replace with...
20198 (_loop_vec_info::_loop_vec_info): ...this.
20199 (destroy_loop_vec_info): Replace with...
20200 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
20201 the stmt_vec_infos. Leave handling of vec_info information to its
20202 destructor. Remove explicit vector releases.
20203 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
20204 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
20205 * tree-vect-slp.c (new_bb_vec_info): Replace with...
20206 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
20207 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
20208 (destroy_bb_vec_info): Replace with...
20209 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
20210 information to its destructor.
20211 (vect_slp_analyze_bb_1): Use new and delete instead of
20212 new_bb_vec_info and destroy_bb_vec_info.
20213 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
20216 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
20218 * tree-data-ref.h (subscript): Add access_fn field.
20219 (data_dependence_relation): Add could_be_independent_p.
20220 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
20221 (same_access_functions): Move to tree-data-ref.c.
20222 * tree-data-ref.c (ref_contains_union_access_p): New function.
20223 (access_fn_component_p): Likewise.
20224 (access_fn_components_comparable_p): Likewise.
20225 (dr_analyze_indices): Add a reference to access_fn_component_p.
20226 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
20228 (constant_access_functions): Likewise.
20229 (add_other_self_distances): Likewise.
20230 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
20231 (initialize_data_dependence_relation): Use XCNEW and remove
20232 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
20233 of access functions that have the same type. Allow the
20234 subsequence to end with different bases in some circumstances.
20235 Record the chosen access functions in SUB_ACCESS_FN.
20236 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
20237 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
20238 (subscript_dependence_tester_1): Likewise dra and drb.
20239 (build_classic_dist_vector): Update calls accordingly.
20240 (subscript_dependence_tester): Likewise.
20241 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
20242 DDR_COULD_BE_INDEPENDENT_P.
20243 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
20244 comp_alias_ddrs instead of may_alias_ddrs.
20245 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
20247 (vect_analyze_data_ref_dependence): Use it if
20248 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
20249 distance vectors if that fails.
20250 (dependence_distance_ge_vf): New function.
20251 (vect_prune_runtime_alias_test_list): Use it. Don't clear
20252 LOOP_VINFO_MAY_ALIAS_DDRS.
20254 2017-08-04 Richard Biener <rguenther@suse.de>
20256 PR middle-end/81705
20257 * fold-const.c (fold_binary_loc): Properly restrict
20258 minus_var0 && minus_var1 case when associating undefined overflow
20261 2017-08-04 Tom de Vries <tom@codesourcery.com>
20263 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
20265 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20267 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20268 Don't start diagnostic messages with a capital letter.
20269 * config/rs6000/rs6000.c (rs6000_option_override_internal):
20271 (rs6000_invalid_builtin): Likewise.
20272 (rs6000_trampoline_init): Likewise.
20274 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20277 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
20278 after setting changeable df flags.
20280 2017-08-03 Richard Biener <rguenther@suse.de>
20282 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
20283 up if the use is in USE - X.
20285 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
20287 * toplev.c (dumpfile.h): New include.
20288 (internal_error_reentered): New static function. Use it...
20289 (internal_error_function): ...here to handle reentered internal_error.
20291 2017-08-03 Richard Biener <rguenther@suse.de>
20293 PR middle-end/81148
20294 * fold-const.c (split_tree): Add minus_var and minus_con
20295 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
20296 here but always use minus_*.
20297 (associate_trees): Assert we never associate with MINUS_EXPR
20298 and NULL first operand. Do not recurse for PLUS_EXPR operands
20299 when associating as MINUS_EXPR either.
20300 (fold_binary_loc): Track minus_var and minus_con.
20302 2017-08-03 Tom de Vries <tom@codesourcery.com>
20305 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
20306 ACCEL_COMPILER, apply finish_options on
20307 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
20309 2017-08-03 Tom de Vries <tom@codesourcery.com>
20312 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
20313 function_entry_patch_area_size > 0.
20315 2017-08-03 Jakub Jelinek <jakub@redhat.com>
20318 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
20319 instead of 10??LU, perform unit multiplication in wide_int,
20320 don't change alloc_object_size_limit if the limit is larger
20323 PR tree-optimization/81655
20324 PR tree-optimization/81588
20325 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
20326 the case when ranges[i].low and high are 1 for unsigned type with
20329 PR middle-end/81052
20330 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
20331 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
20333 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20335 * tree-vrp.h: Add include guard.
20337 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
20340 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
20341 (ud2): New insn pattern.
20342 * config/i386/i386.c (ix86_expand_epilogue):
20343 For naked functions, generate ud2 instead of trap insn.
20345 2017-08-02 Marek Polacek <polacek@redhat.com>
20348 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
20350 2017-08-02 Tom de Vries <tom@codesourcery.com>
20351 Cesar Philippidis <cesar@codesourcery.com>
20353 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
20354 Add missing edge probabilities.
20356 2017-08-02 Tamar Christina <tamar.christina@arm.com>
20358 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
20359 Correct endianness.
20361 2017-08-02 Jakub Jelinek <jakub@redhat.com>
20363 PR middle-end/79499
20364 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
20365 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
20366 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
20367 prologue_seq sequences - if any.
20369 2017-08-02 Richard Biener <rguenther@suse.de>
20371 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
20372 via vectors if supported, integer extracts via punning if supported
20373 or otherwise vector extracts.
20375 2017-08-02 Richard Biener <rguenther@suse.de>
20377 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
20379 (bitmap_insert_into_set): ... this.
20381 2017-08-02 Richard Biener <rguenther@suse.de>
20383 PR tree-optimization/81633
20385 2015-08-17 Alan Hayward <alan.hayward@arm.com>
20387 PR tree-optimization/71752
20388 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
20390 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
20392 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
20393 (machine_function::call_ms2sysv_pad_out): Remove field.
20394 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
20395 (ix86_compute_frame_layout): Likewise.
20397 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
20400 * config/i386/i386.c (ix86_set_func_type): Disallow naked
20401 attribute with interrupt attribute.
20403 2017-08-01 Andrew Pinski <apinski@cavium.com>
20405 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
20406 BIT_INSERT_EXPR's operand 1
20407 to see if the types precision matches.
20409 2017-08-01 Martin Liska <mliska@suse.cz>
20411 PR middle-end/70140
20412 * builtins.c (expand_builtin_memcpy_args): Remove.
20413 (expand_builtin_memcpy): Call newly added function
20414 expand_builtin_memory_copy_args.
20415 (expand_builtin_memcpy_with_bounds): Likewise.
20416 (expand_builtin_mempcpy): Remove last argument.
20417 (expand_builtin_mempcpy_with_bounds): Likewise.
20418 (expand_builtin_memory_copy_args): New function created from
20419 expand_builtin_mempcpy_args with small modifications.
20420 (expand_builtin_mempcpy_args): Remove.
20421 (expand_builtin_stpcpy): Remove unused argument.
20422 (expand_builtin): Likewise.
20423 (expand_builtin_with_bounds): Likewise.
20425 2017-08-01 Martin Liska <mliska@suse.cz>
20428 Make mempcpy more optimal (PR middle-end/70140).
20430 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20433 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
20434 __builtin_vec_cmpne verify both arguments are compatible vectors
20435 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
20436 verify arg1_type is a pointer or array type. For __builtin_vec_st,
20437 move computation of aligned to after checking the argument types.
20441 * config/rs6000/vsx.md (vextract_fp_from_shorth,
20442 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
20445 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
20446 Doug Rupp <rupp@adacore.com>
20447 Olivier Hainque <hainque@adacore.com>
20449 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
20450 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
20452 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
20453 for TARGET_OS_CPP_BUILTIN.
20454 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
20455 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
20457 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
20458 passing required abi options to the assembler for EABI configurations.
20459 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
20460 of .text.hot and .text.unlikely sections for kernel modules when
20461 using ARM style exceptions.
20462 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
20463 options. Add EXTRA_CC1_SPEC.
20464 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
20466 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
20468 (ARM_TARGET2_DWARF_FORMAT): Define.
20469 * config/arm/t-vxworks: Adjust multilib control to removal of the
20470 Diab command line options.
20472 2017-08-01 Martin Liska <mliska@suse.cz>
20474 PR gcov-profile/81561
20475 * gcov.c (unblock): Make unblocking safe as we need to preserve
20476 index correspondence of blocks and block_lists.
20478 2017-08-01 Richard Biener <rguenther@suse.de>
20480 PR tree-optimization/81181
20481 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
20482 (compute_antic): ... end of iteration here.
20484 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
20486 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
20487 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
20488 (ftree-slp-vectorize): Likewise.
20489 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
20490 can no longer be set independent of flag_tree_loop_vectorize.
20491 * omp-general.c (emp_max_vf): Likewise.
20492 * opts.c (enable_fdo_optimizations): Remove references to
20493 flag_tree_vectorize, these are now implicit.
20494 (common_handle_option): Remove handling for OPT_ftree_vectorize,
20495 and leave it for the options machinery.
20497 2017-08-01 Martin Liska <mliska@suse.cz>
20499 PR middle-end/70140
20500 * builtins.c (expand_builtin_memcpy_args): Remove.
20501 (expand_builtin_memcpy): Call newly added function
20502 expand_builtin_memory_copy_args.
20503 (expand_builtin_memcpy_with_bounds): Likewise.
20504 (expand_builtin_mempcpy): Remove last argument.
20505 (expand_builtin_mempcpy_with_bounds): Likewise.
20506 (expand_builtin_memory_copy_args): New function created from
20507 expand_builtin_mempcpy_args with small modifications.
20508 (expand_builtin_mempcpy_args): Remove.
20509 (expand_builtin_stpcpy): Remove unused argument.
20510 (expand_builtin): Likewise.
20511 (expand_builtin_with_bounds): Likewise.
20513 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20516 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
20517 print "ds:" only for immediates in generic address space.
20519 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
20522 * config/i386/i386.c (ix86_funciton_naked): New prototype.
20523 (ix86_function_ok_for_sibcall): Return false for naked functions.
20525 2017-08-01 Richard Biener <rguenther@suse.de>
20527 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
20528 (compute_antic): Seed worklist with exit block predecessors.
20529 * cfganal.c (dfs_find_deadend): For a cycle return the source
20530 of the edge closing it.
20532 2017-08-01 Tamar Christina <tamar.christina@arm.com>
20534 * config/aarch64/aarch64.c
20535 (aarch64_can_const_movi_rtx_p): Move 0 check.
20537 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20539 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
20540 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
20542 * match.pd: Ditto in address comparison pattern.
20544 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20546 PR tree-optimization/81627
20547 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
20548 closed ssa form for store-store chain.
20550 2017-08-01 Bin Cheng <bin.cheng@arm.com>
20552 PR tree-optimization/81620
20553 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
20554 for store-store chain.
20556 2017-08-01 Jakub Jelinek <jakub@redhat.com>
20558 PR tree-optimization/81588
20559 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
20560 ranges[i].in_p, invert comparison code ccode. For >/>=,
20561 swap rhs1 and rhs2 and comparison code unconditionally,
20562 for </<= don't do that. Don't swap rhs1/rhs2 again if
20563 ranges[i].in_p, instead invert comparison code ccode if
20564 opcode or oe->rank is BIT_IOR_EXPR.
20567 * optabs.def (vec_extract_optab, vec_init_optab): Change from
20568 a direct optab to conversion optab.
20569 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
20570 with GET_MODE_INNER as last argument instead of optab_handler.
20571 * expmed.c (extract_bit_field_1): Likewise. Use vector from
20572 vector extraction if possible and optab is available.
20573 * expr.c (store_constructor): Use convert_optab_handler instead
20574 of optab_handler. Use vector initialization from smaller
20575 vectors if possible and optab is available.
20576 * tree-vect-stmts.c (vectorizable_load): Likewise.
20577 * doc/md.texi (vec_extract, vec_init): Document that the optabs
20578 now have two modes.
20579 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
20580 of vec_init from half-sized vectors with the same element mode.
20581 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
20582 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
20583 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
20584 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
20585 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
20586 after mode in gen_vec_extract* calls.
20587 (vec_extract<mode>): Renamed to ...
20588 (vec_extract<mode><ssescalarmodelower>): ... this.
20589 (vec_extract<mode><ssehalfvecmodelower>): New expander.
20590 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
20591 element mode after mode in gen_vec_init* calls.
20592 (VEC_INIT_HALF_MODE): New mode iterator.
20593 (vec_init<mode>): Renamed to ...
20594 (vec_init<mode><ssescalarmodelower>): ... this.
20595 (vec_init<mode><ssehalfvecmodelower>): New expander.
20596 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
20597 (vec_extractv2sfsf): ... this.
20598 (vec_initv2sf): Renamed to ...
20599 (vec_initv2sfsf): ... this.
20600 (vec_extractv2si): Renamed to ...
20601 (vec_extractv2sisi): ... this.
20602 (vec_initv2si): Renamed to ...
20603 (vec_initv2sisi): ... this.
20604 (vec_extractv4hi): Renamed to ...
20605 (vec_extractv4hihi): ... this.
20606 (vec_initv4hi): Renamed to ...
20607 (vec_initv4hihi): ... this.
20608 (vec_extractv8qi): Renamed to ...
20609 (vec_extractv8qiqi): ... this.
20610 (vec_initv8qi): Renamed to ...
20611 (vec_initv8qiqi): ... this.
20612 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
20613 (vec_init<mode>): Renamed to ...
20614 (vec_init<mode><VEC_base_l>): ... this.
20615 (vec_extract<mode>): Renamed to ...
20616 (vec_extract<mode><VEC_base_l>): ... this.
20617 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
20618 (vec_initv2sfsf): ... this.
20619 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
20620 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
20621 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
20622 element mode after mode in gen_vec_init* calls.
20623 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
20624 (vec_init<mode><Vel>): ... this.
20625 (vec_extract<mode>): Renamed to ...
20626 (vec_extract<mode><Vel>): ... this.
20627 * config/aarch64/iterators.md (Vel): New mode attribute.
20628 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
20629 Add element mode after mode in gen_vec_extract* calls.
20630 * config/s390/vector.md (non_vec_l): New mode attribute.
20631 (vec_extract<mode>): Renamed to ...
20632 (vec_extract<mode><non_vec_l>): ... this.
20633 (vec_init<mode>): Renamed to ...
20634 (vec_init<mode><non_vec_l>): ... this.
20635 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
20636 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
20638 * config/arm/iterators.md (V_elem_l): New mode attribute.
20639 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
20640 (vec_extract<mode><V_elem_l>): ... this.
20641 (vec_extractv2di): Renamed to ...
20642 (vec_extractv2didi): ... this.
20643 (vec_init<mode>): Renamed to ...
20644 (vec_init<mode><V_elem_l>): ... this.
20645 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
20646 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
20647 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
20648 Add element mode after gen_vec_extract* calls.
20649 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
20650 (vec_init<mode><unitmode>): ... this.
20651 (vec_extract<mode>): Renamed to ...
20652 (vec_extract<mode><unitmode>): ... this.
20653 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
20654 (vec_init<mode><unitmode>): ... this.
20655 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
20656 (vec_initv2sfsf): ... this.
20657 (vec_extractv2sf): Renamed to ...
20658 (vec_extractv2sfsf): ... this.
20659 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
20660 Add element mode after gen_vec_extract* calls.
20661 * config/mips/mips.md (unitmode): New mode iterator.
20662 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
20663 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
20664 * config/spu/spu.md (inner_l): New mode attribute.
20665 (vec_init<mode>): Renamed to ...
20666 (vec_init<mode><inner_l>): ... this.
20667 (vec_extract<mode>): Renamed to ...
20668 (vec_extract<mode><inner_l>): ... this.
20669 * config/sparc/sparc.md (veltmode): New mode iterator.
20670 (vec_init<VMALL:mode>): Renamed to ...
20671 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
20672 * config/ia64/vect.md (vec_initv2si): Renamed to ...
20673 (vec_initv2sisi): ... this.
20674 (vec_initv2sf): Renamed to ...
20675 (vec_initv2sfsf): ... this.
20676 (vec_extractv2sf): Renamed to ...
20677 (vec_extractv2sfsf): ... this.
20678 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
20679 (vec_init<mode>): Renamed to ...
20680 (vec_init<mode><VEC_base_l>): ... this.
20681 (vec_extract<mode>): Renamed to ...
20682 (vec_extract<mode><VEC_base_l>): ... this.
20683 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
20684 (vec_initv2sfsf): ... this.
20685 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
20686 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
20687 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
20688 gen_vec_init* calls.
20690 2017-08-01 Richard Biener <rguenther@suse.de>
20692 PR tree-optimization/81297
20693 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
20694 TREE_OVERFLOW from INTEGER_CSTs.
20696 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
20698 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
20700 2017-07-31 Carl Love <cel@us.ibm.com>
20702 * config/rs6000/rs6000-c: Add support for built-in functions
20703 vector signed char vec_xl_be (signed long long, signed char *);
20704 vector unsigned char vec_xl_be (signed long long, unsigned char *);
20705 vector signed int vec_xl_be (signed long long, signed int *);
20706 vector unsigned int vec_xl_be (signed long long, unsigned int *);
20707 vector signed long long vec_xl_be (signed long long, signed long long *);
20708 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
20709 vector signed short vec_xl_be (signed long long, signed short *);
20710 vector unsigned short vec_xl_be (signed long long, unsigned short *);
20711 vector double vec_xl_be (signed long long, double *);
20712 vector float vec_xl_be (signed long long, float *);
20713 * config/rs6000/altivec.h (vec_xl_be): Add #define.
20714 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
20715 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
20717 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
20718 (altivec_expand_builtin): Add switch statement to call
20719 altivec_expand_xl_be for each builtin.
20720 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
20721 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
20722 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
20723 __builtin_vsx_le_be_v16qi.
20724 * doc/extend.texi: Update the built-in documentation file for the
20725 new built-in functions.
20727 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
20730 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
20732 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
20734 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20736 * config.gcc: Add z14.
20737 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
20738 CPU model numbers for z13s and z14.
20739 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
20741 * config/s390/s390-opts.h (enum processor_type): Rename
20742 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
20743 * config/s390/s390.c (processor_table): Add field for CPU name to
20744 be passed to Binutils.
20745 (s390_asm_output_machine_for_arch): Use the new field in
20746 processor_table for Binutils.
20747 (s390_expand_builtin): Replace arch12 with z14.
20748 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
20749 (s390_get_sched_attrmask): Likewise.
20750 (s390_get_unit_mask): Likewise.
20751 * config/s390/s390.opt: Add z14 to processor_type enum.
20753 2017-07-31 Martin Jambor <mjambor@suse.cz>
20756 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
20757 regardless of optimization level.
20759 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
20760 Martin Liska <mliska@suse.cz>
20762 * predict.def: Remove old comment and adjust probability.
20763 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
20764 PREDICT statements.
20766 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
20769 * config/i386/i386.c (ix86_function_naked): New function.
20770 (ix86_can_use_return_insn_p): Return false for naked functions.
20771 (ix86_expand_prologue): Skip prologue for naked functions.
20772 (ix86_expand_epilogue): Skip epilogue for naked functions
20773 and emit trap instruction.
20774 (ix86_warn_func_return): New function.
20775 (ix86_attribute_table): Add "naked" attribute specification.
20776 (TARGET_WARN_FUNC_RETURN): Define.
20777 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
20779 2017-07-31 Martin Liska <mliska@suse.cz>
20781 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
20782 (dump_gimple_bb_header): Always dump BB info.
20783 (pp_cfg_jump): Do not append info about BB when dumping a jump.
20785 2017-07-31 Martin Liska <mliska@suse.cz>
20788 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
20789 also with current_function_decl non-null equality.
20791 2017-07-31 Jakub Jelinek <jakub@redhat.com>
20794 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
20795 change type to the element type, instead add eltype variable and
20796 use it where we are interested in the element type.
20798 PR tree-optimization/81603
20799 * ipa-polymorphic-call.c
20800 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
20801 offset arithmetic in offset_int, bail out if the resulting bit offset
20802 doesn't fit into shwi.
20804 2017-07-31 Martin Liska <mliska@suse.cz>
20806 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
20807 (gimplify_save_expr): Fix comment.
20809 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
20812 * config/i386/i386.c (ix86_function_arg): Update arguments for
20814 (ix86_compute_frame_layout): Set the initial stack offset to
20815 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
20816 INCOMING_FRAME_SP_OFFSET.
20817 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
20818 stack before exception handler returns.
20819 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
20820 the 'ERROR_CODE' for exception handler.
20822 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
20824 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
20825 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
20826 (ASM_OUTPUT_REG_POP): Ditto.
20827 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
20828 instead of asm_fprintf to output pure string.
20830 2017-07-29 Jakub Jelinek <jakub@redhat.com>
20832 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
20833 to imported_module_or_decl hook.
20834 (debug_nothing_tree_tree_tree_bool): Remove.
20835 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
20836 * debug.c (do_nothing_debug_hooks): Use
20837 debug_nothing_tree_tree_tree_bool_bool instead of
20838 debug_nothing_tree_tree_tree_bool.
20839 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20840 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
20841 * sdbout.c (sdb_debug_hooks): Likewise.
20842 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
20843 (gen_namespace_die): Add DW_AT_export_symbols attribute if
20845 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
20846 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
20847 attribute, don't add anything.
20849 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20851 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
20852 (fold_build2_stat_loc): Likewise.
20853 (fold_build3_stat_loc): Likewise.
20854 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
20855 (fold_build1_loc): Remove macro.
20856 (fold_build2_loc): Likewise.
20857 (fold_build3_loc): Likewise.
20859 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20861 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
20862 (gimple_build_debug_bind_source_stat): Likewise.
20863 * gimple.h (gimple_build_debug_bind): Remove macro.
20864 (gimple_build_debug_bind_source): Likewise.
20866 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20868 * bitmap.c (bitmap_alloc): Adjust.
20869 (bitmap_gc_alloc): Likewise.
20870 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
20872 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20874 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
20875 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
20876 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
20877 (bitmap_gc_alloc_stat): Likewise.
20878 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
20880 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20882 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
20883 * rtl.h (shallow_copy_rtx): Remove macro.
20885 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20887 * emit-rtl.c (gen_raw_REG): Adjust.
20888 * gengenrtl.c (gendef): Likewise.
20889 * rtl.c (rtx_alloc_stat): Remove _stat from name.
20890 * rtl.h (rtx_alloc): Remove macro.
20892 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20894 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
20895 (build_tree_list_stat): Likewise.
20896 * tree.h (build_tree_list): Remove macro.
20897 (build_tree_list_vec): Likewise.
20899 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20901 * tree.c (make_vector_stat): Remove _stat from name.
20902 (build_vector_stat): Likewise.
20903 * tree.h (make_vector_stat): Remove macro.
20904 (build_vector_stat): Likewise.
20906 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20908 * tree.h (build_var_debug_value): Remove prototype.
20910 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20912 * tree.c (tree_cons_stat): Remove _stat from name.
20913 * tree.h (tree_cons): Remove macro.
20915 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20917 * tree.c (build_vl_exp_stat): Remove _stat from name.
20918 * tree.h (build_vl_exp): Remove macro.
20920 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20922 * tree.c (build_decl_stat): Remove _stat from name.
20923 * tree.h (build_decl): Remove macro.
20925 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20927 * gimple.c (gimple_build_with_ops_stat): Adjust.
20928 (gimple_alloc_stat): Remove _stat from name.
20929 * gimple.h (gimple_alloc): Remove macro.
20931 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20933 * tree.c (make_tree_vec_stat): Remove _stat from name.
20934 (grow_tree_vec_stat): Likewise.
20935 * tree.h (make_tree_vec_stat): Adjust prototype.
20936 (grow_tree_vec_stat): Likewise.
20937 (make_tree_vec): Remove macro.
20938 (grow_tree_vec): Likewise.
20940 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20942 * fold-const.c (fold_build1_stat_loc): Adjust.
20943 (fold_build2_stat_loc): Likewise.
20944 (fold_build3_stat_loc): Likewise.
20945 * tree.c (build0_stat): Remove _stat from name.
20946 (build1_stat): Likewise.
20947 (build2_stat): Likewise.
20948 (build3_stat): Likewise.
20949 (build4_stat): Likewise.
20950 (build5_stat): Likewise.
20951 * tree.h (build1_loc): Remove macro, and rename _stat function
20953 (build2_loc): Likewise.
20954 (build3_loc): Likewise.
20955 (build4_loc): Likewise.
20956 (build5_loc): Likewise.
20958 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20960 * tree.c (make_int_cst_stat): Remove _stat from name.
20961 * tree.h (make_int_cst_stat): Adjust prototype.
20962 (make_int_cst): Remove macro.
20964 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20966 * tree.c (make_tre_binfo_stat): Remove _stat from name.
20967 * tree.h (make_tree_binfo_stat): Adjust prototype.
20968 (make_tree_binfo): Remove.
20970 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20972 * tree.c (copy_node_stat): Rename to copy_node.
20973 (build_distinct_type_copy): Adjust.
20974 * tree.h (copy_node_stat): Adjust prototype.
20975 (copy_node): Remove macro.
20977 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20979 * tree.c (make_node_stat): rename to make_node.
20980 (build_tree_list_stat): Adjust.
20981 (build0_stat): Likewise.
20982 (build2_stat): Likewise.
20983 (build3_stat): Likewise.
20984 (build4_stat): Likewise.
20985 (build5_stat): Likewise.
20986 (build_decl_stat): Likewise.
20987 * tree.h (make_node_stat): Adjust prototype.
20988 (make_node): remove macro.
20990 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
20992 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
20993 (PPC_FEATURE2_SCV): Likewise.
20994 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
20996 2017-07-28 Tamar Christina <tamar.christina@arm.com>
20998 * config/aarch64/aarch64.c
20999 (aarch64_internal_mov_immediate): Add new special pattern.
21000 * config/aarch64/aarch64.md (*movdi_aarch64):
21001 Add reg/32bit const mov case.
21003 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21004 Richard Sandiford <richard.sandiford@linaro.org>
21006 * config/aarch64/aarch64.md (mov<mode>): Generalize.
21007 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21008 Add integer and movi cases.
21009 (movi-split-hf-df-sf split, fp16): New.
21010 (enabled): Added TARGET_FP_F16INST.
21011 * config/aarch64/iterators.md (GPF_HF): New.
21012 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
21014 2017-07-28 Tamar Christina <tamar.christina@arm.com>
21016 * config/aarch64/aarch64.c
21017 (aarch64_simd_container_mode): Add prototype.
21018 (aarch64_expand_mov_immediate): Add HI support.
21019 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
21020 (aarch64_can_const_movi_rtx_p): New.
21021 (aarch64_preferred_reload_class):
21022 Remove restrictions of using FP registers for certain SIMD operations.
21023 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
21024 (aarch64_valid_floating_const): Add integer move validation.
21025 (aarch64_simd_imm_scalar_p): Remove.
21026 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
21027 (aarch64_legitimate_constant_p): Expand list of supported cases.
21028 * config/aarch64/aarch64-protos.h
21029 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
21030 (aarch64_reinterpret_float_as_int): New.
21031 (aarch64_simd_imm_scalar_p): Remove.
21032 * config/aarch64/constraints.md (Uvi): New.
21033 (Dd): Split into Ds and new Dd.
21034 * config/aarch64/aarch64.md (*movsi_aarch64):
21036 (*movdi_aarch64): Add SIMD mov case.
21038 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21040 * tree-predcom.c: (struct chain): Handle store-store chain in which
21041 stores for elimination only store loop invariant values.
21042 (execute_pred_commoning_chain): Ditto.
21043 (prepare_initializers_chain_store_elim): Ditto.
21044 (prepare_finalizers): Ditto.
21045 (is_inv_store_elimination_chain): New function.
21046 (initialize_root_vars_store_elim_1): New function.
21048 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21050 * tree-predcom.c: Revise general description of the pass.
21051 (enum chain_type): New enum type for store elimination.
21052 (struct chain): New field supporting store elimination.
21053 (struct component): Ditto.
21054 (dump_chain): Dump store-stores chain.
21055 (release_chain): Release resources.
21056 (split_data_refs_to_components): Compute and create component
21057 contains only stores for elimination.
21058 (get_chain_last_ref_at): New function.
21059 (make_invariant_chain): Initialization.
21060 (make_rooted_chain): Specify chain type in parameter and record it.
21061 (add_looparound_copies): Skip for store-stores chain.
21062 (determine_roots_comp): Compute type of chain and pass it to
21064 (initialize_root_vars_store_elim_2): New function.
21065 (finalize_eliminated_stores): New function.
21066 (remove_stmt): Handle store for elimination.
21067 (execute_pred_commoning_chain): Execute predictive commoning on
21068 store-store chains.
21069 (determine_unroll_factor): Skip unroll for store-stores chain.
21070 (prepare_initializers_chain_store_elim): New function.
21071 (prepare_initializers_chain): Hanlde store-store chain.
21072 (prepare_finalizers_chain, prepare_finalizers): New function.
21073 (tree_predictive_commoning_loop): Return integer value indicating
21074 if loop is unrolled or lcssa form is corrupted.
21075 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
21077 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21079 * tree-predcom.c (initialize_root): Delete.
21080 (execute_pred_commoning_chain): Initialize root vars and replace
21081 reference of non-combined chain directly, rather than call above
21084 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21086 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
21087 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
21089 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21091 * tree-predcom.c (struct chain): New field init_seq.
21092 (release_chain): Release init_seq.
21093 (prepare_initializers_chain): Record intialization stmts in above
21095 (insert_init_seqs): New function.
21096 (tree_predictive_commoning_loop): Call insert_init_seqs.
21098 2017-07-28 Bin Cheng <bin.cheng@arm.com>
21100 * tree-predcom.c (determine_roots_comp): Skip trivial components.
21102 2017-07-28 Richard Biener <rguenther@suse.de>
21104 * match.pd: Remove superfluous :c.
21105 * genmatch.c (simplify::id): Add member.
21106 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
21108 (current_id): New global.
21109 (dt_node::parent): Move from ...
21110 (dt_operand::parent): ... here. Add for_id member.
21111 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
21112 (decision_tree::find_node): Relax order requirement when
21113 merging DT_TRUE nodes to ones inbetween the current simplify
21114 and the one we try to merge with. Add diagnostic whenever
21115 we need to enforce pattern order by not merging.
21116 (decision_tree::insert): Set current_id.
21117 (decision_tree::print_node): Dump parent node and for_id.
21118 (parser::last_id): Add member.
21119 (parser::push_simplify): Assign unique id.
21120 (parser::parser): Initialize last_id.
21122 2017-07-28 Martin Liska <mliska@suse.cz>
21125 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
21126 gimple_build_debug_bind.
21128 2017-07-28 Richard Biener <rguenther@suse.de>
21130 PR tree-optimization/81502
21131 * match.pd: Add pattern combining BIT_INSERT_EXPR with
21133 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
21135 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
21136 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
21137 for BIT_FIELD_REF args.
21138 * fold-const.c (make_bit_field_ref): Likewise.
21139 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
21141 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21144 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
21145 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
21146 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
21147 Or it into SANITIZER_UNDEFINED.
21148 * ubsan.c: Include gimple-fold.h and varasm.h.
21149 (ubsan_expand_ptr_ifn): New function.
21150 (instrument_pointer_overflow): New function.
21151 (maybe_instrument_pointer_overflow): New function.
21152 (instrument_object_size): Formatting fix.
21153 (pass_ubsan::execute): Call instrument_pointer_overflow
21154 and maybe_instrument_pointer_overflow.
21155 * internal-fn.c (expand_UBSAN_PTR): New function.
21156 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
21157 * sanitizer.def (__ubsan_handle_pointer_overflow,
21158 __ubsan_handle_pointer_overflow_abort): New builtins.
21159 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
21160 * internal-fn.def (UBSAN_PTR): New internal function.
21161 * opts.c (sanitizer_opts): Add pointer-overflow.
21162 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
21163 * fold-const.c (build_range_check): Compute pointer range check in
21164 integral type if pointer arithmetics would be needed. Formatting
21167 2017-07-28 Martin Liska <mliska@suse.cz>
21170 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
21171 parameters that are of a variable-length.
21173 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
21175 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
21177 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
21178 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
21179 (CRT_CALL_STATIC_FUNCTION): Likewise.
21180 (ASM_DEFAULT_SPEC): New define.
21181 (ASM_SPEC32): Likewise.
21182 (ASM_SPEC64): Likewise.
21183 (ASM_SPEC_COMMON): Likewise.
21184 (ASM_SPEC): Likewise.
21185 (INVALID_64BIT): Likewise.
21186 (LINK_OS_DEFAULT_SPEC): Likewise.
21187 (LINK_OS_SPEC32): Likewise.
21188 (LINK_OS_SPEC64): Likewise.
21189 (POWERPC_LINUX): Likewise.
21190 (PTRDIFF_TYPE): Likewise.
21191 (RESTORE_FP_PREFIX): Likewise.
21192 (RESTORE_FP_SUFFIX): Likewise.
21193 (SAVE_FP_PREFIX): Likewise.
21194 (SAVE_FP_SUFFIX): Likewise.
21195 (SIZE_TYPE): Likewise.
21196 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
21197 (TARGET_64BIT): Likewise.
21198 (TARGET_64BIT): Likewise.
21199 (TARGET_AIX): Likewise.
21200 (WCHAR_TYPE_SIZE): Likewise.
21201 (WCHAR_TYPE): Undefine.
21202 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
21203 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
21204 (CPP_OS_RTEMS_SPEC): Delete.
21205 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
21206 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
21208 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
21210 2017-07-28 Jakub Jelinek <jakub@redhat.com>
21212 PR tree-optimization/81578
21213 * tree-parloops.c (build_new_reduction): Bail out if
21214 reduction_code isn't one of the standard OpenMP reductions.
21215 Move the details printing after that decision.
21217 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21219 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
21220 related to reload_in_progress.
21221 (splat_input_operand): Likewise.
21222 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
21224 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
21226 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
21227 (TARGET_INSTANTIATE_DECLS): Likewise.
21228 (legitimate_indexed_address_p): Delete reload_in_progress code.
21229 (rs6000_debug_legitimate_address_p): Likewise.
21230 (rs6000_eliminate_indexed_memrefs): Likewise.
21231 (rs6000_emit_le_vsx_store): Likewise.
21232 (rs6000_emit_move_si_sf_subreg): Likewise.
21233 (rs6000_emit_move): Likewise.
21234 (register_to_reg_type): Likewise.
21235 (rs6000_pre_atomic_barrier): Likewise.
21236 (rs6000_machopic_legitimize_pic_address): Likewise.
21237 (rs6000_allocate_stack_temp): Likewise.
21238 (rs6000_address_for_fpconvert): Likewise.
21239 (rs6000_address_for_altivec): Likewise.
21240 (rs6000_secondary_memory_needed_rtx): Delete function.
21241 (rs6000_check_sdmode): Likewise.
21242 (rs6000_alloc_sdmode_stack_slot): Likewise.
21243 (rs6000_instantiate_decls): Likewise.
21244 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
21245 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
21246 Delete reload_in_progress.
21247 (*vec_reload_and_plus_<mptrsize>): Likewise.
21248 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
21249 (vsx_div_v2di): Likewise.
21250 (vsx_udiv_v2di): Likewise.
21252 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21254 * config/rs6000/rs6000.opt (mlra): Replace with stub.
21255 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
21256 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
21257 (rs6000_debug_reg_global): Delete print of LRA status.
21258 (rs6000_option_override_internal): Delete dead LRA related code.
21259 (rs6000_lra_p): Delete function.
21260 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
21262 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
21264 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
21265 * config/riscv/rtems.h: New file.
21267 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21268 Sudakshina Das <sudi.das@arm.com>
21270 * config/aarch64/aarch64.md
21271 (define_split for and<mode>3nr_compare): Move
21272 non aarch64_logical_operand to a register.
21273 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
21274 register immediate operand to a register.
21275 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
21277 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
21279 PR middle-end/81564
21280 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
21282 2017-07-27 Richard Biener <rguenther@suse.de>
21284 PR tree-optimization/81573
21285 PR tree-optimization/81494
21286 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
21287 multi defuse cycle case.
21289 2017-07-27 Richard Biener <rguenther@suse.de>
21291 PR tree-optimization/81571
21292 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
21295 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
21297 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
21298 earlier and only if MASK_FPU is set. Adjust formatting.
21300 2017-07-27 Martin Liska <mliska@suse.cz>
21302 * opt-functions.awk: Add validation of value of Init.
21303 * optc-gen.awk: Pass new argument.
21305 2017-07-27 Martin Liska <mliska@suse.cz>
21307 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
21308 Fix wrong condition.
21310 2017-07-27 Martin Liska <mliska@suse.cz>
21312 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
21313 BBs and edges seen by autoFDO.
21315 2017-07-27 Richard Biener <rguenther@suse.de>
21317 PR tree-optimization/81502
21318 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
21319 with incompatible but same sized type.
21320 (execute_update_addresses_taken): Likewise.
21322 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
21324 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
21325 flag_tree_loop_vectorize rather than flag_tree_vectorize.
21327 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21330 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
21331 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
21332 Change s_operand to memory_operand.
21334 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
21336 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
21337 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
21338 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
21339 Emit instructions rather than returning an expression. Handle TFmode
21340 and KFmode by casting to TImode.
21341 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
21342 (rs6000_emit_le_vsx_store): Likewise.
21343 * config/rs6000/vsx.md (VSX_TI): New iterator.
21344 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
21345 (*vsx_le_undo_permute_<mode>): Likewise.
21346 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
21347 emit the split sequence.
21348 (*vsx_le_perm_store_<mode>): Likewise.
21350 2017-07-27 Jakub Jelinek <jakub@redhat.com>
21352 PR tree-optimization/81555
21353 PR tree-optimization/81556
21354 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
21355 if true, force CHANGED for the recursive invocation.
21356 (reassociate_bb): Remember original length of ops array, pass
21357 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
21359 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
21360 attributes for noipa attribute. For naked attribute use
21361 lookup_attribute first before lookup_attribute_spec.
21362 * final.c (rest_of_handle_final): Disable IPA RA for functions with
21364 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
21365 for functions with noipa attribute.
21366 (cgraph_externally_visible_p): Return true for functions with noipa
21368 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
21369 for functions with noipa attribute.
21370 * doc/extend.texi: Document noipa function attribute.
21371 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
21372 also for functions with noipa attribute.
21373 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
21375 2017-07-26 Andrew Pinski <apinski@cavium.com>
21377 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
21378 vec_unalign_load_cost and vec_unalign_store_cost.
21380 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
21382 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
21383 -mvsx-small-integer option.
21384 (ISA_3_0_MASKS_IEEE): Likewise.
21385 (OTHER_VSX_VECTOR_MASKS): Likewise.
21386 (POWERPC_MASKS): Likewise.
21387 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
21388 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
21389 code, only testing for DImode being allowed in non-VSX floating
21391 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
21392 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
21394 (rs6000_option_override_internal): Delete -mvsx-small-integer.
21395 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
21396 TARGET_P8_VECTOR test.
21397 (rs6000_secondary_reload_simple_move): Likewise.
21398 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
21399 since TARGET_P9_VECTOR was already tested.
21400 (rs6000_opt_masks): Remove -mvsx-small-integer.
21401 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
21402 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21404 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
21405 test for TARGET_VEXTRACTUB was used, and that uses
21407 (p9 extract splitter): Likewise.
21408 (vsx_extract_<mode>_di_p9): Likewise.
21409 (vsx_extract_<mode>_store_p9): Likewise.
21410 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
21411 for TARGET_P9_VECTOR was used. Delete code that is now dead with
21412 the elimination of TARGET_VSX_SMALL_INTEGER.
21413 (vsx_extract_<mode>_p8): Likewise.
21414 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
21415 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
21416 (vsx_set_<mode>_p9): Likewise.
21417 (vsx_set_v4sf_p9): Likewise.
21418 (vsx_set_v4sf_p9_zero): Likewise.
21419 (vsx_insert_extract_v4sf_p9): Likewise.
21420 (vsx_insert_extract_v4sf_p9_2): Likewise.
21421 * config/rs6000/rs6000.md (sign extend splitter): Change
21422 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
21423 (floatsi<mode>2_lfiwax_mem): Likewise.
21424 (floatunssi<mode>2_lfiwzx_mem): Likewise.
21425 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
21426 since a test for TARGET_P9_VECTOR was used.
21427 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21428 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
21429 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21430 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21431 TARGET_P8_VECTOR test.
21432 (fix_trunc<mode>si2_stfiwx): Likewise.
21433 (fix_trunc<mode>si2_internal): Likewise.
21434 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
21435 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21437 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21438 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
21439 TARGET_P8_VECTOR test.
21440 (fixuns_trunc<mode>si2_stfiwx): Likewise.
21441 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
21442 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
21444 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
21445 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
21446 since a test for TARGET_P9_VECTOR was used.
21447 (splitter for loading small constants): Likewise.
21449 2017-07-26 Andrew Pinski <apinski@cavium.com>
21451 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
21454 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
21457 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
21458 (fp_valid_at): Likewise.
21460 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21462 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
21463 (qdf24xx_addrcost_table): Likewise.
21464 (cortexa57_tunings): Update to use generic_branch_cost.
21465 (cortexa72_tunings): Likewise.
21466 (cortexa73_tunings): Likewise.
21467 (qdf24xx_tunings): Likewise.
21469 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
21471 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
21472 (thunderx2t99_branch_cost): Likewise.
21473 (cortexa35_tunings): Update to use generic_branch_cost.
21474 (cortexa53_tunings): Likewise.
21475 (cortexa57_tunings): Likewise.
21476 (cortexa72_tunings): Likewise.
21477 (cortexa73_tunings): Likewise.
21478 (thunderx2t99_tunings): Likewise.
21480 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21482 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
21483 (sparc_option_override): Honour MASK_FSMULD.
21484 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
21485 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
21486 * config/sparc/sparc.opt (mfsmuld): New option.
21487 * doc/invoke.texi (mfsmuld): Document option.
21489 2017-07-26 Marek Polacek <polacek@redhat.com>
21491 PR middle-end/70992
21492 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
21494 2017-07-26 Richard Biener <rguenther@suse.de>
21496 * gimple-match-head.c (do_valueize): Return OP if valueize
21498 (get_def): New helper to get at the def stmt of a SSA name
21499 if valueize allows.
21500 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
21501 do_valueize to get at the def stmt.
21502 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
21504 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
21506 PR middle-end/46932
21507 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
21509 2017-07-26 Martin Liska <mliska@suse.cz>
21512 * function.c (expand_function_start): Make expansion of
21513 nonlocal_goto_save_area after parm_birth_insn.
21515 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
21517 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
21518 from all CPU target flags enable members.
21520 2017-07-26 Richard Biener <rguenther@suse.de>
21522 * genmatch.c (dt_simplify::gen): Make iterator vars const.
21523 (decision_tree::gen): Make 'type' const.
21524 (write_predicate): Likewise.
21526 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
21528 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
21529 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
21530 (rs6000_option_override_internal): Likewise.
21531 (rs6000_expand_vector_set): Likewise.
21532 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
21533 (TARGET_UPPER_REGS_SF): Likewise.
21534 (TARGET_UPPER_REGS_DI): Likewise.
21535 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
21536 (TARGET_DIRECT_MOVE_64BIT): Likewise.
21537 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
21538 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
21539 (Splitters for DI constants in Altivec registers): Likewise.
21540 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
21541 (vsx_set_v4sf_p9): Likewise.
21542 (vsx_set_v4sf_p9_zero): Likewise.
21543 (vsx_insert_extract_v4sf_p9): Likewise.
21544 (vsx_insert_extract_v4sf_p9_2): Likewise.
21546 2017-07-25 Carl Love <cel@us.ibm.com>
21548 * doc/extend.texi: Update the built-in documentation file for the
21549 existing built-in functions
21550 vector signed char vec_cnttz (vector signed char);
21551 vector unsigned char vec_cnttz (vector unsigned char);
21552 vector signed short vec_cnttz (vector signed short);
21553 vector unsigned short vec_cnttz (vector unsigned short);
21554 vector signed int vec_cnttz (vector signed int);
21555 vector unsigned int vec_cnttz (vector unsigned int);
21556 vector signed long long vec_cnttz (vector signed long long);
21557 vector unsigned long long vec_cnttz (vector unsigned long long);
21559 2017-07-25 Andrew Pinski <apinski@cavium.com>
21561 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
21562 accesses where the use is for the first operand of a BIT_INSERT.
21564 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21567 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21568 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21570 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
21572 * config/i386/gstabs.h: Delete.
21573 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
21575 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
21577 * config/i386/i386.c (ix86_decompose_address): Do not check for
21578 register RTX when looking at index_reg or base_reg.
21579 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
21581 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
21583 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
21584 to update EH info here.
21586 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21588 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
21590 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21592 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
21594 2017-07-25 Torsten Duwe <duwe@suse.de>
21596 * common.opt: Introduce -fpatchable-function-entry
21597 command line option, and its variables function_entry_patch_area_size
21598 and function_entry_patch_area_start.
21599 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
21600 including a two-value parser.
21601 * target.def (print_patchable_function_entry): New target hook.
21602 * targhooks.h (default_print_patchable_function_entry): New function.
21603 * targhooks.c (default_print_patchable_function_entry): Likewise.
21604 * toplev.c (process_options): Switch off IPA-RA if
21605 patchable function entries are being generated.
21606 * varasm.c (assemble_start_function): Look at the
21607 patchable-function-entry command line switch and current
21608 function attributes and maybe generate NOP instructions by
21609 calling the print_patchable_function_entry hook.
21610 * doc/extend.texi: Document patchable_function_entry attribute.
21611 * doc/invoke.texi: Document -fpatchable_function_entry
21612 command line option.
21613 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
21615 * doc/tm.texi: Re-generate.
21617 2017-07-25 Jakub Jelinek <jakub@redhat.com>
21620 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
21621 TARGET_AVX512DQ rather than TARGET_AVX512BW.
21623 2017-07-25 Tamar Christina <tamar.christina@arm.com>
21625 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
21627 2017-07-25 Richard Biener <rguenther@suse.de>
21629 PR tree-optimization/81455
21630 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
21631 not walk in cycles when looking for guards.
21633 2017-07-25 Richard Biener <rguenther@suse.de>
21635 PR tree-optimization/81529
21636 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
21637 when optimizing backedge uses.
21639 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
21641 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
21643 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
21644 to dl_section_ref. On AIX, append an expression to subtract
21645 the size of the section length to dl_section_ref.
21647 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
21649 * configure.ac: If any of the config.* scripts fail, exit 1.
21650 * configure: Regenerate.
21652 2017-07-25 Richard Biener <rguenther@suse.de>
21654 PR middle-end/81546
21655 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
21656 of immediate uses, be more verbose on errors.
21658 2017-07-25 Richard Biener <rguenther@suse.de>
21660 PR tree-optimization/81510
21661 * tree-vect-loop.c (vect_is_simple_reduction): When the
21662 reduction stmt is not inside the loop bail out.
21664 2017-07-25 Richard Biener <rguenther@suse.de>
21666 PR tree-optimization/81303
21667 * tree-vect-loop-manip.c (vect_loop_versioning): Build
21668 profitability check against LOOP_VINFO_NITERSM1.
21670 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
21672 * domwalk.c (cmp_bb_postorder): Simplify.
21673 (sort_bbs_postorder): New function. Use it...
21674 (dom_walker::walk): ...here to optimize common cases.
21676 2017-07-25 Martin Liska <mliska@suse.cz>
21679 * ipa-visibility.c (function_and_variable_visibility): Make the
21680 redirection just on target that supports aliasing.
21681 Fix GNU coding style.
21683 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
21686 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
21688 * config/arm/rtems.h: Likewise.
21689 * config/bfin/rtems.h: Likewise.
21690 * config/i386/rtemself.h: Likewise.
21691 * config/lm32/rtems.h: Likewise.
21692 * config/m32c/rtems.h: Likewise.
21693 * config/m68k/rtemself.h: Likewise.
21694 * config/microblaze/rtems.h: Likewise.
21695 * config/mips/rtems.h: Likewise.
21696 * config/moxie/rtems.h: Likewise.
21697 * config/nios2/rtems.h: Likewise.
21698 * config/powerpcspe/rtems.h: Likewise.
21699 * config/rs6000/rtems.h: Likewise.
21700 * config/rtems.h: Likewise.
21701 * config/sh/rtems.h: Likewise.
21702 * config/sh/rtemself.h: Likewise.
21703 * config/sparc/rtemself.h: Likewise.
21705 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
21708 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
21709 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
21710 * tree-ssa-structalias.c (alias_get_name): Same.
21712 2017-07-25 Bin Cheng <bin.cheng@arm.com>
21715 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
21716 instructions if no du chain is found.
21718 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
21720 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
21722 2017-07-25 Richard Biener <rguenther@suse.de>
21724 PR middle-end/81505
21725 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
21728 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
21730 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
21731 upper-regs options.
21732 (ISA_2_7_MASKS_SERVER): Likewise.
21733 (ISA_3_0_MASKS_IEEE): Likewise.
21734 (OTHER_P8_VECTOR_MASKS): Likewise.
21735 (OTHER_VSX_VECTOR_MASKS): Likewise.
21736 (POWERPC_MASKS): Likewise.
21737 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
21738 duplicate list of options.
21739 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
21740 explicit -mupper-regs options.
21741 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
21742 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
21743 alias for -mupper-regs-df.
21744 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
21745 (rs6000_init_hard_regno_mode_ok): Likewise.
21746 (rs6000_option_override_internal): Likewise.
21747 (rs6000_opt_masks): Likewise.
21748 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
21749 options in terms of whether -mvsx or -mpower8-vector was used.
21750 (TARGET_UPPER_REGS_DI): Likewise.
21751 (TARGET_UPPER_REGS_SF): Likewise.
21752 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
21753 -mupper-regs-* options.
21755 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21757 * passes.c (emergency_dump_function): Print some empty lines and a
21758 header before the RTL dump.
21760 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21762 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
21764 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
21767 * config/aarch64/aarch64.c (aarch64_classify_symbol):
21768 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
21770 2017-07-24 Carl Love <cel@us.ibm.com>
21772 * config/rs6000/rs6000-c.c: Add support for built-in functions
21773 vector float vec_extract_fp32_from_shorth (vector unsigned short);
21774 vector float vec_extract_fp32_from_shortl (vector unsigned short);
21775 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
21776 vec_extract_fp_from_shortl): Add defines for the two builtins.
21777 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
21778 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
21780 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
21781 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
21782 * doc/extend.texi: Update the built-in documentation file for the
21783 new built-in function.
21785 2017-07-24 Jakub Jelinek <jakub@redhat.com>
21788 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
21790 * doc/generic.texi: Likewise.
21791 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21792 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21794 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
21796 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
21797 (aarch64_mls_elt_merge<mode>): Likewise.
21799 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
21801 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21802 having __cxa_atexit.
21804 2017-07-23 Michael Collison <michael.collison@arm.com>
21806 * config/arm/arm.c (arm_option_override): Deprecate
21807 use of -mstructure-size-boundary.
21808 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
21809 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
21811 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21814 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
21815 Reduce cost estimate for direct moves.
21817 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
21820 * config/i386/i386.c (ix86_option_override_internal): Disable
21821 BMI, BMI2 and TBM instructions for -m16.
21823 2017-07-21 Carl Love <cel@us.ibm.com>
21825 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21826 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21827 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21828 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21829 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21830 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21831 VMULOSW): New enum "unspec" values.
21832 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21833 altivec_vmulosw): New patterns.
21834 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21835 VMULOSW): Add definitions.
21837 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
21839 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
21840 (qdf24xx): Likewise.
21841 * config/aarch64/aarch64-options-extensions.def (rdma); New.
21842 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
21843 (AARCH64_FL_V8_1): Renumber.
21844 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
21845 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
21846 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
21847 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
21848 rdma to feature modifiers list.
21850 2017-07-21 Yury Gribov <tetra2005@gmail.com>
21852 PR middle-end/56727
21853 * ipa-visibility (function_and_variable_visibility): Convert
21854 recursive PLT call to direct call if appropriate.
21856 2017-07-21 Andrew Pinski <apinski@cavium.com>
21858 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
21859 operand 1 to see if the types precision matches.
21860 * fold-const.c (operand_equal_p): Likewise.
21862 2017-07-21 Richard Biener <rguenther@suse.de>
21864 PR tree-optimization/81303
21865 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
21866 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
21867 (vect_peeling_hash_get_lowest_cost): Adjust.
21868 (vect_enhance_data_refs_alignment): Likewise. Use
21869 vect_get_peeling_costs_all_drs to compute the penalty for no
21870 peeling to match up costs.
21872 2017-07-21 Richard Biener <rguenther@suse.de>
21874 PR tree-optimization/81500
21875 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
21876 we didn't identify a reduction path.
21878 2017-07-21 Tom de Vries <tom@codesourcery.com>
21879 Cesar Philippidis <cesar@codesourcery.com>
21881 PR gcov-profile/81442
21882 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
21885 2017-07-21 Tom de Vries <tom@codesourcery.com>
21888 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
21890 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
21891 nvptx_override_options_after_change.
21893 2017-07-21 Ulrich Drepper <drepper@redhat.com>
21895 * dwarf2out.c (output_file_names): Avoid double testing for
21896 dwarf_version >= 5.
21898 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
21900 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
21902 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
21904 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
21906 (try_crossjump_to_edge): Do not punt on partitioned functions.
21908 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
21910 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
21911 Put all BBs reachable only via paths crossing cold region to cold
21913 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
21915 2016-07-21 Richard Biener <rguenther@suse.de>
21917 PR tree-optimization/81303
21918 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
21919 into account prologue and epilogue iterations when raising
21920 min_profitable_iters to sth at least covering one vector iteration.
21922 2017-07-21 Tamar Christina <tamar.christina@arm.com>
21924 * config/arm/arm.c (arm_test_cpu_arch_dat):
21927 2017-07-20 Nathan Sidwell <nathan@acm.org>
21929 Remove TYPE_METHODS.
21930 * tree.h (TYPE_METHODS): Delete.
21931 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
21932 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
21933 (dbxout_type_methods): Scan TYPE_FIELDS.
21934 (dbxout_type): Don't check TYPE_METHODS here.
21935 * function.c (use_register_for_decl): Always ignore register for
21936 class types when not optimizing.
21937 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
21938 * tree.c (free_lang_data_in_type): Stitch out member functions and
21939 templates from TYPE_FIELDS.
21940 (build_distinct_type_copy, verify_type_variant,
21941 verify_type): Member fns are on TYPE_FIELDS.
21942 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
21943 * tree-pretty-print.c (dump_generic_node): Likewise.
21945 2017-07-20 Jakub Jelinek <jakub@redhat.com>
21948 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
21949 V2TImode and V4TImode.
21950 (ix86_expand_vector_extract): Likewise.
21951 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
21952 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
21953 (ssescalarmode): Handle V4TImode and V2TImode.
21954 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
21955 (*vec_extractv2ti, *vec_extractv4ti): New insns.
21956 (VEXTRACTI128_MODE): New mode iterator.
21957 (splitter for *vec_extractv?ti first element): New.
21958 (VEC_INIT_MODE): New mode iterator.
21959 (vec_init<mode>): Consolidate 3 expanders into one using
21960 VEC_INIT_MODE mode iterator.
21962 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
21964 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
21965 non_spilled_static_chain_regno_p.
21967 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
21969 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
21971 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
21973 * bb-reorder.c (connect_traces): Allow copying of blocks within
21976 2017-07-20 Richard Biener <rguenther@suse.de>
21978 * gimple.h (gimple_phi_result): Add gphi * overload.
21979 (gimple_phi_result_ptr): Likewise.
21980 (gimple_phi_arg): Likewise. Adjust index assert to only
21981 allow actual argument accesses rather than all slots available
21983 (gimple_phi_arg_def): Add gphi * overload.
21984 * tree-phinodes.c (make_phi_node): Initialize only actual
21986 (resize_phi_node): Clear memory not covered by old node,
21987 do not initialize excess argument slots.
21988 (reserve_phi_args_for_new_edge): Initialize new argument slot
21991 2017-07-20 Bin Cheng <bin.cheng@arm.com>
21993 PR tree-optimization/81388
21995 2016-07-21 Bin Cheng <bin.cheng@arm.com>
21997 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
21998 by removing computation of may_be_zero.
22000 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22001 Tom de Vries <tom@codesourcery.com>
22003 PR middle-end/81030
22004 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
22005 when gimple level profile disagrees with what RTL expander did.
22007 2017-07-20 Richard Biener <rguenther@suse.de>
22009 PR tree-optimization/61171
22010 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
22011 (vect_analyze_stmt): Add slp instance parameter.
22012 (vectorizable_reduction): Likewise.
22013 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
22014 (vect_is_simple_reduction): Deal with chains not detected
22015 as SLP reduction chain, specifically not properly associated
22016 chains containing a mix of plus/minus.
22017 (get_reduction_op): Remove.
22018 (get_initial_defs_for_reduction): Simplify, pass in whether
22019 this is a reduction chain, pass in the SLP node for the PHIs.
22020 (vect_create_epilog_for_reduction): Get the SLP instance as
22022 (vectorizable_reduction): Get the SLP instance as arg.
22023 During analysis remember the SLP node with the PHIs in the
22024 instance. Simplify getting at the vectorized reduction PHIs.
22025 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
22026 through SLP instance.
22027 (vect_slp_analyze_operations): Likewise.
22028 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
22029 (vect_transform_stmt): Likewise.
22031 2017-07-20 Tom de Vries <tom@codesourcery.com>
22033 PR tree-optimization/81489
22034 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
22035 read of phi arg location to before loop that modifies phi.
22037 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22039 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
22042 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22044 PR middle-end/81331
22045 * except.c (execute): Fix ordering issue.
22047 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22049 PR rtl-optimization/81423
22050 * combine.c (make_compound_operation_int): Don't try to optimize
22051 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
22053 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
22055 PR rtl-optimization/81423
22056 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
22057 with a constant that is -1 in the truncated to mode.
22059 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22061 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
22062 (determine_unlikely_bbs): ... here.
22063 * predict.h (propagate_unlikely_bbs_forward): Declare.
22064 * cfgexpand.c (pass_expand::execute): Use it.
22065 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
22067 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
22068 propagate_unlikely_bbs_forward.
22070 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
22072 PR middle-end/81331
22073 * except.c (maybe_add_nop_after_section_switch): New function.
22076 2017-07-19 Tom de Vries <tom@codesourcery.com>
22078 * gimple.h (gimple_phi_set_arg): Make assert more strict.
22080 2017-07-19 Tom de Vries <tom@codesourcery.com>
22082 * gimple.h (gimple_phi_arg): Make assert more strict.
22084 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
22086 * config.gcc (powerpc*-*-*): Add mmintrin.h.
22087 * config/rs6000/mmintrin.h: New file.
22088 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
22090 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22092 PR tree-optimization/81346
22093 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
22095 2017-07-19 Tom de Vries <tom@codesourcery.com>
22097 * config/nvptx/nvptx.md (VECIM): Add V2DI.
22099 2017-07-19 Tom de Vries <tom@codesourcery.com>
22101 * config/nvptx/nvptx-modes.def: Add V2DImode.
22102 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
22103 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
22104 (nvptx_output_mov_insn): Handle lack of mov.b128.
22105 (nvptx_print_operand): Handle 'H' and 'L' codes.
22106 (nvptx_vector_mode_supported): Allow V2DImode.
22107 (nvptx_preferred_simd_mode): New function.
22108 (nvptx_data_alignment): New function.
22109 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
22110 nvptx_preferred_simd_mode.
22111 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
22113 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
22115 2017-07-19 Tom de Vries <tom@codesourcery.com>
22117 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
22118 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
22119 (nvptx_vector_mode_supported): New function. Allow V2SImode.
22120 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
22121 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
22122 (mov<VECIM>_insn): New define_insn.
22123 (define_expand "mov<VECIM>): New define_expand.
22125 2017-07-19 Tom de Vries <tom@codesourcery.com>
22127 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
22129 2017-07-19 Jakub Jelinek <jakub@redhat.com>
22131 PR tree-optimization/81346
22132 * fold-const.h (fold_div_compare, range_check_type): Declare.
22133 * fold-const.c (range_check_type): New function.
22134 (build_range_check): Use range_check_type.
22135 (fold_div_compare): No longer static, rewritten into
22136 a match.pd helper function.
22137 (fold_comparison): Don't call fold_div_compare here.
22138 * match.pd (X / C1 op C2): New optimization using fold_div_compare
22139 as helper function.
22141 2017-07-19 Nathan Sidwell <nathan@acm.org>
22143 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
22144 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
22145 * tree.c (find_decls_types_r, verify_type): Use
22146 TYPE_{MIN,MAX}_VALUE_RAW.
22147 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
22148 (hash_tree): Likewise.
22149 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
22151 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
22154 2017-07-18 Tom de Vries <tom@codesourcery.com>
22156 PR middle-end/81464
22157 * omp-expand.c (expand_omp_for_static_chunk): Handle
22158 equal-argument loop exit phi.
22160 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
22163 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
22164 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
22165 operand 2 predicate.
22166 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
22167 operand 2 predicate.
22168 (ror,rol -> rorx splitters): Use const_int_operand as
22169 operand 2 predicate.
22171 2017-06-18 Richard Biener <rguenther@suse.de>
22173 PR tree-optimization/81410
22174 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
22175 the gap in the ! slp_perm SLP case after each group.
22177 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22179 PR middle-end/81463
22180 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
22183 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
22185 PR middle-end/81462
22186 * predict.c (set_even_probabilities): Cleanup; do not affect
22187 probabilities that are already known.
22188 (combine_predictions_for_bb): Call even when count is set.
22190 2017-07-18 Nathan Sidwell <nathan@acm.org>
22192 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
22195 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22198 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
22199 optimization for loop niter analysis.
22201 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
22204 * config/avr/avr.c (avr_optimize_casesi): Don't use
22205 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
22207 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
22209 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
22210 body_cost_vec from _vect_peel_extended_info.
22211 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
22212 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
22215 2017-07-18 Bin Cheng <bin.cheng@arm.com>
22217 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
22219 2017-07-18 Richard Biener <rguenther@suse.de>
22221 PR tree-optimization/80620
22222 PR tree-optimization/81403
22223 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
22224 info when re-using a VN table entry.
22226 2017-07-18 Richard Biener <rguenther@suse.de>
22228 PR tree-optimization/81418
22229 * tree-vect-loop.c (vectorizable_reduction): Properly compute
22230 vectype_in. Verify that with lane-reducing reduction operations
22231 we have a single def-use cycle.
22233 2017-07-17 Carl Love <cel@us.ibm.com>
22235 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
22237 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22238 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
22239 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
22240 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
22241 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
22242 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
22243 VMULOSW): New enum "unspec" values.
22244 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
22245 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
22246 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
22247 altivec_vmulosw): New patterns.
22248 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
22249 VMULOSW): Add definitions.
22251 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
22253 * config/alpha/alpha.c: Include predict.h.
22255 2017-07-17 Yury Gribov <tetra2005@gmail.com>
22257 * tree-vrp.c (compare_assert_loc): Fix comparison function
22258 to return predictable results.
22260 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22262 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
22264 (subdi3): Likewise.
22265 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
22266 * doc/invoke.texi (mexpand-adddi): Update text.
22268 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22270 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
22271 that also clobbers the CC register. The old expand code is moved
22273 (*arc_clzsi2): ... here.
22274 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
22275 the CC register. The old expand code is moved to ...
22276 (arc_ctzsi2): ... here.
22278 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22280 * config/arc/arc.opt (mindexed-loads): Use initial value
22281 TARGET_INDEXED_LOADS_DEFAULT.
22282 (mauto-modify-reg): Use initial value
22283 TARGET_AUTO_MODIFY_REG_DEFAULT.
22284 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22285 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22286 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
22287 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
22289 2017-07-17 Martin Liska <mliska@suse.cz>
22292 * opts.c (finish_options): Do not allow -fgnu-tm
22293 w/ -fsanitize={kernel-,}address. Say sorry.
22295 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22298 * tree-loop-distribution.c (classify_partition): Only assert on
22299 numer of iterations.
22300 (merge_dep_scc_partitions): Delete prameter. Update function call.
22301 (distribute_loop): Remove code handling loop with unknown niters.
22302 (pass_loop_distribution::execute): Skip loop with unknown niters.
22304 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22307 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
22308 function sort_partitions_by_post_order.
22310 2017-07-17 Bin Cheng <bin.cheng@arm.com>
22312 PR tree-optimization/81374
22313 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
22314 the max index of basic blocks, rather than number of basic blocks.
22316 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22318 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
22320 (arc_legitimate_pic_operand_p): Likewise.
22321 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
22323 (arc_needs_pcl_p): Likewise.
22324 (arc_legitimate_pc_offset_p): Likewise.
22325 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
22326 function is also used in constrains.md.
22327 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
22328 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
22329 PLUS. Only return true/false in known cases, otherwise assert.
22330 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
22331 is already called in arc_legitimate_constant_p.
22332 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
22334 (LEGITIMATE_PIC_OPERAND_P): Use
22335 arc_raw_symbolic_reference_mentioned_p function.
22336 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
22341 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
22342 Andrew Burgess <andrew.burgess@embecosm.com>
22344 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
22345 (arc_return_address_register): New function.
22346 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
22347 (arc_handle_fndecl_attribute): Add naked attribute.
22348 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22349 (TARGET_WARN_FUNC_RETURN): Likewise.
22350 (arc_allocate_stack_slots_for_args): New function.
22351 (arc_warn_func_return): Likewise.
22352 (machine_function): Change type fn_type.
22353 (arc_compute_function_type): Consider new naked function type,
22354 change function return type.
22355 (arc_must_save_register): Adapt to handle new
22356 arc_compute_function_type's return type.
22357 (arc_expand_prologue): Likewise.
22358 (arc_expand_epilogue): Likewise.
22359 (arc_return_address_regs): Delete.
22360 (arc_return_address_register): New function.
22361 (arc_epilogue_uses): Use above function.
22362 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
22363 (arc_function_type): Change encoding, add naked type.
22364 (ARC_INTERRUPT_P): Change to handle the new encoding.
22365 (ARC_FAST_INTERRUPT_P): Likewise.
22366 (ARC_NORMAL_P): Define.
22367 (ARC_NAKED_P): Likewise.
22368 (arc_compute_function_type): Delete prototype.
22369 * config/arc/arc.md (in_ret_delay_slot): Use
22370 arc_return_address_register function.
22371 (simple_return): Likewise.
22372 (p_return_i): Likewise.
22374 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22376 PR tree-optimization/81428
22377 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
22378 can't be built for those types.
22380 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22382 Remove stuff dead since r239246.
22384 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
22385 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
22386 (avr_inform_devices): Remove dead stuff.
22388 2017-07-17 Tamar Christina <tamar.christina@arm.com>
22390 * config/arm/arm_neon.h: Fix softp typo.
22392 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22394 PR tree-optimization/81365
22395 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
22396 aggregate moves onto bb predecessor edges, make sure there are no
22397 loads that could alias the lhs in between the start of bb and the
22400 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
22403 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
22404 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
22405 [LSHIFTRT, outer_code = TRUNCATE]: Same.
22407 2017-07-17 Jakub Jelinek <jakub@redhat.com>
22409 PR tree-optimization/81396
22410 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
22411 (init_symbolic_number): Initialize it to 1.
22412 (perform_symbolic_merge): Add n_ops from both operands into the new
22414 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
22415 without base_addr as useless if they need more than one operation.
22416 (bswap_replace): Handle !bswap case for NULL base_addr.
22418 2017-07-17 Tom de Vries <tom@codesourcery.com>
22421 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
22424 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
22426 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
22427 conditional builtin define __FIX_LEON3FT_B2BST.
22429 2017-07-17 Daniel Cederman <cederman@gaisler.com>
22431 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
22432 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
22435 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22437 PR rtl-optimization/81424
22438 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
22439 to remove potential trapping from operands if -fnon-call-exceptions.
22441 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22443 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
22444 profile_proability for scalling.
22445 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
22447 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22449 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
22451 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22453 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
22454 fixpoint arithmetics.
22456 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22458 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
22459 fixpoint arithmetics.
22461 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22463 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
22464 fixpoint arithmetics.
22466 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
22468 * profile-count.h (profile_probability::from_reg_br_prob_note,
22469 profile_probability::to_reg_br_prob_note): New functions.
22470 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
22471 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
22472 * predict.c (probability_reliable_p): Update.
22473 (edge_probability_reliable_p): Update.
22474 (br_prob_note_reliable_p): Update.
22475 (invert_br_probabilities): Update.
22476 (add_reg_br_prob_note): New function.
22477 (combine_predictions_for_insn): Update.
22478 * asan.c (asan_clear_shadow): Update.
22479 * cfgbuild.c (compute_outgoing_frequencies): Update.
22480 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22481 (update_br_prob_note): Update.
22482 (rtl_verify_edges): Update.
22483 (purge_dead_edges): Update.
22484 (fixup_reorder_chain): Update.
22485 * emit-rtl.c (try_split): Update.
22486 * ifcvt.c (cond_exec_process_insns): Update.
22487 (cond_exec_process_if_block): Update.
22488 (dead_or_predicable): Update.
22489 * internal-fn.c (expand_addsub_overflow): Update.
22490 (expand_neg_overflow): Update.
22491 (expand_mul_overflow): Update.
22492 * loop-doloop.c (doloop_modify): Update.
22493 * loop-unroll.c (compare_and_jump_seq): Update.
22494 * optabs.c (emit_cmp_and_jump_insn_1): Update.
22495 * predict.h: Update.
22496 * reorg.c (mostly_true_jump): Update.
22498 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
22499 * config/alpha/alpha.c (emit_unlikely_jump): Update.
22500 * config/arc/arc.c: (emit_unlikely_jump): Update.
22501 * config/arm/arm.c: (emit_unlikely_jump): Update.
22502 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
22503 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
22504 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
22505 (ix86_print_operand): Update.
22506 (ix86_split_fp_branch): Update.
22507 (predict_jump): Update.
22508 * config/ia64/ia64.c (ia64_print_operand): Update.
22509 * config/mmix/mmix.c (mmix_print_operand): Update.
22510 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
22511 (rs6000_expand_split_stack_prologue): Update.
22512 * config/rs6000/rs6000.c: Update.
22513 * config/s390/s390.c (s390_expand_vec_strlen): Update.
22514 (s390_expand_vec_movstr): Update.
22515 (s390_expand_cs_tdsi): Update.
22516 (s390_expand_split_stack_prologue): Update.
22517 * config/sh/sh.c (sh_print_operand): Update.
22518 (expand_cbranchsi4): Update.
22519 (expand_cbranchdi4): Update.
22520 * config/sparc/sparc.c (output_v9branch): Update.
22521 * config/spu/spu.c (get_branch_target): Update.
22522 (ea_load_store_inline): Update.
22523 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
22524 * config/tilepro/tilepro.c: Update.
22526 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
22528 * gimplify.c (mostly_copy_tree_r): Revert latest change.
22529 (gimplify_save_expr): Likewise.
22531 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22533 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
22535 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22537 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
22539 * timevar.def (TV_IPA_FNSUMMARY): Define.
22541 2017-07-16 Daniel Cederman <cederman@gaisler.com>
22543 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
22544 to back store errata sensitive sequence from being generated.
22545 (sqrtdf2_fix): Likewise.
22547 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22549 * tree-ssa-threadupdate.c (compute_path_counts,
22550 update_joiner_offpath_counts): Use profile_probability.
22552 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
22555 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22557 * config/arm/arm-c.c (arm_cpu_builtins): Define
22558 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
22560 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
22562 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22563 array entries to represent __ieee128 versions of the
22564 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
22565 scalar_extract_sig, and scalar_insert_exp built-in functions.
22566 (altivec_resolve_overloaded_builtin): Add special case handling
22567 for the __builtin_scalar_insert_exp function, as represented by
22568 the P9V_BUILTIN_VEC_VSIEDP constant.
22569 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
22570 exponent support for __ieee128 argument.
22571 (VSESQP): Add scalar extract signature support for __ieee128
22573 (VSTDCNQP): Add scalar test negative support for __ieee128
22575 (VSIEQP): Add scalar insert exponent support for __int128 argument
22576 with __ieee128 result.
22577 (VSIEQPF): Add scalar insert exponent support for __ieee128
22578 argument with __ieee128 result.
22579 (VSTDCQP): Add scalar test data class support for __ieee128
22581 (VSTDCNQP): Add overload support for scalar test negative with
22582 __ieee128 argument.
22583 (VSTDCQP): Add overload support for scalar test data class
22584 __ieee128 argument.
22585 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
22586 UNSPEC_VSX_SXSIGDP.
22587 (UNSPEC_VSX_SIEXPQP): New constant.
22588 (xsxexpqp): New insn for VSX scalar extract exponent quad
22590 (xsxsigqp): New insn for VSX scalar extract significand quad
22592 (xsiexpqpf): New insn for VSX scalar insert exponent quad
22593 precision with floating point argument.
22594 (xststdcqp): New expand for VSX scalar test data class quad
22596 (xststdcnegqp): New expand for VSX scalar test negative quad
22598 (xststdcqp): New insn to match expansions for VSX scalar test data
22599 class quad precision and VSX scalar test negative quad precision.
22600 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
22601 special case operand checking to enforce that second operand of
22602 VSX scalar test data class with quad precision argument is a 7-bit
22604 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22605 prototypes and descriptions of __ieee128 versions of
22606 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
22607 scalar_test_data_class, and scalar_test_neg built-in functions.
22609 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22611 PR tree-optimization/81162
22612 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
22613 replace a negate with an add.
22615 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
22617 * doc/invoke.texi (arm/-mcpu): Document +crypto.
22619 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22621 * config/arm/arm-c.c (arm_cpu_builtins): Define
22622 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
22624 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22626 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
22627 (armv8-r): Set ARM Cortex-R52 as default CPU.
22628 * config/arm/arm-tables.opt: Regenerate.
22629 * config/arm/arm-tune.md: Regenerate.
22630 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
22632 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
22633 extension for -mcpu=cortex-r52.
22635 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
22637 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
22638 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
22639 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
22640 (fp-armv8): Define it as FP_ARMv8 only.
22641 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
22642 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
22644 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
22645 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
22646 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
22647 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
22648 than TARGET_FPU_ARMV8.
22649 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
22650 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
22651 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
22653 * config/arm/neon.md (neon_vrint): Likewise.
22654 (neon_vcvt): Likewise.
22655 (neon_<fmaxmin_op><mode>): Likewise.
22656 (<fmaxmin><mode>3): Likewise.
22657 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
22658 * config/arm/predicates.md (arm_cond_move_operator): Check against
22659 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
22661 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
22663 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
22664 to top of function.
22666 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22668 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
22669 loop in comment with memset.
22671 2017-07-14 Martin Liska <mliska@suse.cz>
22673 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
22674 * dwarf2out.c (is_java): Remove the function.
22675 (output_pubname): Remove usage of the function.
22676 (lower_bound_default): Remove usage of DW_LANG_Java.
22677 (gen_compile_unit_die): Likewise.
22678 * gcc.c: Remove compiler defaults for .java and .zip files.
22679 * gimple-expr.c (remove_suffix): Change as there's no longer
22680 extension than 4-letter one.
22681 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
22682 (gimplify_save_expr): Likewise.
22683 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
22684 as it's possible even for other languages than Java.
22685 * langhooks.h (struct lang_hooks): Remove Java from a comment.
22686 * lto-opts.c (lto_write_options): Remove reference to Java.
22687 * opts.c (strip_off_ending): Update file extension handling.
22688 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
22689 * tree-eh.c (lower_resx): Likewise.
22690 * tree.c (free_lang_data_in_type): Remove dead code.
22691 (find_decls_types_r): Likewise.
22692 (build_common_builtin_nodes): Remove Java from a comment.
22693 (verify_type): Remove dead code.
22694 * varasm.c (assemble_external): Remove Java from a comment.
22696 2017-07-14 Martin Liska <mliska@suse.cz>
22698 * opts.c (finish_options): Add quotes.
22699 (common_handle_option): Likewise.
22701 2017-07-14 Martin Liska <mliska@suse.cz>
22703 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
22704 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
22705 Remove N_SO_PASCAL.
22706 * dwarf2out.c (lower_bound_default): Do not handle
22708 (gen_compile_unit_die): Likewise.
22709 * gcc.c: Remove default extension binding for GNU Pascal.
22710 * stmt.c: Remove Pascal language from a comment.
22711 * xcoffout.c: Likewise.
22713 2017-07-13 David Malcolm <dmalcolm@redhat.com>
22716 * diagnostic-show-locus.c (fixit_cmp): New function.
22717 (layout::layout): Sort m_fixit_hints.
22718 (column_range::column_range): Assert that the values are valid.
22719 (struct char_span): New struct.
22720 (correction::overwrite): New method.
22721 (struct source_line): New struct.
22722 (line_corrections::add_hint): Add assertions. Reimplement memcpy
22723 calls in terms of classes source_line and char_span, and
22724 correction::overwrite.
22725 (selftest::test_overlapped_fixit_printing_2): New function.
22726 (selftest::diagnostic_show_locus_c_tests): Call it.
22728 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
22730 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
22731 early if there is no lhs.
22733 2017-07-13 Martin Liska <mliska@suse.cz>
22735 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
22736 (gen_reference_type_die): Likewise.
22737 * stor-layout.c: Remove Pascal-related comment.
22739 2017-07-13 Martin Liska <mliska@suse.cz>
22741 * opts.c (finish_options): Add quotes to error messages.
22742 (parse_sanitizer_options): Likewise.
22744 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
22746 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
22748 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
22750 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
22752 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22754 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
22756 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
22758 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22761 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
22762 provides the hardware capability bits, define the macro
22763 __BUILTIN_CPU_SUPPORTS__.
22764 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
22765 if GLIBC does not provide the hardware capability bits. Add a
22766 gcc_unreachable call if the built-in cpu function is neither
22767 __builtin_cpu_is nor __builtin_cpu_supports.
22768 (rs6000_get_function_versions_dispatcher): Change the warning
22769 that an old GLIBC is used which does not export the capability
22770 bits to be an error.
22771 * doc/extend.texi (target_clones attribute): Document the
22772 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
22773 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
22774 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
22775 the macros defined by GCC if the newer GLIBC is available.
22777 2017-07-12 Jeff Law <law@redhat.com>
22779 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
22780 remaining includes slightly.
22781 * config/riscv/riscv-builtins.c: Include profile-count.h.
22783 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22786 * config/avr/avr.c (avr_set_current_function): In diagnostic
22787 messages: Quote keywords and (parts of) identifiers.
22788 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
22791 2017-07-12 Carl Love <cel@us.ibm.com>
22793 * config/rs6000/rs6000-c.c: Add support for built-in functions
22794 vector bool char vec_revb (vector bool char);
22795 vector bool short vec_revb (vector short char);
22796 vector bool int vec_revb (vector bool int);
22797 vector bool long long vec_revb (vector bool long long);
22798 * doc/extend.texi: Update the built-in documentation file for the
22799 new built-in functions.
22801 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22803 * config/s390/s390.md: Remove movcc splitter.
22805 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22807 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
22808 load/store on condition.
22810 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22813 * config/avr/avr.c (avr_encode_section_info)
22814 [progmem && !TREE_READONLY]: Error if progmem object needs
22817 2017-07-11 Michael Collison <michael.collison@arm.com>
22819 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
22822 2017-07-11 Carl Love <cel@us.ibm.com>
22824 * config/rs6000/rs6000-c.c: Add support for builtins
22825 vector unsigned int vec_parity_lsbb (vector signed int);
22826 vector unsigned int vec_parity_lsbb (vector unsigned int);
22827 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
22828 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
22829 vector unsigned long long vec_parity_lsbb (vector signed long long);
22830 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
22831 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
22832 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
22833 * doc/extend.texi: Update the built-in documentation file for the
22834 new built-in functions.
22836 2017-07-11 David Malcolm <dmalcolm@redhat.com>
22838 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
22839 (layout::m_primary_loc): New field.
22840 (layout::layout): Initialize new field. Move location filtering
22841 logic from here to...
22842 (layout::maybe_add_location_range): ...this new method. Add
22843 support for filtering to just the lines already specified by other
22845 (layout::will_show_line_p): New method.
22846 (gcc_rich_location::add_location_if_nearby): New method.
22847 (selftest::test_add_location_if_nearby): New test function.
22848 (selftest::diagnostic_show_locus_c_tests): Call it.
22849 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
22852 2017-07-11 Tom de Vries <tom@codesourcery.com>
22854 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
22855 (bb_first_real_insn): New function.
22856 (nvptx_single): Add extra initialization of broadcasted condition
22859 2017-07-11 Nathan Sidwell <nathan@acm.org>
22861 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
22863 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
22865 * doc/extend.texi (AVR Function Attributes): Remove weblink to
22866 Binutils doc as TEXI will mess them up.
22867 * doc/invoke.texi (AVR Options): Same here.
22869 2017-07-11 Daniel Cederman <cederman@gaisler.com>
22871 * config/sparc/sparc.opt (mfix-ut700): New option.
22872 (mfix-gr712rc): Likewise.
22873 (sparc_fix_b2bst): New variable.
22874 * doc/invoke.texi (SPARC options): Document them.
22875 (ARM options): Fix warnings.
22876 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
22877 instructions to prevent sequences that can trigger the store-store
22878 errata for certain LEON3FT processors.
22879 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
22880 (sparc_option_override): Set sparc_fix_b2bst appropriately.
22881 * config/sparc/sparc.md (fix_b2bst): New attribute.
22882 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
22884 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
22887 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
22889 (*rsqrtsf2_sse): Ditto.
22891 (div<mode>3): Macroize insn from divdf3 and divsf3
22892 using MODEF mode iterator.
22894 2017-07-10 Martin Sebor <msebor@redhat.com>
22896 PR tree-optimization/80397
22897 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
22898 instead of testing for equality to INTEGER_TYPE.
22900 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
22902 * config.gcc: Remove uclibc from arc target spec.
22904 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
22906 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
22908 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22911 * lto-wrapper.c (remove_option): New function.
22912 (merge_and_complain): Merge PIC/PIE options more realistically.
22914 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
22916 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
22920 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
22921 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
22922 * config.in: Regenerate.
22923 * configure: Regenerate.
22924 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
22925 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
22926 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
22927 (TARGET_GASISR_PROLOGUES): ...target mask.
22928 * common/config/avr/avr-common.c
22929 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
22930 Set -mgas-isr-prologues.
22931 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
22932 INSERT_PASS_BEFORE for it.
22933 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
22934 * config/avr/avr.c (avr_option_override)
22935 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
22936 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
22937 (avr_attribute_table) <no_gccisr>: Add new function attribute.
22938 (avr_set_current_function) <is_no_gccisr>: Init machine field.
22939 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
22941 (make_avr_pass_pre_proep): New function.
22942 (emit_push_sfr) <treg>: Add argument to function and use it
22943 instead of TMP_REG.
22944 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
22945 and set machine->gasisr.yes.
22946 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
22947 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
22948 __gcc_isr.n_pushed to .L__stack_usage.
22949 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
22950 (avr_asm_final_postscan_insn): ...this new static function.
22951 * config/avr/avr.h (machine_function)
22952 <is_no_gccisr, use_L__stack_usage>: New fields.
22953 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
22954 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
22955 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
22956 (gasisr, *gasisr): New expander and insn.
22957 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
22958 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
22959 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
22961 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
22963 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
22966 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
22968 Move jump-tables out of .text again.
22971 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
22972 (ASM_OUTPUT_ADDR_VEC): New function.
22973 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
22974 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
22975 INSN_ADDRESSes as asm comment.
22976 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
22977 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
22978 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
22979 * config/avr/avr.md (*tablejump): Adjust comment.
22980 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
22981 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
22983 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
22984 (avr_output_addr_vec): New proto.
22985 (avr_log_t) <insn_addresses>: New field.
22987 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
22990 * config/i386/i386.c (ix86_function_arg_advance): Set
22991 outgoing_args_on_stack to true if there are outgoing arguments
22993 (ix86_function_arg): Likewise.
22994 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
22995 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
22996 * config/i386/i386.h (machine_function): Add
22997 outgoing_args_on_stack.
22999 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
23001 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23002 supporting pthreds.
23003 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
23005 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
23007 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
23008 (REAL_H): Remove $(MACHMODE_H).
23009 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
23011 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
23012 $(MACHMODE_H) and double-int.h.
23013 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
23015 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
23018 2017-07-07 Andrew Pinski <apinski@cavium.com>
23020 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
23021 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
23023 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23025 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
23026 Add warning if GCC was not configured to link against a GLIBC that
23027 exports the hardware capability bits.
23028 (make_resolver_func): Make resolver function private and not a
23029 COMDAT function. Create the name with clone_function_name instead
23030 of make_unique_name.
23033 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
23034 correct operand in doing the split.
23036 2017-07-07 Carl Love <cel@us.ibm.com>
23038 * config/rs6000/rs6000-c: Add support for built-in function
23039 vector unsigned short vec_pack_to_short_fp32 (vector float,
23041 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
23042 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
23043 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
23044 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
23045 (convert_4f32_8i16): Add define_expand.
23046 * doc/extend.texi: Update the built-in documentation file for the
23047 new built-in function.
23049 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23051 * config/sparc/m8.md: New file.
23052 * config/sparc/sparc.md: Include m8.md.
23054 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23056 * config/sparc/sparc.opt: New option -mvis4b.
23057 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
23058 (sparc_option_override): Handle VIS4B.
23059 (enum sparc_builtins): Define
23060 SPARC_BUILTIN_DICTUNPACK{8,16,32},
23061 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
23062 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
23063 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
23064 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
23065 (check_constant_argument): New function.
23066 (sparc_vis_init_builtins): Define builtins
23067 __builtin_vis_dictunpack{8,16,32},
23068 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
23069 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
23070 __builtin_vis_fpcmpde{8,16,32}shl and
23071 __builtin_vis_fpcmpur{8,16,32}shl.
23072 (sparc_expand_builtin): Check that the constant operands to
23073 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
23074 constant and in range.
23075 * config/sparc/sparc-c.c (sparc_target_macros): Handle
23077 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
23078 (SPARC_IMM5_P): Likewise.
23079 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
23080 (enabled): Handle vis4b.
23081 (UNSPEC_DICTUNPACK): New unspec.
23082 (UNSPEC_FPCMPSHL): Likewise.
23083 (UNSPEC_FPUCMPSHL): Likewise.
23084 (UNSPEC_FPCMPDESHL): Likewise.
23085 (UNSPEC_FPCMPURSHL): Likewise.
23086 (cpu_feature): New CPU feature `vis4b'.
23087 (dictunpack{8,16,32}): New insns.
23088 (FPCSMODE): New mode iterator.
23089 (fpcscond): New code iterator.
23090 (fpcsucond): Likewise.
23091 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
23092 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
23093 (fpcmpde{8,16,32}{si,di}shl): Likewise.
23094 (fpcmpur{8,16,32}{si,di}shl): Likewise.
23095 * config/sparc/constraints.md: Define constraints `q' for unsigned
23096 2-bit integer constants and `t' for unsigned 5-bit integer
23098 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
23100 (imm5_operand_dictunpack16): Likewise.
23101 (imm5_operand_dictunpack32): Likewise.
23102 (imm2_operand): Likewise.
23103 * doc/invoke.texi (SPARC Options): Document -mvis4b.
23104 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
23105 ditunpack* and fpcmp*shl builtins.
23107 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23109 * config.gcc: Handle m8 in --with-{cpu,tune} options.
23110 * config.in: Add HAVE_AS_SPARC6 define.
23111 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
23113 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
23115 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
23116 (CPP_CPU_SPEC): Handle m8.
23117 (ASM_CPU_SPEC): Likewise.
23118 * config/sparc/sparc-opts.h (enum processor_type): Add
23120 * config/sparc/sparc.c (m8_costs): New struct.
23121 (sparc_option_override): Handle TARGET_CPU_m8.
23122 (sparc32_initialize_trampoline): Likewise.
23123 (sparc64_initialize_trampoline): Likewise.
23124 (sparc_issue_rate): Likewise.
23125 (sparc_register_move_cost): Likewise.
23126 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
23127 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
23128 (ASM_CPU64_DEFAULT_SPEC): Likewise.
23129 (CPP_CPU_SPEC): Handle M8.
23130 (ASM_CPU_SPEC): Likewise.
23131 (AS_M8_FLAG): Define.
23132 * config/sparc/sparc.md: Add m8 to the cpu attribute.
23133 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
23134 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
23136 * configure: Regenerate.
23137 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
23140 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23142 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
23144 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
23145 ("*movdi_insn_sp32"): Do not set v3pipe.
23146 ("*movsi_insn"): Likewise.
23147 ("*movdi_insn_sp64"): Likewise.
23148 ("*movsf_insn"): Likewise.
23149 ("*movdf_insn_sp32"): Likewise.
23150 ("*movdf_insn_sp64"): Likewise.
23151 ("*zero_extendsidi2_insn_sp64"): Likewise.
23152 ("*sign_extendsidi2_insn"): Likewise.
23153 ("*mov<VM32:mode>_insn"): Likewise.
23154 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23155 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23156 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23157 ("<vlop:code><VL:mode>3"): Likewise.
23158 ("*not_<vlop:code><VL:mode>3"): Likewise.
23159 ("*nand<VL:mode>_vis"): Likewise.
23160 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
23161 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
23162 ("one_cmpl<VL:mode>2"): Likewise.
23163 ("faligndata<VM64:mode>_vis"): Likewise.
23164 ("alignaddrsi_vis"): Likewise.
23165 ("alignaddrdi_vis"): Likweise.
23166 ("alignaddrlsi_vis"): Likewise.
23167 ("alignaddrldi_vis"): Likewise.
23168 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23169 ("bmaskdi_vis"): Likewise.
23170 ("bmasksi_vis"): Likewise.
23171 ("bshuffle<VM64:mode>_vis"): Likewise.
23172 ("cmask8<P:mode>_vis"): Likewise.
23173 ("cmask16<P:mode>_vis"): Likewise.
23174 ("cmask32<P:mode>_vis"): Likewise.
23175 ("pdistn<P:mode>_vis"): Likewise.
23176 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23178 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23180 * config/sparc/sparc.md ("subtype"): New insn attribute.
23181 ("*wrgsr_sp64"): Set insn subtype.
23182 ("*rdgsr_sp64"): Likewise.
23183 ("alignaddrsi_vis"): Likewise.
23184 ("alignaddrdi_vis"): Likewise.
23185 ("alignaddrlsi_vis"): Likewise.
23186 ("alignaddrldi_vis"): Likewise.
23187 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
23188 ("fexpand_vis"): Likewise.
23189 ("fpmerge_vis"): Likewise.
23190 ("faligndata<VM64:mode>_vis"): Likewise.
23191 ("bshuffle<VM64:mode>_vis"): Likewise.
23192 ("cmask8<P:mode>_vis"): Likewise.
23193 ("cmask16<P:mode>_vis"): Likewise.
23194 ("cmask32<P:mode>_vis"): Likewise.
23195 ("fchksm16_vis"): Likewise.
23196 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
23197 ("fmean16_vis"): Likewise.
23198 ("fp<plusminus_insn>64_vis"): Likewise.
23199 ("<plusminus_insn>v8qi3"): Likewise.
23200 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
23201 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
23202 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
23203 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
23204 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
23205 ("*movqi_insn"): Likewise.
23206 ("*movhi_insn"): Likewise.
23207 ("*movsi_insn"): Likewise.
23208 ("movsi_pic_gotdata_op"): Likewise.
23209 ("*movdi_insn_sp32"): Likewise.
23210 ("*movdi_insn_sp64"): Likewise.
23211 ("movdi_pic_gotdata_op"): Likewise.
23212 ("*movsf_insn"): Likewise.
23213 ("*movdf_insn_sp32"): Likewise.
23214 ("*movdf_insn_sp64"): Likewise.
23215 ("*zero_extendhisi2_insn"): Likewise.
23216 ("*zero_extendqihi2_insn"): Likewise.
23217 ("*zero_extendqisi2_insn"): Likewise.
23218 ("*zero_extendqidi2_insn"): Likewise.
23219 ("*zero_extendhidi2_insn"): Likewise.
23220 ("*zero_extendsidi2_insn_sp64"): Likewise.
23221 ("ldfsr"): Likewise.
23222 ("prefetch_64"): Likewise.
23223 ("prefetch_32"): Likewise.
23224 ("tie_ld32"): Likewise.
23225 ("tie_ld64"): Likewise.
23226 ("*tldo_ldub_sp32"): Likewise.
23227 ("*tldo_ldub1_sp32"): Likewise.
23228 ("*tldo_ldub2_sp32"): Likewise.
23229 ("*tldo_ldub_sp64"): Likewise.
23230 ("*tldo_ldub1_sp64"): Likewise.
23231 ("*tldo_ldub2_sp64"): Likewise.
23232 ("*tldo_ldub3_sp64"): Likewise.
23233 ("*tldo_lduh_sp32"): Likewise.
23234 ("*tldo_lduh1_sp32"): Likewise.
23235 ("*tldo_lduh_sp64"): Likewise.
23236 ("*tldo_lduh1_sp64"): Likewise.
23237 ("*tldo_lduh2_sp64"): Likewise.
23238 ("*tldo_lduw_sp32"): Likewise.
23239 ("*tldo_lduw_sp64"): Likewise.
23240 ("*tldo_lduw1_sp64"): Likewise.
23241 ("*tldo_ldx_sp64"): Likewise.
23242 ("*mov<VM32:mode>_insn"): Likewise.
23243 ("*mov<VM64:mode>_insn_sp64"): Likewise.
23244 ("*mov<VM64:mode>_insn_sp32"): Likewise.
23246 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23248 * config/sparc/sparc.md ("type"): New insn type viscmp.
23249 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
23251 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
23252 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
23253 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
23254 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
23256 ("n7_vis_logical_11cycle"): Likewise.
23257 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
23258 * config/sparc/niagara2.md ("niag3_vis": Likewise.
23259 * config/sparc/niagara.md ("niag_vis"): Likewise.
23260 * config/sparc/ultra3.md ("us3_fga"): Likewise.
23261 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
23263 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
23265 * config/sparc/sparc.md: New instruction type `bmask'.
23266 (bmaskdi_vis): Use the `bmask' type.
23267 (bmasksi_vis): Likewise.
23268 * config/sparc/ultra3.md (us3_array): Likewise.
23269 * config/sparc/niagara7.md (n7_array): Likewise.
23270 * config/sparc/niagara4.md (n4_array): Likewise.
23271 * config/sparc/niagara2.md (niag2_vis): Likewise.
23272 (niag3_vis): Likewise.
23273 * config/sparc/niagara.md (niag_vis): Likewise.
23275 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23277 * ipa-comdats.c: Remove optimize check from gate.
23278 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
23279 for functions not optimized.
23280 (ipa_fn_summary_read): Skip optimize check.
23281 (ipa_fn_summary_write): Likewise.
23282 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
23284 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
23286 (can_inline_edge_p): Check flag_pcc_struct_return for match.
23287 (check_callers): Give up on caller which is not optimized.
23288 (inline_small_functions): Likewise.
23289 (ipa_inline): Do not give up when not optimizing.
23290 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
23291 away unoptimizes cdtors.
23292 (whole_program_function_and_variable_visibility): Do
23293 ipa_discover_readonly_nonaddressable_vars in LTO mode.
23294 * ipa.c (process_references): Do not check optimize.
23295 (symbol_table::remove_unreachable_nodes): Update optimize check.
23296 (set_writeonly_bit): Update optimize check.
23297 (pass_ipa_cdtor_merge::gate): Do not check optimize.
23298 (pass_ipa_single_use::gate): Remove.
23300 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23302 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
23303 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
23304 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
23305 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
23306 permute_load, permute_store, adjust_extract, adjust_splat,
23307 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
23308 replace_swap_with_copy, dump_swap_insn_table,
23309 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
23310 recombine_lvx_pattern, recombine_stvx_pattern,
23311 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
23312 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
23313 to file rs6000-p8swap.c.
23314 * config/rs6000/rs6000-p8swap.c: New file.
23315 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
23316 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
23317 and rs6000*-*-* targets.
23319 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23321 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
23323 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23325 * lto-wrapper.c (merge_and_complain): Do not merge
23326 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
23327 fsigned_zeros, ftrapping_math, fwrapv.
23328 (append_compiler_options): Do not track these options.
23329 (append_linker_options): Likewie
23331 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23333 * cgraphunit.c (cgraph_node::finalize_function): When
23334 !flag_toplevel_reorde set no_reorder flag.
23335 (varpool_node::finalize_decl): Likewise.
23336 (symbol_table::compile): Drop no toplevel reorder path.
23338 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23340 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
23341 edges; zero probability is not better than uninitialized.
23343 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23345 * asan.h (asan_sanitize_allocas_p): Declare.
23346 * asan.c (asan_sanitize_allocas_p): New function.
23347 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
23348 (handle_builtin_alloca): Likewise.
23349 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
23350 if !asan_sanitize_allocas_p.
23351 * params.def (asan-instrument-allocas): Add new option.
23352 * params.h (ASAN_PROTECT_ALLOCAS): Define.
23353 * opts.c (common_handle_option): Disable allocas sanitization for
23356 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
23358 * asan.c: Include gimple-fold.h.
23359 (get_last_alloca_addr): New function.
23360 (handle_builtin_stackrestore): Likewise.
23361 (handle_builtin_alloca): Likewise.
23362 (asan_emit_allocas_unpoison): Likewise.
23363 (get_mem_refs_of_builtin_call): Add new parameter, remove const
23364 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
23365 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
23366 (instrument_builtin_call): Pass gimple iterator to
23367 get_mem_refs_of_builtin_call.
23368 (last_alloca_addr): New global.
23369 * asan.h (asan_emit_allocas_unpoison): Declare.
23370 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
23371 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
23372 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
23373 if function calls alloca.
23374 * gimple-fold.c (replace_call_with_value): Remove static keyword.
23375 * gimple-fold.h (replace_call_with_value): Declare.
23376 * internal-fn.c: Include asan.h.
23377 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
23378 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
23380 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23382 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
23383 (C_SELFTEST_FLAGS): New.
23384 (CPP_SELFTEST_FLAGS): New.
23385 (SELFTEST_DEPS): New, from deps of s-selftest.
23386 (C_SELFTEST_DEPS): New, from deps of s-selftest.
23387 (CPP_SELFTEST_DEPS): New.
23388 (selftest): Add dependency on s-selftest-c++.
23389 (s-selftest): Rename to...
23390 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
23391 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
23392 than SELFTEST_FLAGS.
23393 (selftest-gdb): Rename to...
23394 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
23396 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
23397 (selftest-valgrind): Rename to...
23398 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
23400 (selftest-valgrind): Reintroduce as an alias for
23401 selftest-c-valgrind.
23402 (s-selftest-c++): New.
23403 (selftest-c++-gdb): New.
23404 (selftest-c++-valgrind): New.
23406 2017-07-06 Olivier Hainque <hainque@adacore.com>
23408 * gcc.c (process_command): When deciding if undefined variables
23409 should be ignored when processing specs, accept "gcc -v" as well.
23411 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
23413 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
23414 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
23416 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
23418 * config/arm/arm-cpus.in (armv8-r): Add new entry.
23419 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
23420 * config/arm/arm-tables.opt: Regenerate.
23421 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
23423 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
23425 2017-07-06 Carl Love <cel@us.ibm.com>
23427 * ChangeLog: Clean up from mid air collision
23429 2017-07-06 Carl Love <cel@us.ibm.com>
23431 * config/rs6000/rs6000-c.c: Add support for built-in functions
23432 vector signed int vec_subc (vector signed int, vector signed int);
23433 vector signed __int128 vec_subc (vector signed __int128,
23434 vector signed __int128);
23435 vector unsigned __int128 vec_subc (vector unsigned __int128,
23436 vector unsigned __int128);
23437 vector signed int vec_sube (vector signed int, vector signed int,
23438 vector signed int);
23439 vector unsigned int vec_sube (vector unsigned int,
23440 vector unsigned int,
23441 vector unsigned int);
23442 vector signed __int128 vec_sube (vector signed __int128,
23443 vector signed __int128,
23444 vector signed__int128);
23445 vector unsigned __int128 vec_sube (vector unsigned __int128,
23446 vector unsigned __int128,
23447 vector unsigned __int128);
23448 vector signed int vec_subec (vector signed int, vector signed int,
23449 vector signed int);
23450 vector unsigned int vec_subec (vector unsigned int,
23451 vector unsigned int,
23452 vector unsigned int);
23453 vector signed __int128 vec_subec (vector signed __int128,
23454 vector signed __int128,
23455 vector signed__int128);
23456 vector unsigned __int128 vec_subec (vector unsigned __int128,
23457 vector unsigned __int128,
23458 vector unsigned __int128);
23459 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
23460 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
23461 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
23462 BU_ALTIVEC_OVERLOAD_X definitions.
23463 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
23464 * doc/extend.texi: Update the built-in documentation file for the new
23465 built-in functions.
23467 2017-07-06 David Malcolm <dmalcolm@redhat.com>
23470 * diagnostic-show-locus.c (layout::layout): Use start and finish
23471 spelling location for the start and finish of each range.
23472 * genmatch.c (linemap_client_expand_location_to_spelling_point):
23473 Add unused aspect param.
23474 * input.c (expand_location_1): Add "aspect" param, and use it
23475 to access the correct part of the location.
23476 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
23478 (expand_location_to_spelling_point): Likewise.
23479 (linemap_client_expand_location_to_spelling_point): Add "aspect"
23480 param, and pass it to expand_location_1.
23482 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
23484 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
23485 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
23486 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
23487 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
23488 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
23489 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
23490 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
23491 _mm_maskz_getmant_ss): New intrinsics.
23492 (__builtin_ia32_getexpss128_mask): Changed to ...
23493 __builtin_ia32_getexpss128_round ... this.
23494 (__builtin_ia32_getexpsd128_mask): Changed to ...
23495 __builtin_ia32_getexpsd128_round ... this.
23496 * config/i386/i386-builtin-types.def
23497 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
23498 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
23499 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
23500 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
23501 __builtin_ia32_getmantss_mask_round): New builtins.
23502 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
23503 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
23504 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
23505 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
23506 * config/i386/sse.md
23507 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
23508 avx512f_sgetexp<mode><mask_scalar_name>
23509 <round_saeonly_scalar_name> ... this.
23510 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
23511 %0, %1, %2<round_saeonly_op3>}): Changed to ...
23512 vgetexp<ssescalarmodesuffix>
23513 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23514 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
23515 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
23516 avx512f_vgetmant<mode><mask_scalar_name>
23517 <round_saeonly_scalar_name> ... this.
23518 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
23519 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
23520 vgetmant<ssescalarmodesuffix>
23521 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
23522 %0<mask_scalar_operand4>, %1, %2
23523 <round_saeonly_scalar_mask_op4>, %3} ... this.
23524 * config/i386/subst.md (mask_scalar_operand4,
23525 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
23526 round_saeonly_scalar_nimm_predicate): New subst attributes.
23528 2017-07-06 Julia Koval <julia.koval@intel.com>
23530 * config/i386/i386.c (ix86_erase_embedded_rounding):
23531 Remove code for old rounding pattern.
23533 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
23535 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
23537 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
23539 * doc/sourcebuild.texi (Test Directives, Variants of
23540 dg-require-support): Add documentation for dg-require-stack-check.
23542 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
23544 * config/i386/subst.md (mask_scalar, round_scalar,
23545 round_saeonly_scalar): New meta-templates.
23546 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
23547 round_scalar_mask_operand3, round_scalar_mask_op3,
23548 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
23549 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
23550 round_saeonly_scalar_constraint,
23551 round_saeonly_scalar_prefix): New subst attribute.
23552 * config/i386/sse.md
23553 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
23554 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
23555 <round_scalar_name> ... this.
23556 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
23557 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
23558 <round_scalar_name> ... this.
23559 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
23560 <sse>_vm<code><mode>3<mask_scalar_name>
23561 <round_saeonly_scalar_name> ... this.
23562 (v<plusminus_mnemonic><ssescalarmodesuffix>
23563 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23564 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23565 v<plusminus_mnemonic><ssescalarmodesuffix>
23566 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23567 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23568 (v<multdiv_mnemonic><ssescalarmodesuffix>
23569 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
23570 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
23571 v<multdiv_mnemonic><ssescalarmodesuffix>
23572 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23573 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
23574 (v<maxmin_float><ssescalarmodesuffix>
23575 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
23576 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
23577 v<maxmin_float><ssescalarmodesuffix>
23578 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
23579 %0<mask_scalar_operand3>, %1, %<iptr>2
23580 <round_saeonly_scalar_mask_op3>} ... this.
23582 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
23584 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
23585 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
23587 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
23588 Alan Hayward <alan.hayward@arm.com>
23589 David Sherwood <david.sherwood@arm.com>
23591 * combine.c (simplify_if_then_else): Remove "enum" before
23593 * compare-elim.c (can_eliminate_compare): Likewise.
23594 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
23596 (aarch64_lookup_simd_builtin_type): Likewise.
23597 (aarch64_simd_builtin_type): Likewise.
23598 (aarch64_init_simd_builtin_types): Likewise.
23599 (aarch64_simd_expand_args): Likewise.
23600 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
23602 (aarch64_reverse_mask): Likewise.
23603 (aarch64_simd_emit_reg_reg_move): Likewise.
23604 (aarch64_gen_adjusted_ldpstp): Likewise.
23605 (aarch64_ccmp_mode_to_code): Likewise.
23606 (aarch64_operands_ok_for_ldpstp): Likewise.
23607 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23608 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
23610 (aarch64_min_divisions_for_recip_mul): Likewise.
23611 (aarch64_reassociation_width): Likewise.
23612 (aarch64_get_condition_code_1): Likewise.
23613 (aarch64_simd_emit_reg_reg_move): Likewise.
23614 (aarch64_simd_attr_length_rglist): Likewise.
23615 (aarch64_reverse_mask): Likewise.
23616 (aarch64_operands_ok_for_ldpstp): Likewise.
23617 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
23618 (aarch64_gen_adjusted_ldpstp): Likewise.
23619 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
23621 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
23622 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
23623 (arm_lookup_simd_builtin_type): Likewise.
23624 (arm_simd_builtin_type): Likewise.
23625 (arm_init_simd_builtin_types): Likewise.
23626 (arm_expand_builtin_args): Likewise.
23627 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
23628 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
23629 (ft32_setup_incoming_varargs): Likewise.
23630 (ft32_function_arg): Likewise.
23631 (ft32_function_arg_advance): Likewise.
23632 (ft32_pass_by_reference): Likewise.
23633 (ft32_arg_partial_bytes): Likewise.
23634 (ft32_valid_pointer_mode): Likewise.
23635 (ft32_addr_space_pointer_mode): Likewise.
23636 (ft32_addr_space_legitimate_address_p): Likewise.
23637 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
23639 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
23640 (ix86_emit_outlined_ms2sysv_restore): Likewise.
23641 (iamcu_alignment): Likewise.
23642 (canonicalize_vector_int_perm): Likewise.
23643 (ix86_noce_conversion_profitable_p): Likewise.
23644 (ix86_mpx_bound_mode): Likewise.
23645 (ix86_operands_ok_for_move_multiple): Likewise.
23646 * config/microblaze/microblaze-protos.h
23647 (microblaze_expand_conditional_branch_reg): Likewise.
23648 * config/microblaze/microblaze.c
23649 (microblaze_expand_conditional_branch_reg): Likewise.
23650 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
23652 (rs6000_reassociation_width): Likewise.
23653 (rs6000_invalid_binary_op): Likewise.
23654 (fusion_p9_p): Likewise.
23655 (emit_fusion_p9_load): Likewise.
23656 (emit_fusion_p9_store): Likewise.
23657 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
23659 (riscv_hard_regno_mode_ok_p): Likewise.
23660 (riscv_address_insns): Likewise.
23661 (riscv_split_symbol): Likewise.
23662 (riscv_legitimize_move): Likewise.
23663 (riscv_function_value): Likewise.
23664 (riscv_hard_regno_nregs): Likewise.
23665 (riscv_expand_builtin): Likewise.
23666 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
23667 (riscv_build_integer): Likewise.
23668 (riscv_split_integer): Likewise.
23669 (riscv_legitimate_constant_p): Likewise.
23670 (riscv_cannot_force_const_mem): Likewise.
23671 (riscv_regno_mode_ok_for_base_p): Likewise.
23672 (riscv_valid_base_register_p): Likewise.
23673 (riscv_valid_offset_p): Likewise.
23674 (riscv_valid_lo_sum_p): Likewise.
23675 (riscv_classify_address): Likewise.
23676 (riscv_legitimate_address_p): Likewise.
23677 (riscv_address_insns): Likewise.
23678 (riscv_load_store_insns): Likewise.
23679 (riscv_force_binary): Likewise.
23680 (riscv_split_symbol): Likewise.
23681 (riscv_force_address): Likewise.
23682 (riscv_legitimize_address): Likewise.
23683 (riscv_move_integer): Likewise.
23684 (riscv_legitimize_const_move): Likewise.
23685 (riscv_legitimize_move): Likewise.
23686 (riscv_address_cost): Likewise.
23687 (riscv_subword): Likewise.
23688 (riscv_output_move): Likewise.
23689 (riscv_canonicalize_int_order_test): Likewise.
23690 (riscv_emit_int_order_test): Likewise.
23691 (riscv_function_arg_boundary): Likewise.
23692 (riscv_pass_mode_in_fpr_p): Likewise.
23693 (riscv_pass_fpr_single): Likewise.
23694 (riscv_pass_fpr_pair): Likewise.
23695 (riscv_get_arg_info): Likewise.
23696 (riscv_function_arg): Likewise.
23697 (riscv_function_arg_advance): Likewise.
23698 (riscv_arg_partial_bytes): Likewise.
23699 (riscv_function_value): Likewise.
23700 (riscv_pass_by_reference): Likewise.
23701 (riscv_setup_incoming_varargs): Likewise.
23702 (riscv_print_operand): Likewise.
23703 (riscv_elf_select_rtx_section): Likewise.
23704 (riscv_save_restore_reg): Likewise.
23705 (riscv_for_each_saved_reg): Likewise.
23706 (riscv_register_move_cost): Likewise.
23707 (riscv_hard_regno_mode_ok_p): Likewise.
23708 (riscv_hard_regno_nregs): Likewise.
23709 (riscv_class_max_nregs): Likewise.
23710 (riscv_memory_move_cost): Likewise.
23711 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
23712 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
23713 (rl78_addr_space_address_mode): Likewise.
23714 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
23716 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
23717 (rs6000_reassociation_width): Likewise.
23718 (rs6000_invalid_binary_op): Likewise.
23719 (fusion_p9_p): Likewise.
23720 (emit_fusion_p9_load): Likewise.
23721 (emit_fusion_p9_store): Likewise.
23722 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
23723 (ok_for_simple_move_operands): Likewise.
23724 (ok_for_simple_move_strict_operands): Likewise.
23725 (ok_for_simple_arith_logic_operands): Likewise.
23726 (visium_legitimize_reload_address): Likewise.
23727 (visium_select_cc_mode): Likewise.
23728 (output_cbranch): Likewise.
23729 (visium_split_double_move): Likewise.
23730 (visium_expand_copysign): Likewise.
23731 (visium_expand_int_cstore): Likewise.
23732 (visium_expand_fp_cstore): Likewise.
23733 * config/visium/visium.c (visium_pass_by_reference): Likewise.
23734 (visium_function_arg): Likewise.
23735 (visium_function_arg_advance): Likewise.
23736 (visium_libcall_value): Likewise.
23737 (visium_setup_incoming_varargs): Likewise.
23738 (visium_legitimate_constant_p): Likewise.
23739 (visium_legitimate_address_p): Likewise.
23740 (visium_legitimize_address): Likewise.
23741 (visium_secondary_reload): Likewise.
23742 (visium_register_move_cost): Likewise.
23743 (visium_memory_move_cost): Likewise.
23744 (prepare_move_operands): Likewise.
23745 (ok_for_simple_move_operands): Likewise.
23746 (ok_for_simple_move_strict_operands): Likewise.
23747 (ok_for_simple_arith_logic_operands): Likewise.
23748 (visium_function_value_1): Likewise.
23749 (rtx_ok_for_offset_p): Likewise.
23750 (visium_legitimize_reload_address): Likewise.
23751 (visium_split_double_move): Likewise.
23752 (visium_expand_copysign): Likewise.
23753 (visium_expand_int_cstore): Likewise.
23754 (visium_expand_fp_cstore): Likewise.
23755 (visium_split_cstore): Likewise.
23756 (visium_select_cc_mode): Likewise.
23757 (visium_split_cbranch): Likewise.
23758 (output_cbranch): Likewise.
23759 (visium_print_operand_address): Likewise.
23760 * expmed.c (flip_storage_order): Likewise.
23761 * expmed.h (emit_cstore): Likewise.
23762 (flip_storage_order): Likewise.
23763 * genrecog.c (validate_pattern): Likewise.
23764 * hsa-gen.c (gen_hsa_addr): Likewise.
23765 * internal-fn.c (expand_arith_overflow): Likewise.
23766 * ira-color.c (allocno_copy_cost_saving): Likewise.
23767 * lra-assigns.c (find_hard_regno_for_1): Likewise.
23768 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
23769 (process_invariant_for_inheritance): Likewise.
23770 * lra-eliminations.c (move_plus_up): Likewise.
23771 * omp-low.c (lower_oacc_reductions): Likewise.
23772 * simplify-rtx.c (simplify_subreg): Likewise.
23773 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
23774 (TARGET_CHKP_BOUND_MODE): Likewise..
23775 * targhooks.c (default_chkp_bound_mode): Likewise.
23776 (default_setup_incoming_vararg_bounds): Likewise.
23777 * targhooks.h (default_chkp_bound_mode): Likewise.
23778 (default_setup_incoming_vararg_bounds): Likewise.
23779 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
23780 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
23781 (have_whole_vector_shift): Likewise.
23782 * tree-vect-stmts.c (vectorizable_load): Likewise.
23783 * doc/tm.texi: Regenerate.
23785 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23787 Graceful degrade if Binutils PR21472 is not available.
23790 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
23791 .rodata in flash test fails.
23792 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
23793 * confgure: Regenerate.
23794 * config.in: Regenerate.
23795 * config/avr/avr.c (avr_asm_named_section)
23796 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
23797 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
23798 (avr_asm_init_sections): Same.
23800 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23802 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
23803 (fma<VH:mode>4_intrinsic): Likewise.
23804 (*fmsub<VCVTF:mode>4): Likewise.
23805 (*fmsub<VH:mode>4_intrinsic): Likewise.
23807 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23810 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
23811 Don't depend on "optimize > 0".
23812 (out_movhi_r_mr, out_movqi_mr_r): Same.
23813 (out_movhi_mr_r, out_movqi_r_mr): Same.
23814 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
23815 io_address_operand on "optimize > 0".
23817 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23819 * tree-loop-distribution.c: Add general explanantion on the pass.
23820 (generate_loops_for_partition): Mark distributed loop.
23821 (pg_add_dependence_edges): New parameter. Handle alias data
23822 dependence specially and record it in the parameter if asked.
23823 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
23824 (init_partition_graph_vertices, add_partition_graph_edge): New.
23825 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
23826 (free_partition_graph_vdata, build_partition_graph): New.
23827 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
23828 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
23829 (data_ref_segment_size, latch_dominated_by_data_ref): New.
23830 (compute_alias_check_pairs, version_loop_by_alias_check): New.
23831 (version_for_distribution_p, finalize_partitions): New.
23832 (distribute_loop): Handle alias data dependence specially. Factor
23833 out loop fusion code as functions and call these functions.
23835 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23837 * tree-loop-distribution.c (classify_partition): New parameter and
23838 better handle reduction statement.
23839 (rdg_build_partitions): Revise comment.
23840 (distribute_loop): Compute statements in all partitions and pass it
23841 to classify_partition.
23843 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23845 * tree-loop-distribution.c (enum partition_type): New.
23846 (struct partition): New field type.
23847 (partition_merge_into): Add parameter. Update partition type.
23848 (data_dep_in_cycle_p, update_type_for_merge): New functions.
23849 (build_rdg_partition_for_vertex): Compute partition type.
23850 (rdg_build_partitions): Dump partition type.
23851 (distribute_loop): Update calls to partition_merge_into.
23853 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23855 * tree-loop-distribution.c (struct ddr_hasher): New.
23856 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
23858 (classify_partition): Call get_data_dependence.
23859 (pg_add_dependence_edges): Ditto.
23860 (distribute_loop): Release data dependence hash table.
23862 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23864 * tree-loop-distribution.c (ref_base_address): Delete.
23865 (similar_memory_accesses): Rename ...
23866 (share_memory_accesses): ... to this. Check if partitions access
23867 the same memory reference.
23868 (distribute_loop): Call share_memory_accesses.
23870 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23872 * tree-loop-distribution.c (struct partition): New field recording
23873 its data reference.
23874 (partition_alloc, partition_free): Init and release data refs.
23875 (partition_merge_into): Merge data refs.
23876 (build_rdg_partition_for_vertex): Collect data refs for partition.
23877 (pg_add_dependence_edges): Change parameters from vector to bitmap.
23879 (distribute_loop): Remve data refs from vertice data of partition
23882 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23884 * tree-loop-distribution.c (params.h): Include header file.
23885 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
23886 (datarefs_vec): New global var.
23887 (create_rdg_vertices): Use datarefs_vec directly.
23888 (free_rdg): Don't free data references.
23889 (build_rdg): Update use. Don't free data references.
23890 (distribute_loop): Compute global variable for data references.
23891 Bail out if there are too many data references.
23893 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23895 * tree-loop-distribution.c (loop_nest): New global var.
23896 (build_rdg): Use loop directly, rather than loop nest.
23897 (pg_add_dependence_edges): Remove loop nest parameter. Use global
23899 (distribute_loop): Compute global variable loop nest. Update use.
23901 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23903 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
23904 (partition_merge_into): New parameter. Dump reason for fusion.
23905 (distribute_loop): Update use of partition_merge_into.
23907 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23909 * tree-loop-distribution.c (bb_top_order_index): New.
23910 (bb_top_order_index_size, bb_top_order_cmp): New.
23911 (stmts_from_loop): Use topological order.
23912 (pass_loop_distribution::execute): Compute and release topological
23913 order for basic blocks.
23915 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23917 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
23920 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23922 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
23923 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
23924 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
23925 * internal-fn.def (LOOP_DIST_ALIAS): New.
23926 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
23927 (fold_loop_internal_call): ... this.
23928 (vect_loop_dist_alias_call): New function.
23929 (set_uid_loop_bbs): Call fold_loop_internal_call.
23930 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
23933 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
23936 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
23937 Require dead FLAGS_REG at the beginning of a peephole.
23939 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
23942 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
23943 arguments in the call to __builtin_ia32_sbb_u32.
23944 (_subborrow_u64): Swap _X and _Y arguments in the call to
23945 __builtin_ia32_sbb_u64.
23947 2017-07-04 Jakub Jelinek <jakub@redhat.com>
23950 * tree-vrp.c (compare_assert_loc): Turn into a function template
23951 with stable template parameter. Only test if a->e is NULL,
23952 !a->e == !b->e has been verified already. Use e == NULL or
23953 e != NULL instead of e or ! e tests. If stable is true, don't use
23954 iterative_hash_expr, on the other side allow a or b or both NULL
23955 and sort the NULLs last.
23956 (process_assert_insertions): Sort using compare_assert_loc<false>
23957 instead of compare_assert_loc, later sort using
23958 compare_assert_loc<true> before calling process_assert_insertions_for
23959 in a loop. Use break instead of continue once seen NULL pointer.
23961 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23963 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23964 Cortex-R7 and Cortex-R8 processors.
23966 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23968 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
23969 uninitialized while src is not.
23971 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
23973 * common/config/arm/arm-common.c: Adjust include path for
23975 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
23976 (arm-cpu.h): Create in build directory. Adjust dependency rules.
23977 (arm-cpu-data.h): Likewise.
23978 (arm-cpu-cdata.h): Likewise.
23979 * config/arm/arm-cpu.h: Delete.
23980 * config/arm/arm-cpu-cdata.h: Delete.
23981 * config/arm/arm-cpu-data.h: Delete.
23983 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
23985 * config/arm/arm-cpus.in (cortex-a55): New.
23986 (cortex-a75): Likewise.
23987 (cortex-a75.cortex-a55): Likewise.
23988 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
23990 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
23991 * config/arm/arm-cpu-cdata.h: Regenerate.
23992 * config/arm/arm-cpu-data.h: Regenerate.
23993 * config/arm/arm-cpu.h: Regenerate.
23994 * config/arm/arm-tables.opt: Regenerate.
23995 * config/arm/arm-tune.md: Regenerate.
23997 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23999 * haifa-sched.c (sched_create_recovery_edges): Update profile.
24001 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
24003 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
24006 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
24008 PR tree-optimization/81292
24009 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
24010 full_string_p, also call adjust_related_strinfos if the adjustment
24011 is simple, otherwise invalidate related strinfos.
24013 2017-07-04 Martin Liska <mliska@suse.cz>
24016 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
24017 newly created variable as DECL_IGNORED_P.
24019 2017-07-04 Martin Liska <mliska@suse.cz>
24022 * ipa-inline.c (inline_small_functions):
24023 Use xstrdup_for_dump.
24025 2017-07-04 Tom de Vries <tom@codesourcery.com>
24027 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
24029 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24032 * config/darwin.c (darwin_function_switched_text_sections):
24035 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
24037 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
24039 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24041 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
24043 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24045 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
24046 min_profitable_iters, and th as inclusive lower bounds.
24047 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
24048 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
24049 for min_profitable_iters and min_profitable_estimate.
24050 (vect_transform_loop): Treat th as an inclusive lower bound.
24051 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
24053 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
24056 * config/darwin.c (darwin_function_switched_text_sections):
24057 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
24058 in two pieces, and suppress the use of buf.
24060 2017-07-03 Nathan Sidwell <nathan@acm.org>
24062 * hash-table.h (hash_table_mod1): Fix indentation.
24064 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24066 PR middle-end/81290
24067 * predict.c (force_edge_cold): Be more careful about propagation
24069 * profile-count.h (profile_probability::guessed,
24070 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
24072 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
24074 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
24076 * doc/invoke.texi (rcpc architecture extension): Document it.
24078 2017-07-03 Richard Biener <rguenther@suse.de>
24080 PR tree-optimization/60510
24081 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
24082 the scalar reduction PHI and use it.
24083 (vectorizable_reduction): Properly guard the single_defuse_cycle
24084 path for non-SLP reduction chains where we cannot use it.
24085 Rework reduc_def/index and vector type deduction. Rework
24086 vector operand gathering during reduction op code-gen.
24087 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
24088 chains dissolve the chain and leave it to non-SLP reduction
24091 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24093 * tree-data-ref.h (dr_alignment): Declare.
24094 * tree-data-ref.c (dr_alignment): New function.
24095 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
24096 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
24098 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
24100 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24102 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
24103 and base_misalignment fields.
24104 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
24105 * tree-data-ref.c: Include builtins.h.
24106 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
24107 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
24108 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
24109 * tree-vect-data-refs.c: Include tree-cfg.h.
24110 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
24111 fields instead of calculating an alignment here.
24112 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
24113 innermost_loop_behavior fields.
24115 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24117 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
24119 (DR_STEP_ALIGNMENT): New macro.
24120 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
24121 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
24122 (create_data_ref): Print it.
24123 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
24124 to tell whether the step preserves vector (mis)alignment.
24125 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24126 Move the check for an integer step and generalise to all INTEGER_CST.
24127 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
24128 Print the outer step alignment.
24130 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24132 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
24133 with offset_alignment.
24134 (DR_ALIGNED_TO): Delete.
24135 (DR_OFFSET_ALIGNMENT): New macro.
24136 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
24137 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
24138 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
24139 (create_data_ref): Likewise.
24140 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
24141 (vect_analyze_data_refs): Likewise.
24142 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
24143 creating dummy innermost behavior.
24145 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24147 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
24148 with a "innermost_loop_behavior *" and refeence tree.
24149 * tree-data-ref.c (dr_analyze_innermost): Likewise.
24150 (create_data_ref): Update call accordingly.
24151 * tree-predcom.c (find_looparound_phi): Likewise.
24153 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24155 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
24156 fields with dr_wrt_vec_loop.
24157 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
24158 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
24159 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
24160 (vect_dr_behavior): New function.
24161 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24162 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24163 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
24164 track whether the step preserves the misalignment.
24165 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
24166 Use vect_dr_behavior.
24167 (vect_setup_realignment): Update call accordingly.
24168 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
24169 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
24170 call to vect_create_addr_base_for_vector_ref.
24171 (vect_create_cond_for_align_checks): Likewise.
24172 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
24173 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
24174 (vect_recog_mask_conversion_pattern): Likewise.
24175 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
24176 (new_stmt_vec_info): Remove redundant zeroing.
24178 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
24180 * common/config/arm/arm-common.c (arm_be8_option): New function.
24181 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
24182 (ISA_ARMv6): Add isa_bit_be8.
24183 * config/arm/arm.h (arm_be8_option): Add prototype.
24184 (BE8_SPEC_FUNCTION): New define.
24185 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
24186 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
24187 (mlittle-endian): Similarly.
24188 (mbe8, mbe32): New options.
24189 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
24190 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
24192 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24194 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
24196 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24198 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
24199 (cleanup_tree_cfg_bb): Use it.
24200 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
24202 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
24204 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24207 * loop-doloop.c (add_test): Update profile.
24209 2017-07-03 Martin Liska <mliska@suse.cz>
24212 * sanopt.c (rewrite_usage_of_param): New function.
24213 (sanitize_rewrite_addressable_params): Likewise.
24214 (pass_sanopt::execute): Call rewrite_usage_of_param.
24216 2017-07-03 Richard Biener <rguenther@suse.de>
24218 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
24219 back to using VIEW_CONVERT_EXPR.
24221 2017-07-03 Martin Liska <mliska@suse.cz>
24224 * doc/extend.texi: Document when a resolver function is
24225 generated for target_clones.
24227 2017-07-03 Martin Liska <mliska@suse.cz>
24229 * asan.c (asan_emit_stack_protection): Unpoison just red zones
24230 and shadow memory of auto variables which are subject of
24231 use-after-scope sanitization.
24232 (asan_expand_mark_ifn): Add do set only when is_poison.
24234 2016-07-03 Richard Biener <rguenther@suse.de>
24236 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
24238 (vect_force_simple_reduction): Record reduction def -> phi mapping.
24239 (vectorizable_reduction): Perform reduction PHI creation when
24240 visiting a reduction PHI and adjust and simplify code generation
24241 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
24242 (vect_transform_loop): Visit reduction PHIs.
24243 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
24244 defs into the SLP tree.
24245 (vect_build_slp_tree): Reduction defs terminate the recursion.
24246 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
24248 (vect_get_vec_defs_for_stmt_copy): Export.
24249 (vect_get_vec_defs): Likewise.
24250 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
24252 (vect_get_vec_defs_for_stmt_copy): Declare.
24253 (vect_get_vec_defs): Likewise.
24255 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
24257 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
24258 parameter with a "loop" parameter and use it instead of the
24259 loop containing DR_STMT. Don't check simple_iv when doing
24260 BB analysis. Describe the two analysis modes in the comment.
24262 2017-07-03 Tom de Vries <tom@codesourcery.com>
24264 PR tree-optimization/69468
24265 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
24266 (find_same_succ_bb): Handle ignore_edge_flags.
24268 2017-07-03 Tom de Vries <tom@codesourcery.com>
24270 PR tree-optimization/81192
24271 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
24273 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
24275 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
24277 2017-07-03 Tom de Vries <tom@codesourcery.com>
24279 PR tree-optimization/81192
24280 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
24281 BB_SAME_SUCC (bb) == NULL.
24283 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24285 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
24288 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24290 * dumpfile.c: Include profile-count.h
24291 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
24293 (insert_cond_bb): Update profile.
24294 * tree-cfg.h (insert_cond_bb): Update prototype.
24295 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
24296 * tree-dump.c: Do not include tree-cfg.
24298 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24300 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
24302 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24304 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
24307 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
24309 * tree-complex.c (expand_complex_div_wide): update profile.
24311 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24312 Alan Hayward <alan.hayward@arm.com>
24313 David Sherwood <david.sherwood@arm.com>
24315 * Makefile.in (MACHMODE_H): Remove insn-modes.h
24316 (CORETYPES_H): New define.
24317 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
24318 (insn-modes-inline.h, s-modes-inline-h): New rules.
24319 (generated_files): Add insn-modes-inline.h.
24320 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
24321 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
24322 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
24323 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
24324 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
24325 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
24326 (build/gencodes.o, build/genconditions.o): Likewise.
24327 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
24328 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
24329 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
24330 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
24331 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
24332 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
24333 * coretypes.h: Include everything up to real.h for generators.
24334 Include insn-modes.h first. Include wide-int-print.h after
24335 wide-int.h. Include insn-modes-inline.h and then machmode.h.
24336 * machmode.h: Don't include insn-modes.h here.
24337 * function-tests.c: Remove includes of signop.h, machmode.h,
24338 double-int.h and wide-int.h.
24340 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
24342 * optc-save-gen.awk: Likewise.
24343 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
24344 * godump.c: Remove include of wide-int-print.h.
24345 * pretty-print.h: Likewise.
24346 * wide-int-print.cc: Likewise.
24347 * wide-int.cc: Likewise.
24348 * hash-map-tests.c: Remove include of signop.h.
24349 * hash-set-tests.c: Likewise.
24350 * rtl-tests.c: Likewise.
24351 * mkconfig.sh: Remove include of machmode.h.
24352 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
24354 (emit_insn_modes_inline_h): ...this new function. Emit the code
24355 into an insn-modes-inline.h header file, adding appropriate
24356 include guards and end comments.
24357 (emit_insn_modes_c_header): Remove include of machmode.h.
24358 (emit_min_insn_modes_c_header): Include coretypes.h rather than
24360 (main): Handle -i flag and call emit_insn_modes_inline_h when
24363 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24365 * tree-ssa-strlen.c (strinfo): Rename the length field to
24366 nonzero_chars. Add a full_string_p field.
24367 (compare_nonzero_chars, zero_length_string_p): New functions.
24368 (get_addr_stridx): Add an offset_out parameter.
24369 Use compare_nonzero_chars.
24370 (get_stridx): Update accordingly. Use compare_nonzero_chars.
24371 (new_strinfo): Update after above changes to strinfo.
24372 (set_endptr_and_length): Set full_string_p.
24373 (get_string_length): Update after above changes to strinfo.
24374 (unshare_strinfo): Update call to new_strinfo.
24375 (maybe_invalidate): Likewise.
24376 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
24377 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
24378 as a uhwi instead of an shwi. Update after above changes to
24379 strinfo and new_strinfo.
24380 (zero_length_string): Assert that chainsi contains full strings.
24381 Use zero_length_string_p. Update call to new_strinfo.
24382 (adjust_related_strinfos): Update after above changes to strinfo.
24383 Copy full_string_p from origsi.
24384 (adjust_last_stmt): Use zero_length_string_p.
24385 (handle_builtin_strlen): Update after above changes to strinfo and
24386 new_strinfo. Install the lhs as the string length if the previous
24387 entry didn't describe a full string.
24388 (handle_builtin_strchr): Update after above changes to strinfo
24390 (handle_builtin_strcpy): Likewise.
24391 (handle_builtin_strcat): Likewise.
24392 (handle_builtin_malloc): Likewise.
24393 (handle_pointer_plus): Likewise.
24394 (handle_builtin_memcpy): Likewise. Track nonzero characters
24395 that aren't necessarily followed by a nul terminator.
24396 (handle_char_store): Likewise.
24398 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24400 PR tree-optimization/80769
24401 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
24402 for malloc and calloc. Document the new invariant that all related
24403 strinfos have delayed lengths or none do.
24404 (verify_related_strinfos): Move earlier in file.
24405 (set_endptr_and_length): New function, split out from...
24406 (get_string_length): ...here. Also set the lengths of related
24408 (zero_length_string): Assert that chainsi has known (rather than
24410 (adjust_related_strinfos): Likewise.
24412 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
24414 PR tree-optimization/81136
24415 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
24416 assert that two references with the same misalignment have the same
24417 compile-time misalignment if those compile-time misalignments
24420 2017-07-01 Andi Kleen <ak@linux.intel.com>
24422 * print-tree.c (print_node): Print all attributes.
24424 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24426 * cfg.c (scale_bbs_frequencies): New function.
24427 * cfg.h (scale_bbs_frequencies): Declare it.
24428 * cfgloopanal.c (single_likely_exit): Cleanup.
24429 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
24431 (scale_loop_profile): Likewise.
24432 (loop_version): Likewise.
24433 (create_empty_loop_on_edge): Update.
24434 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
24435 scale_loop_frequencies, scale_loop_profile, loopify,
24436 loop_version): Update prototypes.
24437 * modulo-sched.c (sms_schedule): Update.
24438 * predict.c (unlikely_executed_edge_p): Also check probability.
24439 (probably_never_executed_edge_p): Fix typo.
24440 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24441 * tree-parloops.c (gen_parallel_loop): Update.
24442 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
24443 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24444 * tree-ssa-loop-split.c (split_loop): Update.
24445 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24446 * tree-vect-loop-manip.c (vect_do_peeling): Update.
24447 (vect_loop_versioning): Update.
24448 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24450 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24452 * trans-mem.c (split_bb_make_tm_edge): Update profile.
24454 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24456 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
24457 to keep profile consistent.
24459 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24461 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
24462 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
24463 * profile-count.h (max_safe_multiplier): Make unsigned.
24464 (profile_count::guessed_zero): New.
24466 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24468 * bb-reorder.c (fix_up_crossing_landing_pad,
24469 fix_crossing_conditional_branches): Use make_single_succ_edge
24470 to keep profile consistent.
24472 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
24474 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
24477 2017-07-01 Jakub Jelinek <jakub@redhat.com>
24480 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
24481 the right scopes, make sure cond_jump isn't preserved between multiple
24482 iterations. Search for fallthru edge whenever there are 3+ edges and
24483 use find_fallthru_edge for it.
24485 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24487 Patch by Alexander Monakov <amonakov@ispras.ru>
24488 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
24489 probabilities consistently.
24491 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24493 * pa.c (pa_expand_compare_and_swap_loop): Update call of
24494 emit_cmp_and_jump_insns.
24496 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24499 * tree-inline.c (expand_call_inline): Combine profile statuses.
24501 2017-06-30 Andrew Pinski <apinski@cavium.com>
24503 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
24504 fold_stmt returned true.
24506 2017-06-30 Nathan Sidwell <nathan@acm.org>
24508 * ggc.h (empty_string): Delete.
24509 * cfgexpand.c (expand_asm_stmt): Use plain "".
24510 * optabs.c (expand_asm_memory_barrier): Likewise.
24511 * stringpool.c (empty_string): Delete.
24512 (digit_vector, digit_string): Delete.
24513 (ggc_alloc_string): Use plain "", don't optimize single digit
24514 strings. Use ggc_alloc_atomic.
24516 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
24518 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
24519 comparison set and one other set, use the cost of the non-comparison
24522 2017-06-30 Nathan Sidwell <nathan@acm.org>
24524 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
24527 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
24529 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
24530 loops. Remove now unneeded calls to gimple_switch_set_label() that
24531 just set removed labels to NULL_TREE.
24533 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
24535 * tree-ssanames.c (set_range_info_raw): Abstract from ...
24536 (set_range_info): ...here. Only call set_range_info_raw if domain
24538 (set_nonzero_bits): Call set_range_info_raw.
24539 * tree-ssanames.h (set_range_info_raw): New.
24541 2017-06-30 Jakub Jelinek <jakub@redhat.com>
24544 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
24545 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
24546 of nonimmediate_operand and <store_mask_constraint> instead of m
24547 for the input operand. For V8FI iterator, always split if input
24548 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
24549 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
24550 <store_mask_predicate> instead of register_operand and
24551 <store_mask_constraint> instead of v for the input operand. Make
24552 sure both operands aren't MEMs for if not <mask_applied>.
24554 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
24556 * lto-wrapper.c (copy_file) Close both file descriptors before
24559 2017-06-30 Martin Liska <mliska@suse.cz>
24562 * multiple_target.c (create_dispatcher_calls): Make ifunc
24563 also for function that don't have calls or are not referenced.
24565 2017-06-30 Richard Biener <rguenther@suse.de>
24567 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
24568 analyze the first scalar stmt. Move vector type computation
24569 for the BB case here from ...
24570 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
24571 live operation processing in the SLP case properly.
24573 2017-06-30 Richard Biener <rguenther@suse.de>
24575 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
24577 2017-06-30 Martin Liska <mliska@suse.cz>
24580 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
24581 before BUILT_IN_UNWIND_RESUME when ASAN is used.
24583 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
24585 * doc/invoke.texi (AArch64): Add missing options and remove redundant
24588 2017-06-30 Richard Biener <rguenther@suse.de>
24590 PR tree-optimization/81249
24591 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
24592 condition reduction result to original scalar type.
24594 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24596 * profile-count.h (enum profile_quality): Fix typos and whitespace
24599 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24601 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
24602 type for branch probabilities.
24604 2017-06-29 Julian Brown <julian@codesourcery.com>
24605 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24607 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
24608 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
24609 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
24610 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
24612 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24614 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
24615 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
24616 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
24617 CC usage from generic code to here.
24618 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
24619 CC usage into the target macros.
24621 2017-06-29 Maya Rashish <coypu@sdf.org>
24623 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
24626 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24628 * arm/arm-builtins.c: Include profile-count.h
24629 * except.c (sjlj_emit_function_enter): Use
24630 profile_probability::unlikely.
24632 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24634 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
24635 and tocrel_offset be pointer args rather than implicitly using
24637 (legitimate_constant_pool_address_p, rs6000_emit_move,
24638 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
24639 tocrel_offset and use in toc_relative_expr_p call.
24640 (print_operand, print_operand_address): Use static tocrel_base_oac
24641 and tocrel_offset_oac.
24642 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
24645 2017-06-29 Maya Rashish <coypu@sdf.org>
24647 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
24649 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
24651 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
24652 objects, take into account only the alignment of 'op0' and 'mode1' if
24655 2017-06-29 Steve Ellcey <sellcey@cavium.com>
24657 * ccmp.c (ccmp_tree_comparison_p): New function.
24658 (ccmp_candidate_p): Update to use above function.
24659 (get_compare_parts): New function.
24660 (expand_ccmp_next): Update to use new functions.
24661 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
24663 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
24664 take mode as argument.
24665 * ccmp.h (expand_ccmp_expr): Add mode as argument.
24666 * expr.c (expand_expr_real_1): Pass mode as argument.
24668 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
24670 * combine.c (combine_instructions): Print insns to dump_file, together
24673 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
24675 * asan.c (asan_emit_stack_protection): Update.
24676 (create_cond_insert_point): Update.
24677 * auto-profile.c (afdo_propagate_circuit): Update.
24678 * basic-block.h (struct edge_def): Turn probability to
24679 profile_probability.
24680 (EDGE_FREQUENCY): Update.
24681 * bb-reorder.c (find_traces_1_round): Update.
24682 (better_edge_p): Update.
24683 (sanitize_hot_paths): Update.
24684 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
24685 (make_single_succ_edge): Update.
24686 (check_bb_profile): Update.
24687 (dump_edge_info): Update.
24688 (update_bb_profile_for_threading): Update.
24689 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
24690 probabilitycount to 0.
24691 * cfgbuild.c (compute_outgoing_frequencies): Update.
24692 * cfgcleanup.c (try_forward_edges): Update.
24693 (outgoing_edges_match): Update.
24694 (try_crossjump_to_edge): Update.
24695 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
24696 (expand_gimple_tailcall): Update.
24697 (construct_init_block): Use make_single_succ_edge.
24698 (construct_exit_block): Use make_single_succ_edge.
24699 * cfghooks.c (verify_flow_info): Update.
24700 (redirect_edge_succ_nodup): Update.
24701 (split_edge): Update.
24702 (account_profile_record): Update.
24703 * cfgloopanal.c (single_likely_exit): Update.
24704 * cfgloopmanip.c (scale_loop_profile): Update.
24705 (set_zero_probability): Remove.
24706 (duplicate_loop_to_header_edge): Update.
24707 * cfgloopmanip.h (loop_version): Update prototype.
24708 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
24709 (force_nonfallthru_and_redirect): Update.
24710 (update_br_prob_note): Update.
24711 (rtl_verify_edges): Update.
24712 (purge_dead_edges): Update.
24713 (rtl_lv_add_condition_to_bb): Update.
24714 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
24715 * cgraphunit.c (init_lowered_empty_function): Update.
24716 (cgraph_node::expand_thunk): Update.
24717 * cilk-common.c: Include profile-count.h
24718 * dojump.c (inv): Remove.
24719 (jumpifnot): Update.
24720 (jumpifnot_1): Update.
24721 (do_jump_1): Update.
24723 (do_jump_by_parts_greater_rtx): Update.
24724 (do_compare_rtx_and_jump): Update.
24725 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
24726 do_jump_1. do_compare_rtx_and_jump): Update prototype.
24727 * dwarf2cfi.c: Include profile-count.h
24728 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
24729 (sjlj_emit_dispatch_table): Likewise.
24730 * explow.c: Include profile-count.h
24731 * expmed.c (emit_store_flag_force): Update.
24732 (do_cmp_and_jump): Update.
24733 * expr.c (compare_by_pieces_d::generate): Update.
24734 (compare_by_pieces_d::finish_mode): Update.
24735 (emit_block_move_via_loop): Update.
24736 (store_expr_with_bounds): Update.
24737 (store_constructor): Update.
24738 (expand_expr_real_2): Update.
24739 (expand_expr_real_1): Update.
24740 * expr.h (try_casesi, try_tablejump): Update prototypes.
24741 * gimple-pretty-print.c (dump_probability): Update.
24742 (dump_profile): New.
24743 (dump_gimple_label): Update.
24744 (dump_gimple_bb_header): Update.
24745 * graph.c (draw_cfg_node_succ_edges): Update.
24746 * hsa-gen.c (convert_switch_statements): Update.
24747 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
24748 (find_if_case_1): Update.
24749 (find_if_case_2): Update.
24750 * internal-fn.c (expand_arith_overflow_result_store): Update.
24751 (expand_addsub_overflow): Update.
24752 (expand_neg_overflow): Update.
24753 (expand_mul_overflow): Update.
24754 (expand_vector_ubsan_overflow): Update.
24755 * ipa-cp.c (good_cloning_opportunity_p): Update.
24756 * ipa-split.c (split_function): Use make_single_succ_edge.
24757 * ipa-utils.c (ipa_merge_profiles): Update.
24758 * loop-doloop.c (add_test): Update.
24759 (doloop_modify): Update.
24760 * loop-unroll.c (compare_and_jump_seq): Update.
24761 (unroll_loop_runtime_iterations): Update.
24762 * lra-constraints.c (lra_inheritance): Update.
24763 * lto-streamer-in.c (input_cfg): Update.
24764 * lto-streamer-out.c (output_cfg): Update.
24765 * mcf.c (adjust_cfg_counts): Update.
24766 * modulo-sched.c (sms_schedule): Update.
24767 * omp-expand.c (expand_omp_for_init_counts): Update.
24768 (extract_omp_for_update_vars): Update.
24769 (expand_omp_ordered_sink): Update.
24770 (expand_omp_for_ordered_loops): Update.
24771 (expand_omp_for_generic): Update.
24772 (expand_omp_for_static_nochunk): Update.
24773 (expand_omp_for_static_chunk): Update.
24774 (expand_cilk_for): Update.
24775 (expand_omp_simd): Update.
24776 (expand_omp_taskloop_for_outer): Update.
24777 (expand_omp_taskloop_for_inner): Update.
24778 * omp-simd-clone.c (simd_clone_adjust): Update.
24779 * optabs.c (expand_doubleword_shift): Update.
24780 (expand_abs): Update.
24781 (emit_cmp_and_jump_insn_1): Update.
24782 (expand_compare_and_swap_loop): Update.
24783 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
24784 * predict.c (predictable_edge_p): Update.
24785 (edge_probability_reliable_p): Update.
24786 (set_even_probabilities): Update.
24787 (combine_predictions_for_insn): Update.
24788 (combine_predictions_for_bb): Update.
24789 (propagate_freq): Update.
24790 (estimate_bb_frequencies): Update.
24791 (force_edge_cold): Update.
24792 * profile-count.c (profile_count::dump): Add missing space into dump.
24793 (profile_count::debug): Add newline.
24794 (profile_count::differs_from_p): Explicitly convert to unsigned.
24795 (profile_count::stream_in): Update.
24796 (profile_probability::dump): New member function.
24797 (profile_probability::debug): New member function.
24798 (profile_probability::differs_from_p): New member function.
24799 (profile_probability::differs_lot_from_p): New member function.
24800 (profile_probability::stream_in): New member function.
24801 (profile_probability::stream_out): New member function.
24802 * profile-count.h (profile_count_quality): Rename to ...
24803 (profile_quality): ... this one.
24804 (profile_probability): New.
24805 (profile_count): Update.
24806 * profile.c (compute_branch_probabilities): Update.
24807 * recog.c (peep2_attempt): Update.
24808 * sched-ebb.c (schedule_ebbs): Update.
24809 * sched-rgn.c (find_single_block_region): Update.
24810 (compute_dom_prob_ps): Update.
24811 (schedule_region): Update.
24812 * sel-sched-ir.c (compute_succs_info): Update.
24813 * stmt.c (struct case_node): Update.
24814 (do_jump_if_equal): Update.
24815 (get_outgoing_edge_probs): Update.
24816 (conditional_probability): Update.
24817 (emit_case_dispatch_table): Update.
24818 (expand_case): Update.
24819 (expand_sjlj_dispatch_table): Update.
24820 (emit_case_nodes): Update.
24821 * targhooks.c: Update.
24822 * tracer.c (better_p): Update.
24823 (find_best_successor): Update.
24824 * trans-mem.c (expand_transaction): Update.
24825 * tree-call-cdce.c: Update.
24826 * tree-cfg.c (gimple_split_edge): Upate.
24827 (move_sese_region_to_fn): Upate.
24828 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
24829 * tree-eh.c (lower_resx): Upate.
24830 (cleanup_empty_eh_move_lp): Upate.
24831 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24832 * tree-inline.c (copy_edges_for_bb): Update.
24833 (copy_cfg_body): Update.
24834 * tree-parloops.c (gen_parallel_loop): Update.
24835 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
24836 (gimple_gen_time_profiler): Update.
24837 * tree-ssa-dce.c (remove_dead_stmt): Update.
24838 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
24839 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
24840 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
24841 (unloop_loops): Update.
24842 (try_peel_loop): Update.
24843 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24844 * tree-ssa-loop-split.c (connect_loops): Update.
24845 (split_loop): Update.
24846 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24847 (hoist_guard): Update.
24848 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
24849 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
24850 (value_replacement): Update.
24851 * tree-ssa-reassoc.c (branch_fixup): Update.
24852 * tree-ssa-tail-merge.c (replace_block_by): Update.
24853 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
24854 (create_edge_and_update_destination_phis): Update.
24855 (compute_path_counts): Update.
24856 (recompute_probabilities): Update.
24857 (update_joiner_offpath_counts): Update.
24858 (freqs_to_counts_path): Update.
24859 (duplicate_thread_path): Update.
24860 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
24861 (struct switch_conv_info): Update.
24862 (gen_inbound_check): Update.
24863 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24864 (vect_do_peeling): Update.
24865 (vect_loop_versioning): Update.
24866 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24867 (optimize_mask_stores): Update.
24868 * ubsan.c (ubsan_expand_null_ifn): Update.
24869 * value-prof.c (gimple_divmod_fixed_value): Update.
24870 (gimple_divmod_fixed_value_transform): Update.
24871 (gimple_mod_pow2): Update.
24872 (gimple_mod_pow2_value_transform): Update.
24873 (gimple_mod_subtract): Update.
24874 (gimple_mod_subtract_transform): Update.
24875 (gimple_ic): Update.
24876 (gimple_stringop_fixed_value): Update.
24877 (gimple_stringops_transform): Update.
24878 * value-prof.h: Update.
24880 2017-06-29 Carl Love <cel@us.ibm.com>
24882 * config/rs6000/rs6000-c.c: Add support for built-in functions
24883 vector signed int vec_signed (vector float);
24884 vector signed long long vec_signed (vector double);
24885 vector signed int vec_signed2 (vector double, vector double);
24886 vector signed int vec_signede (vector double);
24887 vector signed int vec_signedo (vector double);
24888 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
24889 instruction generator.
24890 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24891 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
24892 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
24894 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
24895 vunsignede_v2df): Add define_expands.
24896 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
24897 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
24898 VEC_UNSIGNEDO): Add definitions.
24899 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24900 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
24901 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
24902 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
24903 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
24904 * config/rs6000/altivec.h (vec_signed, vec_signed2,
24905 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
24906 vec_unsignede, vec_unsignedo): Add builtin defines.
24907 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
24909 * doc/extend.texi: Update the built-in documentation file for the
24910 new built-in functions.
24912 2017-06-29 Richard Biener <rguenther@suse.de>
24914 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
24915 reduction chains to LOOP_VINFO_REDUCTIONS.
24916 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
24917 SLP reductions after processing reduction chains.
24919 2017-06-29 Nathan Sidwell <nathan@acm.org>
24921 * builtins.c (fold_builtin_FUNCTION): Use
24922 lang_hooks.decl_printable_name.
24924 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
24926 PR middle-end/81194
24927 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
24928 with only one label.
24929 * stmt.c (expand_case): Assert NCASES is greater than one.
24931 2017-06-29 Richard Biener <rguenther@suse.de>
24933 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
24935 (group_case_labels): Likewise.
24936 (find_taken_edge): Push sanity checking on val to workers...
24937 (find_taken_edge_cond_expr): ... here
24938 (find_taken_edge_switch_expr): ... and here, handle cases
24939 with just a default label.
24940 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
24941 (group_case_labels): Likewise.
24942 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
24943 group_case_labels does anything cleanup the CFG again.
24945 2017-06-29 Bin Cheng <bin.cheng@arm.com>
24947 PR tree-optimization/81196
24948 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
24949 exit condition comparing two IVs.
24951 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
24953 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
24954 profile to the dummy entry at the end of the list of architectures.
24955 * config/arm/arm-cpu-cdata.h: Regenerated.
24957 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24958 Michael Collison <michael.collison@arm.com>
24961 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
24963 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
24964 (*aarch64_reg_<mode>3_minus_mask): New pattern.
24965 (*aarch64_<optab>_reg_di3_mask2): New pattern.
24966 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
24967 of shift when the shift amount is masked with constant equal to
24968 the size of the mode.
24969 * config/aarch64/predicates.md (subreg_lowpart_operator): New
24972 2017-06-29 Martin Liska <mliska@suse.cz>
24974 * config/i386/i386.opt: Change range from [1,5] to [0,5].
24976 2017-06-29 Yury Gribov <tetra2005@gmail.com>
24979 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
24982 (edge_growth_cache_entry::edge_growth_cache_entry): New
24984 (reset_edge_growth_cache): Update to use constructor.
24986 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24988 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
24989 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
24990 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
24992 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
24994 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
24995 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
24997 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
24999 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
25000 (*-linux-uclibc*): Add t-uclibc tmake_file.
25001 * config/t-musl: New.
25002 * config/t-uclibc: New.
25004 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
25006 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
25008 (gen_comm_data): Emit architectural setting of arch_prof.
25009 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
25011 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
25012 (armv8-m.base, armv8-m.main): Likewise.
25013 * arm-protos.h (arm_build_target): Add profile field.
25014 (arch_option): Likewise.
25015 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
25017 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
25018 arm_active_target.profile.
25020 2017-06-28 Richard Biener <rguenther@suse.de>
25022 PR middle-end/81227
25023 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
25024 TYPE_OVERFLOW_WRAPS.
25025 * match.pd (negate_expr_p): Likewise.
25026 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
25027 fold_build2, not fold_binary.
25029 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25031 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
25032 Convert memory address to Pmode.
25033 (aarch64_print_operand): Assert MEM operands are always Pmode.
25035 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
25038 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
25039 Remove redundant if.
25040 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
25041 * config/arm/aarch-common-protos.h
25042 (aarch_forward_to_shift_is_not_shifted_re): Remove.
25043 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
25045 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
25048 * multiple_target.c (create_dispatcher_calls): Set the default
25049 clone to be static, not public.
25051 2017-06-28 Richard Biener <rguenther@suse.de>
25053 * tree-vect-loop.c (vectorizable_reduction): Move special
25054 cond reduction IV var creation ...
25055 (vect_create_epilog_for_reduction): ... here. Remove induction_index
25056 parameter. Use STMT_VINFO_VECTYPE.
25057 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
25060 2017-06-28 Martin Liska <mliska@suse.cz>
25063 * ipa-visibility.c (non_local_p): Handle visibility.
25065 2017-06-28 Martin Liska <mliska@suse.cz>
25068 * common.opt: Add IntegerRange to various options.
25069 * opt-functions.awk (integer_range_info): New function.
25070 * optc-gen.awk: Add integer_range_info to cl_options struct.
25071 * opts-common.c (decode_cmdline_option): Handle
25072 CL_ERR_INT_RANGE_ARG.
25073 (cmdline_handle_error): Likewise.
25074 * opts.c (print_filtered_help): Show valid interval in
25075 when --help is provided.
25076 * opts.h (struct cl_option): Add range_min and range_max fields.
25077 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
25079 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
25081 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
25082 (x * C EQ/NE y * C): New transformation.
25084 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
25086 * genmultilib (combination_space): Accept '+' in option names.
25088 2017-06-28 Martin Liska <mliska@suse.cz>
25091 * asan.c (instrument_derefs): Bail out inner references
25092 that are hard register variables.
25094 2017-06-28 Jakub Jelinek <jakub@redhat.com>
25097 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
25098 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
25100 2017-06-28 Richard Biener <rguenther@suse.de>
25102 * tree-vectorizer.h (vect_get_vec_defs): Remove.
25103 (vect_get_slp_defs): Adjust.
25104 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
25106 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
25108 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
25109 get_initial_defs_for_reduction instead of vect_get_vec_defs.
25110 (vectorizable_reduction): Adjust.
25111 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
25113 (vect_get_slp_defs): Likewise.
25114 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
25115 (vectorizable_bswap): Adjust.
25116 (vectorizable_call): Likewise.
25117 (vectorizable_conversion): Likewise.
25118 (vectorizable_assignment): Likewise.
25119 (vectorizable_shift): Likewise.
25120 (vectorizable_operation): Likewise.
25121 (vectorizable_store): Likewise.
25122 (vectorizable_condition): Likewise.
25123 (vectorizable_comparison): Likewise.
25125 2017-06-28 Michael Collison <michael.collison@arm.com>
25128 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
25130 (arm_gen_movmemqi): Removed unused variable 'i'.
25131 Convert 'for' loop into 'while' loop.
25132 (arm_expand_prologue): Remove last unnecessary set of insn.
25133 (thumb_pop): Remove unused variable 'pushed_words'.
25134 (thumb_exit): Remove last unnecessary set of regs_to_pop.
25136 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25138 * config/s390/predicates.md: Use s390_rel_address_ok_p.
25139 * config/s390/s390-protos.h: Add prototype of
25140 s390_rel_address_ok_p.
25141 * config/s390/s390.c (s390_got_symbol): New function.
25142 (s390_rel_address_ok_p): New function.
25143 (legitimize_pic_address): Use s390_rel_address_ok_p.
25144 (s390_load_got): Use s390_got_symbol.
25145 (s390_option_override): Issue error if
25146 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
25147 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
25149 * config/s390/s390.opt: New option mpic-data-is-text-relative.
25151 2017-06-27 Andrew Pinski <apinski@cavium.com>
25153 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
25154 (X * copysign (1.0, X)): New pattern.
25155 (X * copysign (1.0, -X)): New pattern.
25156 (copysign (-1.0, CST)): New pattern.
25158 2017-06-27 Joseph Myers <joseph@codesourcery.com>
25160 * genmultilib (combination_space): Remove variable.
25161 Validate reuse rules against regular expression for any sequence
25162 of multilib options in any order.
25164 2017-06-27 Michael Collison <michael.collison@arm.com>
25166 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
25167 call aarch64_split_simd_combine.
25168 * (aarch64_combine_internal<mode>): Delete pattern.
25169 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
25170 Allow register and subreg operands.
25172 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25174 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
25175 specific need, just fallback on defaults.
25176 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
25178 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25179 Olivier Hainque <hainque@adacore.com>
25181 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
25183 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
25184 targets. Pick a default if no particular attempt applied.
25185 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
25188 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25190 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
25191 (x86_64-wrs-vxworks7): Likewise.
25193 2017-06-27 Marek Polacek <polacek@redhat.com>
25196 * ubsan.c (instrument_null): Check get_base_address's result for null.
25198 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25200 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
25202 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
25204 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
25205 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
25206 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
25207 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
25208 New function types.
25209 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
25210 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
25211 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
25212 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
25213 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
25214 BUILT_IN_FEUPDATEENV): New builtins.
25215 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
25216 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
25217 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
25218 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
25220 (builtin_structptr_types): Adjust size.
25221 * tree.c (builtin_structptr_types): Add four entries.
25223 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25224 Olivier Hainque <hainque@adacore.com>
25226 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
25227 (TLS_SYM): New local macro, forcing reference to __tls__ on
25228 link command lines for VxWorks 7 RTPs, triggering initialization
25230 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
25231 OS features TLS support, true for RTPs on VxWorks 7.
25232 * config/vxworks.c (vxworks_override_options): Setup emutls
25235 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25237 * predict.c (test_prediction_value_range): Use -1U instead of -1
25238 to avoid narrowing conversion warning.
25239 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
25240 to avoid narrowing conversion warning.
25241 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
25243 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
25245 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
25247 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
25248 64bit configurations.
25249 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
25250 (SIZE_TYPE): Likewise.
25251 * config/vxworks.c (vxworks_emutls_var_fields): Use
25252 long_unsigned_type_node instead of unsigned_type_node as the offset
25253 field type, which is "pointer" mode in emutls.c.
25255 2017-06-27 Jakub Jelinek <jakub@redhat.com>
25258 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
25260 PR middle-end/81207
25261 * gimple-fold.c (replace_call_with_call_and_fold): Handle
25262 gimple_vuse copying separately from gimple_vdef copying.
25264 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25266 * value-prof.c (free_hist): Remove call to memset and the enclosing if
25269 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25270 Olivier Hainque <hainque@adacore.com>
25272 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
25273 for all vxworks7 targets.
25274 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
25275 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
25276 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
25277 variations for VX6/VX7 and 32/64bits later on in ...
25278 (VXWORKS_LIB_SPEC): Leverage new macros.
25279 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
25280 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
25282 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
25284 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
25285 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
25287 2017-06-26 Carl Love <cel@us.ibm.com>
25289 * config/rs6000/rs6000-c.c: Add support for built-in functions
25290 vector bool char vec_reve (vector bool char);
25291 vector signed char vec_reve (vector signed char);
25292 vector unsigned char vec_reve (vector unsigned char);
25293 vector bool int vec_reve (vector bool int);
25294 vector signed int vec_reve (vector signed int);
25295 vector unsigned int vec_reve (vector unsigned int);
25296 vector bool long long vec_reve (vector bool long long);
25297 vector signed long long vec_reve (vector signed long long);
25298 vector unsigned long long vec_reve (vector unsigned long long);
25299 vector bool short vec_reve (vector bool short);
25300 vector signed short vec_reve (vector signed short);
25301 vector double vec_reve (vector double);
25302 vector float vec_reve (vector float);
25303 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
25304 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
25305 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
25306 (altivec_vreve): New pattern.
25307 * config/rs6000/altivec.h (vec_reve): New define.
25308 * doc/extend.texi (vec_rev): Update the built-in documentation file
25309 for the new built-in functions.
25311 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25313 PR tree-optimization/71815
25314 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
25316 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
25318 (slsr_process_phi): Likewise.
25319 (replace_uncond_cands_and_profitable_phis): Don't replace a
25320 multiply candidate with a stride of 1 (copy or cast).
25321 (phi_incr_cost): Call uses_consumed_by_stmt rather than
25323 (lowest_cost_path): Likewise.
25324 (total_savings): Likewise.
25326 2017-06-26 Richard Biener <rguenther@suse.de>
25329 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
25330 Use def_builtin_pure for all gather builtins.
25332 2017-06-26 Richard Biener <rguenther@suse.de>
25334 PR tree-optimization/81203
25335 * tree-tailcall.c (find_tail_calls): Do not move stmts into
25336 non-dominating BBs.
25338 2017-06-26 Marek Polacek <polacek@redhat.com>
25341 * doc/invoke.texi: Document -Wmultistatement-macros.
25343 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
25345 * doc/sourcebuild.texi (ARM-specific attributes): Document new
25346 arm_neon_ok_no_float_abi effective target.
25348 2017-06-26 Richard Biener <rguenther@suse.de>
25350 PR tree-optimization/80928
25351 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
25352 (copy_bbs): Set BB_DUPLICATED flag early.
25353 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
25355 (execute_on_shrinking_pred): Likewise.
25356 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
25357 BB_DUPLICATED blocks.
25358 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
25359 iterate over all PHIs considering removal of *gsi.
25361 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
25363 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
25366 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25368 * config/rs6000/rs6000-string.c: (expand_block_clear,
25369 do_load_for_compare, select_block_compare_mode,
25370 compute_current_alignment, expand_block_compare,
25371 expand_strncmp_align_check, expand_strn_compare,
25372 expand_block_move, rs6000_output_load_multiple)
25373 Move functions related to string/block move/compare
25374 to a separate file.
25375 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
25376 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
25377 for this function which is now used in two files.
25378 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
25379 * config.gcc: Add rs6000-string.o to extra_objs for
25380 targets powerpc*-*-* and rs6000*-*-*.
25382 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
25385 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
25386 32-bit, since indexed is not valid for DImode.
25387 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
25388 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
25389 (define_peephole2 for Altivec d-form load): Add 32-bit support.
25390 (define_peephole2 for Altivec d-form store): Likewise.
25393 * multiple_target.c (create_dispatcher_calls): Only create the
25394 dispatcher call if the function is the default clone of a
25395 versioned function.
25397 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
25399 PR middle-end/80902
25400 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
25401 a call, force the call to not be a tail call.
25403 2017-06-23 Jeff Law <law@redhat.com>
25405 * doc/contrib.texi: Add entry for Steven Pemberton's work on
25408 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
25410 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
25411 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
25412 handling for early expansion of vector shifts (sl,sr,sra,rl).
25413 (builtin_function_type): Add vector shift right instructions
25414 to the unsigned argument list.
25416 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
25418 rtl-optimizatoin/79286
25419 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
25420 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
25421 trap. PIC register plus a const unspec without offset can never trap.
25423 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
25425 * tree.h (builtin_structptr_type): New type.
25426 (builtin_structptr_types): Declare new array.
25427 * tree.c (builtin_structptr_types): New array.
25428 (free_lang_data, build_common_tree_nodes): Use it.
25430 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
25433 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
25436 2017-06-22 Matt Turner <mattst88@gmail.com>
25438 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
25439 Lake models to skylake case. Assume skylake for unknown
25440 models with clflushopt.
25442 2017-06-22 Jeff Law <law@redhat.com>
25444 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
25445 frame sizes that do not satisfy aarch64_uimm12_shift.
25447 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
25449 * profile-count.h (apply_probability,
25450 apply_scale, probability_in): Fix checks for zero.
25452 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25454 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
25455 * doc/cppdiropts.texi (-I @var{dir}): Document it.
25457 2016-06-22 Richard Biener <rguenther@suse.de>
25459 * tree-vect-loop.c (vect_model_reduction_cost): Handle
25460 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
25461 REDUC_MAX_EXPR support.
25462 (vectorizable_reduction): Likewise.
25463 (vect_create_epilog_for_reduction): Likewise.
25465 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
25467 * match.pd (A / (1 << B) -> A >> B): New.
25468 * generic-match-head.c: Include optabs-tree.h.
25469 * gimple-match-head.c: Likewise.
25470 * optabs-tree.h (target_supports_op_p): New.
25471 * optabs-tree.c (target_supports_op_p): New.
25473 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25475 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
25476 $gcc_cv_ld --help output.
25477 (gcc_cv_ld_demangle): Likewise.
25478 (gcc_cv_ld_eh_frame_hdr): Likewise.
25479 (gcc_cv_ld_pie): Likewise.
25480 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
25481 (gcc_cv_ld_buildid): Likewise.
25482 (gcc_cv_ld_sysroot): Likewise.
25483 (ld_bndplt_support): Likewise.
25484 (ld_pushpopstate_support): Likewise.
25485 * configure: Regenerate.
25486 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
25488 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25491 * config/i386/sse.md (round<mode>2): Renumber match_dup and
25492 operands indexes to avoid gap between operands and match_dups.
25494 2017-06-21 Andrew Pinski <apinski@cavium.com>
25496 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
25497 Increment Arith_shift and Arith_shift_reg by 1.
25498 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
25500 * config/aarch64/aarch64.c (thunderx_tunings): Enable
25501 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
25502 (aarch64_strip_extend): Add new argument and test for it.
25503 (aarch64_cheap_mult_shift_p): New function.
25504 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
25505 add a cost if it is true.
25506 Update calls to aarch64_strip_extend.
25507 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
25509 2017-06-21 Andrew Pinski <apinski@cavium.com>
25511 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
25513 (thunderxt88): Likewise.
25514 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
25515 (thunderx_prefetch_tune): New variable.
25516 (thunderx2t99_prefetch_tune): Update for the correct values.
25517 (thunderxt88_tunings): New variable.
25518 (thunderx_tunings): Use thunderx_prefetch_tune instead of
25519 generic_prefetch_tune.
25520 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
25522 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25524 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
25525 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
25526 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
25527 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
25528 (aarch64_atomic_cas<mode>, GPI): Likewise.
25530 2017-06-21 Martin Liska <mliska@suse.cz>
25532 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
25533 statements on cold and hot labels.
25534 * predict.c (tree_estimate_probability_bb): Remove the
25535 prediction from this place.
25537 2017-06-21 Martin Liska <mliska@suse.cz>
25539 PR tree-optimization/79489
25540 * gimplify.c (maybe_add_early_return_predict_stmt): New
25542 (gimplify_return_expr): Call the function.
25543 * predict.c (tree_estimate_probability_bb): Remove handling
25545 * predict.def: Update comment about early return predictor.
25546 * gimple-predict.h (is_gimple_predict): New function.
25547 * predict.def: Change default value of early return to 66.
25548 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
25550 * passes.def: Put pass_strip_predict_hints to the beginning of
25553 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
25555 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
25556 FUNCTION_DECL declarations.
25557 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
25559 (dwaf2out_decl): Likewise.
25560 * godump.c (go_early_global_decl): Skip call to the real debug hook
25561 for FUNCTION_DECL declarations.
25562 * passes.c (rest_of_decl_compilation): Skip call to the
25563 early_global_decl debug hook for FUNCTION_DECL declarations, unless
25564 -fdump-go-spec is passed.
25566 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25568 * config/i386/i386.c (struct builtin_isa): New field pure_p.
25569 Reorder for compactness.
25570 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
25571 (def_builtin_pure, def_builtin_pure2): New functions.
25572 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
25574 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
25576 * match.pd (nop_convert): New predicate.
25577 ((A +- CST1) +- CST2): Allow some NOP conversions.
25579 2017-06-21 Jakub Jelinek <jakub@redhat.com>
25582 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
25583 with ctors/dtors if GOVD_SHARED is set.
25585 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25587 * config/aarch64/aarch64.md (movti_aarch64):
25588 Emit mov rather than orr.
25589 (movtf_aarch64): Likewise.
25590 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
25591 Emit mov rather than orr.
25593 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25595 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
25596 Swap alternatives, make integer dup more expensive.
25598 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
25600 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
25601 Return true for non-tls symbols.
25603 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
25605 * config/aarch64/aarch64-cores.def (cortex-a55): New.
25606 (cortex-a75): Likewise.
25607 (cortex-a75.cortex-a55): Likewise.
25608 * config/aarch64/aarch64-tune.md: Regenerate.
25609 * doc/invoke.texi (-mtune): Document new values for -mtune.
25611 2017-06-21 Tom de Vries <tom@codesourcery.com>
25613 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
25614 stack_size feature.
25615 (Effective-Target Keywords, Other attributes): Suggest using
25616 dg-add-options stack_size feature to get stack limit in stack_size
25617 effective target documentation.
25619 2017-06-21 Julian Brown <julian@codesourcery.com>
25620 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25622 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
25623 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
25624 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
25626 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
25627 attribute type list for neon_multiply.
25628 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
25629 attribute type list for neon_multiply.
25630 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
25631 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
25632 attribute type list for neon_multiply.
25633 * config/arm/types.md (crypto_pmull): Add.
25634 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
25635 attribute type list.
25637 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
25639 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
25642 2017-06-20 Jakub Jelinek <jakub@redhat.com>
25644 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
25645 to make sure not to dereference a NULL cost_classes_ptr pointer.
25647 2017-06-20 Carl Love <cel@us.ibm.com>
25649 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25650 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
25651 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
25652 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
25653 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
25654 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
25655 VMULOSW): New enum "unspec" values.
25656 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
25657 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
25658 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
25659 altivec_vmulosw): New patterns.
25660 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
25661 VMULOSW): Add definitions.
25663 2017-06-20 Julia Koval <julia.koval@intel.com>
25665 * config/i386/i386.c: Fix rounding expand for new pattern.
25666 * config/i386/subst.md: Fix pattern (parallel -> unspec).
25668 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25670 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
25671 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
25673 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25675 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
25678 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
25680 * config/aarch64/aarch64-cores.def: Rearrange to sort by
25681 architecture, then by implementer ID.
25682 * config/aarch64/aarch64-tune.md: Regenerate.
25684 2017-06-20 Richard Biener <rguenther@suse.de>
25686 PR middle-end/81097
25687 * fold-const.c (split_tree): Fold to type before negating.
25689 2017-06-20 David Malcolm <dmalcolm@redhat.com>
25691 * diagnostic-show-locus.c
25692 (selftest::test_fixit_deletion_affecting_newline): New function.
25693 (selftest::diagnostic_show_locus_c_tests): Call it.
25695 2017-06-20 Andreas Schwab <schwab@suse.de>
25698 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
25701 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25703 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
25704 __ARM_FEATURE_COPROC according to support.
25706 2017-06-20 Jakub Jelinek <jakub@redhat.com>
25708 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
25709 Rewritten to avoid overflow for > 32-bit pointers.
25712 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
25713 by removing enum keyword.
25714 (ubsan_type_descriptor): Likewise. Formatting fix.
25717 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
25718 splitter): Require TARGET_SSE2 in the condition.
25720 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
25723 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
25724 for doing vector set of SFmode on ISA 3.0.
25725 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
25726 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
25728 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
25729 SFmode value into a V4SF variable that was extracted from another
25730 V4SF variable without converting the element to double precision
25731 and back to single precision vector format.
25732 (vsx_insert_extract_v4sf_p9_2): Likewise.
25734 2017-06-19 Jakub Jelinek <jakub@redhat.com>
25736 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
25737 in UWHI to avoid undefined overflow.
25740 * ubsan.h (enum ubsan_encode_value_phase): New.
25741 (ubsan_encode_value): Change second argument to
25742 enum ubsan_encode_value_phase with default value of
25743 UBSAN_ENCODE_VALUE_GENERIC.
25744 * ubsan.c (ubsan_encode_value): Change second argument to
25745 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
25746 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
25747 create_tmp_var_raw instead of create_tmp_var and use a
25749 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
25750 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
25751 ubsan_encode_value callers.
25754 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
25755 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
25756 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
25758 2017-06-19 Richard Biener <rguenther@suse.de>
25760 PR middle-end/81118
25761 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
25762 estimates if we changed anything.
25764 2017-06-19 Richard Biener <rguenther@suse.de>
25766 PR tree-optimization/80887
25767 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
25768 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
25769 simplified lookups, then reset mprts_hook.
25770 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
25772 (try_to_simplify): Likewise.
25774 2017-06-19 Martin Liska <mliska@suse.cz>
25777 * gimplify.c (gimplify_switch_expr):
25778 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
25780 2017-06-19 Martin Liska <mliska@suse.cz>
25782 * doc/install.texi: Document that PGO runs in 4 stages.
25784 2017-06-19 Martin Liska <mliska@suse.cz>
25787 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
25788 to dispatcher function name.
25789 * multiple_target.c (replace_function_decl): New function.
25790 (create_dispatcher_calls): Redirect both edges and references.
25792 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
25794 * profile-count.c (profile_count::dump): Dump quality.
25795 (profile_count::differs_from_p): Update for unsigned val.
25796 * profile-count.h (profile_count_quality): New enum.
25797 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
25799 2017-06-19 Richard Biener <rguenther@suse.de>
25801 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
25802 struct function as arg.
25803 (estimate_numbers_of_iterations): Export overload with loop arg.
25804 (free_numbers_of_iterations_estimates_loop): Use an overload of
25805 free_numbers_of_iterations_estimates instead.
25806 * tree-cfg.c (remove_bb): Adjust.
25807 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
25808 * tree-parloops.c (gen_parallel_loop): Likewise.
25809 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
25811 (tree_unroll_loops_completely): Likewise.
25812 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
25813 Use an overload instead and export.
25814 (estimated_loop_iterations): Adjust.
25815 (max_loop_iterations): Likewise.
25816 (likely_max_loop_iterations): Likewise.
25817 (estimate_numbers_of_iterations): Take struct function as arg
25819 (loop_exits_before_overflow): Adjust.
25820 (free_numbers_of_iterations_estimates_loop): Use an overload.
25821 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
25822 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
25824 2017-06-19 Richard Biener <rguenther@suse.de>
25827 * ipa-prop.c (find_constructor_constant_at_offset): Handle
25828 RANGE_EXPR conservatively.
25830 2017-06-16 Carl Love <cel@us.ibm.com>
25832 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25833 definitions for vec_float, vec_float2, vec_floato,
25834 vec_floate built-ins.
25835 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
25836 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
25838 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
25839 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
25840 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
25841 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
25842 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
25843 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
25844 vec_floato): Add builtin defines.
25845 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
25846 Update the built-in documentation file for the new built-in
25849 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25851 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
25852 (mthumb): Mark as the negative of -marm.
25854 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25856 * doc/invoke.texi (ARM Options, -mcpu): Document supported
25858 (ARM Options, -mtune): Document that this accepts the same
25859 extension options as -mcpu.
25860 (ARM Options, -mfpu): Document addition of -mfpu=auto.
25862 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25864 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
25865 permitted extensions.
25867 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25869 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
25870 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
25871 (armv8-m.main): Add option +nodsp.
25872 * config/arm/arm-cpu-cdata.h: Regenerated.
25874 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25876 * config/arm/t-fuchsia: New file.
25877 * config.gcc (arm*-*-fuchsia*): Use it.
25879 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25881 * config/arm/t-symbian: Rewrite for new option infrastructure.
25883 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25885 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
25886 (MULTILIB_REQUIRED): Likewise.
25888 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25890 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
25891 (MULTILIB_RESUE): Likewise.
25892 (MULTILIB_MATCHES): Likewise.
25893 (MULTLIB_REQUIRED): Likewise.
25895 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25897 * config/arm/t-rtems: Rewrite for new option framework.
25899 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25901 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
25902 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
25903 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
25904 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
25905 * config/arm/t-multilib: ... here.
25906 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
25907 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
25908 armv7-a and armv8*-a when A-profile libraries have not been built.
25909 * config/arm/t-rmprofile: Rewrite.
25911 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25913 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
25914 with a backslash. Remove the backslash after substituting unescaped
25916 * doc/fragments.texi (MULTILIB_REUSE): Document it.
25918 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25920 * config.gcc: (arm*-*-*): When building a-profile libraries, force
25921 the driver to pass through the default setting of -mfloat-abi.
25922 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
25924 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
25925 (all_feat_combs): New rule.
25926 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
25928 * config/arm/t-aprofile: Rewrite.
25930 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25932 * config/arm/arm.h (FPUTYPE_AUTO): Define.
25933 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
25934 fpu is not specified by the user/command-line.
25935 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
25936 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
25937 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
25938 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
25939 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
25940 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
25942 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25944 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
25945 * config/arm/t-arm-elf: Rewritten.
25947 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25949 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
25950 have some floating-point instructions.
25951 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
25952 (TARGET_MAYBE_HARD_FLOAT): New macro.
25953 * config/arm/arm-builtins.c (arm_init_builtins): Use
25954 TARGET_MAYBE_HARD_FLOAT.
25955 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
25957 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25959 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
25960 (configargs.h): Include it.
25961 (arm_print_hint_for_fpu_option): New function.
25962 (arm_parse_fpu_option): New function.
25963 (candidate_extension): New class.
25964 (arm_canon_for_multilib): New function.
25965 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
25966 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
25967 (ARCH_CANONICAL_SPECS): New macro.
25968 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
25970 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25972 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
25973 are set after handling multilib fragments. Set target_cpu_default2
25976 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25978 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
25980 (arm*-*-*): Set target_cpu_default2 to a quoted string.
25981 * config/arm/parsecpu.awk (check_cpu): Validate any extension
25983 (check_arch): Likewise.
25984 * config/arm/arm.c (arm_configure_build_target): Handle
25985 TARGET_CPU_DEFAULT being a string constant. Scan any feature
25986 options in the default.
25988 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25990 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
25991 when an option is an alias of another.
25992 * config/arm/parsecpu.awk (optalias): New parser token.
25993 (gen_comm_data): Mark non-alias options as such. Emit entries
25994 for extension aliases.
25995 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
25996 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
25997 (armv6kz, armv6zk, armv6t2): Likewise.
25998 (armv7): Make vfpv3-d16 an alias.
25999 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
26001 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
26002 Sort in canonical order.
26003 (armv8-a): Sort in canonical order.
26004 (armv8.1-a, armv8.2-a): Likewise.
26005 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
26007 (cortex-a9): Sort in canonical order.
26008 * config/arm/arm.c (selftests.h): Include it.
26009 (arm_test_cpu_arch_data): New function.
26010 (arm_run_self_tests): New function.
26011 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
26012 (targetm): Move declaration to the end of the file.
26013 * arm-cpu-cdata.h: Regenerated.
26015 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26017 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
26018 call to target_mode_check describing the type of option passed.
26019 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
26020 (arm_target_thumb_only): Use arm_parse_arch_option_name or
26021 arm_parse_cpu_option_name to match parameters against list of
26023 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
26024 arm_arch_core_flags data structure.
26025 * config/arm/arm-cpu_cdata.h: Regenerated.
26027 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26029 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
26031 (arm_print_hint_for_cpu_option): Likewise.
26032 (arm_print_hint_for_arch_option): Likewise.
26033 (arm_parse_cpu_option_name): Likewise.
26034 (arm_parse_arch_option_name): Likewise.
26035 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
26036 of entries in the all_fpus list.
26037 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
26038 (arm_parse_cpu_option_name): Declare.
26039 (arm_parse_arch_option_name): Declare.
26040 (arm_parse_option_features): Declare.
26041 (arm_intialize_isa): Declare.
26042 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
26044 (gen_comm_data): ... here. Make definitions non-static.
26045 * config/arm/arm-cpu-data.h: Regenerated.
26046 * config/arm/arm-cpu-cdata.h: Regenerated.
26048 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26050 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
26051 (cpu_arch_extension): New structure.
26052 (cpu_arch_option, arch_option, cpu_option): New structures.
26053 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
26054 architecture types.
26055 (gen_data): Generate new format data tables.
26056 * config/arm/arm.c (cpu_tune): New structure.
26057 (cpu_option, processors): Delete.
26058 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
26059 (arm_print_hint_for_cpu_option): ... this and ...
26060 (arm_print_hint_for_arch_option): ... this.
26061 (arm_parse_arch_cpu_name): Delete. Replace with ...
26062 (arm_parse_cpu_option_name): ... this and ...
26063 (arm_parse_arch_option_name): ... this.
26064 (arm_unrecognized_feature): Change type of target parameter to
26066 (arm_parse_arch_cpu_features): Delete. Replace with ...
26067 (arm_parse_option_features): ... this.
26068 (arm_configure_build_target): Rework to use new configuration data
26070 (arm_print_tune_info): Rework for new configuration data tables.
26071 * config/arm/arm-cpu-data.h: Regenerated.
26072 * config/arm/arm-cpu.h: Regenerated.
26074 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26076 * Makefile.in (OBJS): Move sbitmap.o from here ...
26077 (OBJS-libcommon): ... to here.
26079 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26081 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
26082 (ISA_ALL_CRYPTO): New macro.
26083 (ISA_ALL_SIMD): New macro
26084 (ISA_ALL_FP): New macro.
26085 * config/arm/arm.c (fpu_bitlist): Update initializer.
26086 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
26088 (arm9e): Add fpu. Add option for nofp
26089 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
26090 (arm926ej-s, arm1026ej-s): Likewise.
26091 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
26092 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
26093 neon-fp16, neon-vfpv4, nofp and nosimd.
26094 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
26095 (cortex-a8): Add fpu. Add option for nofp.
26096 (cortex-a9): Add fpu. Add options for nosimd and nofp.
26097 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
26098 (cortex-r4f): Add fpu.
26099 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
26100 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
26102 (cortex-r8): Likewise.
26103 (cortex-m4): Add fpu. Add option for nofp.
26104 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
26105 (cortex-a17.cortex-a7): Likewise.
26106 (cortex-a32): Add fpu. Add options for crypto and nofp.
26107 (cortex-a35, cortex-a53): Likewise.
26108 (cortex-a57): Add fpu. Add option for crypto.
26109 (cortex-a72, cortex-a73): Likewise.
26110 (exynos-m1): Likewise.
26111 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
26112 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
26113 (cortex-m33): Add fpu. Add option for nofp.
26114 * config/arm/arm-cpu-cdata.h: Regenerated
26115 * config/arm/arm-cpu-data.h: Regenerated.
26117 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26119 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
26120 (armv5te, armv5tej): Likewise.
26121 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
26122 (armv7): Add options fp and vfpv3-d16.
26123 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
26124 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
26126 (armv7ve): Likewise.
26127 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
26128 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
26129 (armv8-a): Add nocrypto option.
26130 (armv8.1-a, armv8.2-a): Likewise.
26131 (armv8-m.main): add options fp, fp.dp and nofp.
26133 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26135 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
26137 (armv8-a+crc): Delete.
26138 (armv8.1-a): Add options simd, crypto and nofp.
26139 (armv8.2-a): Add options fp16, simd, crypto and nofp.
26140 (armv8.2-a+fp16): Delete.
26141 (armv8-m.main): Add option dsp.
26142 (armv8-m.main+dsp): Delete.
26143 (cortex-a8): Add fpu. Add nofp option.
26144 (cortex-a9): Add fpu. Add nofp and nosimd options.
26145 * config/arm/parsecpu.awk (gen_data): Generate option tables and
26146 link to main cpu and architecture data structures.
26147 (gen_comm_data): Only put isa attributes from the main architecture
26149 (option): New statement for architecture and CPU entries.
26150 * arm.c (struct cpu_option): New structure.
26151 (struct processors): Add entry for options.
26152 (arm_unrecognized_feature): New function.
26153 (arm_parse_arch_cpu_name): Ignore any characters after the first
26155 (arm_parse_arch_cpu_feature): New function.
26156 (arm_configure_build_target): Separate out any CPU and architecture
26157 features and parse separately. Don't error out if -mfpu=auto is
26158 used with only an architecture string.
26159 (arm_print_asm_arch_directives): New function.
26160 (arm_file_start): Call it.
26161 * config/arm/arm-cpu-cdata.h: Regenerated.
26162 * config/arm/arm-cpu-data.h: Likewise.
26163 * config/arm/arm-tables.opt: Likewise.
26165 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26167 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
26168 assembler when it is not -mfpu=auto.
26170 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26172 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
26173 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
26174 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
26175 (ASM_CPU_SPEC): Rewrite.
26176 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
26177 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
26178 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
26179 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
26180 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
26181 copied string is NUL-terminated. Also strip any characters prefixed
26183 (arm_rewrite_selected_arch): New function.
26184 (arm_rewrite_march): New function.
26186 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
26188 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
26189 (x_arm_cpu_string, x_arm_tune_string): Likewise.
26190 (march, mcpu, mtune): Convert to string-based options.
26191 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
26192 (arm_parse_arch_cpu_name): New function.
26193 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
26194 identify selected architecture or CPU.
26195 (arm_option_save): New function.
26196 (TARGET_OPTION_SAVE): Redefine.
26197 (arm_option_restore): Restore string options.
26198 (arm_option_print): Print string options.
26200 2017-06-16 Martin Sebor <msebor@redhat.com>
26202 PR tree-optimization/80933
26203 PR tree-optimization/80934
26204 * builtins.c (fold_builtin_3): Do not handle bcmp here.
26205 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
26206 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
26207 (gimple_fold_builtin): Call them.
26209 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26211 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
26212 as unlikely; update profile.
26214 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26216 * predict.c (force_edge_cold): Handle declaring edges impossible
26219 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26221 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
26223 (try_unroll_loop_completely): Fix reporting.
26225 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26227 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
26229 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
26232 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
26233 if given a non-constant argument for an intrinsic which requires a
26236 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26238 * profile.c (compare_freqs): New function.
26239 (branch_prob): Sort edge list.
26240 (find_spanning_tree): Assume that the list is priority sorted.
26242 2017-06-16 Richard Biener <rguenther@suse.de>
26244 PR tree-optimization/81090
26245 * passes.def (pass_record_bounds): Remove.
26246 * tree-pass.h (make_pass_record_bounds): Likewise.
26247 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
26248 make_pass_record_bounds): Likewise.
26249 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
26250 not free niter estimates at the beginning but at the end.
26251 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
26253 2017-06-16 Richard Biener <rguenther@suse.de>
26255 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
26256 initializer to workaround ICE in host GCC 4.8.
26258 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
26260 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
26262 (clone_inlined_nodes): Update.
26264 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26266 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
26267 prefetch settings, and enable prefetching by default at -O3.
26269 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26271 * config/aarch64/aarch64.c (aarch64_override_options_internal):
26272 Set flag_prefetch_loop_arrays according to tuning data.
26274 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26276 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
26277 New tune structure.
26278 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
26279 [Unrelated to main purpose of the patch] Place the pointer field last
26280 to enable type checking errors when tune structure are wrongly merged.
26281 * config/aarch64/aarch64.c (generic_prefetch_tune,)
26282 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
26283 (thunderx2t99_prefetch_tune): New tune constants.
26284 (tune_params *_tunings): Update all tunings (no functional change).
26285 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
26286 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
26287 from tunings structures.
26289 2017-06-16 Jakub Jelinek <jakub@redhat.com>
26292 * ubsan.c (instrument_null): Add T argument, use it instead
26293 of computing it based on IS_LHS.
26294 (instrument_object_size): Likewise.
26295 (pass_ubsan::execute): Adjust instrument_null and
26296 instrument_object_size callers to pass gimple_get_lhs or
26297 gimple_assign_rhs1 result to it. Use instrument_null instead of
26298 calling get_base_address and instrument_mem_ref. Handle
26299 aggregate call arguments for object-size sanitization.
26301 2017-06-16 Yury Gribov <tetra2005@gmail.com>
26303 PR tree-optimization/81089
26304 * tree-vrp.c (is_masked_range_test): Validate operands of
26307 2017-06-15 Martin Sebor <msebor@redhat.com>
26310 * dumpfile.c (dump_register): Avoid calling memset to initialize
26311 a class with a default ctor.
26312 * gcc.c (struct compiler): Remove const qualification.
26313 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
26314 * hash-table.h: Ditto.
26315 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
26317 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
26318 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
26320 * params.h (struct param_info): Make struct members non-const.
26321 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
26322 with default initialization.
26323 * vec.h (vec_copy_construct, vec_default_construct): New helper
26325 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
26326 with vec_copy_construct.
26327 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
26328 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
26329 * doc/invoke.texi (-Wclass-memaccess): Document.
26331 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26333 * emit-rtl.h (is_leaf): Update comment about local
26334 register allocator.
26336 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
26339 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
26340 for a variable to have a section before checking if the section has a
26342 Set section to.persistent if persistent attribute is set.
26343 Warn if .persistent attribute is used on an automatic variable.
26345 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
26347 PR rtl-optimization/80474
26348 * reorg.c (update_block): Do not ignore instructions in a delay slot.
26350 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
26352 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
26355 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
26357 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
26358 (casesi): Emit bounds checking as RTL.
26359 (casesi_internal_mips16_<mode>): Remove bounds checking.
26361 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
26363 * config/xtensa/xtensa.c (xtensa_option_override): Append
26364 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
26365 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
26366 xtensa_doloop_hooks): Define unconditionally.
26367 (xtensa_reorg_loops): Only call reorg_loops in the presence of
26369 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
26370 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
26371 for it in xtensa_option_override.
26372 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
26373 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
26375 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
26377 * doc/cppopts.texi: Document '-' special value to -MF.
26379 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
26381 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
26382 (cortex_a53_fconst): Likewise.
26383 (cortex_a53_fpmul): Likewise.
26384 (cortex_a53_f_load_64): Likewise.
26385 (cortex_a53_f_load_many): Likewise.
26386 (cortex_a53_advsimd_alu): Likewise.
26387 (cortex_a53_advsimd_alu_q): Likewise.
26388 (cortex_a53_advsimd_mul): Likewise.
26389 (cortex_a53_advsimd_mul_q): Likewise.
26390 (fpmac bypass): Add new bypass for fpmac-fpmac case.
26391 Add missing fmul, r2f_cvt and fconst cases.
26393 2017-06-14 Richard Biener <rguenther@suse.de>
26395 PR middle-end/81088
26396 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
26398 (fold_binary_loc): When associating do not treat pre-existing
26399 TREE_OVERFLOW on literal constants as a reason to allow
26400 TREE_OVERFLOW on associated literal constants.
26402 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26404 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
26405 (MASK_FEATURES): New macro.
26406 * config/sparc/sparc.c (sparc_option_override): Remove the special
26407 handling of -mfpu and generalize it to all MASK_FEATURES switches.
26409 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26411 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
26412 a division of 0 if non-call exceptions are enabled.
26414 2017-06-14 Andrew Pinski <apinski@cavium.com>
26415 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26418 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
26419 Improve vector initialization code gen for only variable case.
26421 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
26423 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
26425 2017-06-14 Richard Biener <rguenther@suse.de>
26427 PR tree-optimization/81083
26428 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
26431 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26433 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
26434 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
26435 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
26436 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
26437 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
26438 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
26440 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26442 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
26443 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
26445 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26447 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
26449 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26451 * config/rs6000/t-rtems: Don't handle SPE.
26453 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26455 * config/rs6000/t-linux: Don't handle SPE.
26457 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26459 * config/rs6000/eabispe.h: Delete file.
26461 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26463 * config/rs6000/t-spe: Delete file.
26465 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26467 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
26468 (rs6000_legitimate_offset_address_p): Return false for anything in
26469 V2SImode or V2SFmode.
26471 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
26473 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
26474 except V2SF and V2SI. Rearrange the vector modes, and add comments.
26475 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
26477 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
26478 (rs6000_legitimate_offset_address_p): Ditto.
26479 (rs6000_emit_move): Ditto.
26480 (rs6000_init_builtins): Remove V4HI_type_node.
26482 2017-06-13 Martin Liska <mliska@suse.cz>
26485 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
26486 (gate_asan): Likewise.
26487 * asan.h (asan_no_sanitize_address_p): Remove the function.
26488 (sanitize_flags_p): New function.
26489 * builtins.def: Fix coding style.
26490 * common.opt: Use renamed enum value.
26491 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
26492 * doc/extend.texi: Document no_sanitize attribute.
26493 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
26494 to SANITIZE_UNDEFINED_NONDEFAULT.
26495 * gcc.c (sanitize_spec_function): Use the renamed enum value.
26496 * gimple-fold.c (optimize_atomic_compare_exchange_p):
26497 Use sanitize_flags_p.
26498 * gimplify.c (gimplify_function_tree): Likewise.
26499 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
26500 * opts.c (parse_no_sanitize_attribute): New function.
26501 (common_handle_option): Use renamed enum value.
26502 * opts.h (parse_no_sanitize_attribute): Declare.
26503 * tree.c (sanitize_flags_p): New function.
26504 * tree.h: Declared here.
26505 * tsan.c: Use sanitize_flags_p.
26506 * ubsan.c (ubsan_expand_null_ifn): Likewise.
26507 (instrument_mem_ref): Likewise.
26508 (instrument_bool_enum_load): Likewise.
26509 (do_ubsan_in_current_function): Remove the function.
26510 (pass_ubsan::execute): Use sanitize_flags_p.
26511 * ubsan.h: Remove do_ubsan_in_current_function
26512 * tree-cfg.c (print_no_sanitize_attr_value): New function.
26513 (dump_function_to_file): Use it here.
26515 2017-06-13 Martin Jambor <mjambor@suse.cz>
26517 PR tree-optimization/80803
26518 PR tree-optimization/81063
26519 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
26520 (propagate_subaccesses_across_link): Enqueue subtree whenever
26521 necessary instead of relying on the caller.
26523 2017-06-13 Martin Jambor <mjambor@suse.cz>
26525 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
26526 that have a first_link.
26527 (sort_and_splice_var_accesses): Do not check first_link before
26529 (subtree_mark_written_and_enqueue): Likewise.
26530 (propagate_all_subaccesses): Likewise and do not stop at first
26531 parent with a first_link.
26533 2017-06-13 Martin Jambor <mjambor@suse.cz>
26535 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
26538 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26540 * match.pd: New pattern.
26542 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26544 * tree-vrp.c (is_masked_range_test): New function.
26545 (register_edge_assert_for): Determine ranges for
26548 2017-06-13 Yury Gribov <tetra2005@gmail.com>
26550 PR tree-optimization/67328
26551 * fold-const.c (maskable_range_p): New function.
26552 (build_range_check): Generate bittests if possible.
26554 2017-06-13 Martin Liska <mliska@suse.cz>
26556 * gimple-pretty-print.c (dump_probability): Add new argument.
26557 (dump_edge_probability): Dump both probability and count.
26558 (dump_gimple_label): Likewise.
26559 (dump_gimple_bb_header): Likewise.
26561 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
26564 * config/avr/avr-devices.c: Fix indentation.
26565 * config/avr/gen-avr-mmcu-specs.c: Dito.
26567 2017-06-13 Richard Biener <rguenther@suse.de>
26569 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
26570 instead get vector type from stmt_info.
26571 (vectorizable_reduction): Adjust. Remove dead code.
26573 2017-06-13 Richard Biener <rguenther@suse.de>
26575 PR middle-end/81065
26576 * fold-const.c (extract_muldiv_1): Remove bogus distribution
26577 case of C * (x * C2 + C3).
26578 (fold_addr_of_array_ref_difference): Properly fold index difference.
26580 2017-06-12 David S. Miller <davem@davemloft.net>
26583 * config/sparc/sparc.md (return expander): Emit frame blockage if
26584 function uses alloca.
26586 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
26588 * combine.c (make_field_assignment): Check len rather than the mode
26589 precision when calling force_to_mode.
26591 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
26593 Support multilibs and devices that see flash in RAM address range.
26596 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
26597 (avr_mcu_t) <flash_pm_offset>: New field.
26598 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
26599 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
26600 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
26601 (AVR_TINY_PM_OFFSET): Remove macro.
26602 * config/avr/avr.opt (-mshort-calls): New option.
26603 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
26604 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
26605 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
26606 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
26607 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
26608 instead of avr_arch->have_jmp_call.
26609 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
26610 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
26611 avr_arch->flash_pm_offset to define.
26612 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
26613 new field flash_pm_offset. Add entry for avrxmega3.
26614 (avr_texinfo): Add entry for avrxmega3.
26615 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
26616 attiny212, attiny214,
26617 attiny412, attiny414, attiny416, attiny417,
26618 attiny814, attiny816, attiny817,
26619 attiny1614, attiny1616, attiny1617,
26620 attiny3214, attiny3216, attiny3217.
26621 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
26622 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
26623 (avr_print_operand_address) [AVR_TINY]: Same.
26624 (avr_asm_init_sections) <readonly_data_section>: Only patch
26625 callback if avr_arch->flash_pm_offset = 0.
26626 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
26627 for rodata if avr_arch->flash_pm_offset != 0.
26628 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
26629 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
26630 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
26631 (m_options): Append opt_rcall.
26632 (m_dirnames): Append dir_rcall.
26633 * config/avr/t-multilib: Regenerate.
26635 * configure.ac [target=avr]: Check whether avrxmega3 default
26636 linker description file works as needed.
26637 * configure: Regenerate.
26638 * doc/avr-mmcu.texi: Regenerate.
26639 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
26640 <__AVR_ARCH__>: Document avrxmega3 and 103.
26641 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
26642 <__AVR_SHORT_CALLS__>: Document it.
26643 <__AVR_PM_BASE_ADDRESS__>: Document it.
26644 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
26645 (AVR Variable Attributes) <progmem>: Document this is
26646 not needed for avrxmega3.
26647 (AVR Named Address Spaces) <__flash>: Dito.
26649 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
26651 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
26653 2017-06-12 Doug Rupp <rupp@adacore.com>
26655 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
26656 Append vxworks-stdint.h to the tm_file list.
26657 * config/vxworks-stdint.h: New file.
26659 2017-06-12 Martin Liska <mliska@suse.cz>
26661 PR tree-optimization/81041
26662 * tree-profile.c (gimple_gen_ic_func_profiler):
26663 Create an extra BB in profile-generate
26664 (gimple_gen_time_profiler): Likewise.
26666 2017-06-12 Jakub Jelinek <jakub@redhat.com>
26668 PR tree-optimization/81003
26669 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
26670 (update_range_test): Use it instead of force_gimple_operand_gsi.
26672 2017-06-12 Richard Biener <rguenther@suse.de>
26674 PR tree-optimization/81053
26675 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
26676 with backedge value not defined in loop. Simplify def stmt
26679 2017-06-11 Tom de Vries <tom@codesourcery.com>
26682 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
26684 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
26685 nvptx_cannot_force_const_mem.
26687 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26689 * opts.c (finish_options): Move test for flag_split_stack after
26690 it has been initialized.
26692 2017-06-11 Jason Merrill <jason@redhat.com>
26694 * tree.h (id_equal): New.
26695 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
26696 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
26697 instead of strcmp of IDENTIFIER_POINTER.
26699 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26701 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
26702 (mark_all_inlined_calls_cdtor): Fix formating.
26703 (inline_transform): Rescale profile before inlining.
26705 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26707 * cgraph.h (cgraph_edge::clone): Update prototype.
26708 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
26709 (cgraph_node::create_clone): Update.
26710 (cgraph_node::create_version_clone): Update.
26711 * tree-inline.c (copy_bb): Update.
26712 (expand_call_inline): Update.
26714 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
26716 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
26717 factored out from ...
26718 (rs6000_emit_prologue): ... here.
26720 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
26722 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
26723 factored out from ...
26724 (rs6000_emit_prologue): ... here.
26726 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26728 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
26730 (handle_missing_profiles): Fix computation of tp_first_run.
26731 (counts_to_freqs): Do not touch freqs when count is 0.
26733 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
26735 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
26738 2017-06-10 Tom de Vries <tom@codesourcery.com>
26740 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
26741 attributes): Document signal effective target.
26743 2017-06-10 Tom de Vries <tom@codesourcery.com>
26745 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
26746 Document effective target stack_size.
26748 2017-06-09 David Malcolm <dmalcolm@redhat.com>
26750 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
26751 to the edit_context if they can be auto-applied.
26753 2017-06-9 Ian Lance Taylor <iant@golang.org>
26755 * opts.c (finish_options): If -fsplit-stack, disable implicit
26756 -forder-blocks-and-partition.
26757 * doc/invoke.texi (Optimize Options): Document that when using
26758 -fsplit-stack -forder-blocks-and-partition is not implicitly
26761 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26763 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
26764 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
26765 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
26766 * builtins.def (abort, trap, unreachable): Declare cold.
26767 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
26768 * tree-core.h (ECF_COLD): New.
26769 * tree.c (set_call_expr_flags): Handle ECF_COLD.
26770 (build_common_builtin_nodes): Mark unreachable and abort as cold.
26772 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26774 * predict.c (unlikely_executed_stmt_p): Cleanup.
26776 2017-06-09 Richard Biener <rguenther@suse.de>
26778 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
26779 model if the ref is always written to.
26781 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26783 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
26785 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26787 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
26790 2017-06-09 Tom de Vries <tom@codesourcery.com>
26793 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
26794 "target cannot support label values" when encountering LABEL_REF.
26796 2017-06-09 Martin Liska <mliska@suse.cz>
26798 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
26799 (gimple_gen_ic_func_profiler): Emit direct comparison
26800 of __gcov_indirect_call_callee with NULL.
26801 (gimple_gen_time_profiler): Change probability from
26802 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
26804 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26806 * profile.c (edge_gcov_counts): Turn to pointer.
26807 (compute_branch_probabilities, compute_branch_probabilities): Update.
26808 (branch_prob): Do not clear edge_gcov_count.
26809 * profile.h (edge_gcov_counts): Turn to pointer.
26810 (edge_gcov_count): Update.
26812 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26814 * gimple.h (gimple_check_failed): Mark cold.
26816 2017-06-09 Richard Biener <rguenther@suse.de>
26818 PR tree-optimization/66623
26819 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
26820 refactor check_reduction into two parts, properly computing
26821 whether we have to check reduction validity for outer loop
26824 2017-06-09 Richard Biener <rguenther@suse.de>
26826 PR tree-optimization/79483
26827 * graphite-scop-detection.c (order): New global.
26828 (get_order): Compute bb to order mapping that satisfies code
26829 generation constraints.
26830 (cmp_pbbs): New helper.
26831 (build_scops): Start domwalk at entry block, sort generated
26834 2017-06-09 Richard Biener <rguenther@suse.de>
26836 PR middle-end/81007
26837 * ipa-polymorphic-call.c
26838 (ipa_polymorphic_call_context::restrict_to_inner_class):
26839 Skip FIELD_DECLs with error_mark_node type.
26840 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
26843 2017-06-09 Martin Liska <mliska@suse.cz>
26845 * predict.c (struct branch_predictor): New struct.
26846 (test_prediction_value_range): New test.
26847 (predict_c_tests): New function.
26848 * selftest-run-tests.c (selftest::run_tests): Run the function.
26849 * selftest.h: Declare new tests.
26851 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
26854 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
26855 gen_add3_insn did not fail.
26856 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
26857 r0, construct that number in a temporary reg and add that reg to r0.
26858 If asked to put the result in r0 as well, fail.
26860 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
26862 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26863 for early expansion of vec_eqv.
26865 2017-06-08 Jakub Jelinek <jakub@redhat.com>
26867 PR middle-end/81005
26868 * ubsan.c (instrument_null): Avoid pointless code temporary.
26869 (pass_ubsan::execute): Instrument aggregate arguments of calls.
26871 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
26875 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
26878 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
26881 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26883 * predict.c (unlikely_executed_edge_p): Move ahead.
26884 (probably_never_executed_edge_p): Use it.
26886 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
26888 PR middle-end/79988
26889 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
26890 gimple_call_builtin_p call.
26892 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26894 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
26895 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
26896 rtl_check_failed_type2, rtl_check_failed_code1,
26897 rtl_check_failed_code2, rtl_check_failed_code_mode,
26898 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
26899 rtvec_check_failed_bounds, rtl_check_failed_flag,
26900 _fatal_insn_not_found, _fatal_insn): Likewise.
26901 * tree.h (tree_contains_struct_check_failed,
26902 tree_check_failed, tree_not_check_failed,
26903 tree_class_check_failed, tree_range_check_failed,
26904 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
26905 tree_vec_elt_check_failed, phi_node_elt_check_failed,
26906 tree_operand_check_failed, omp_clause_check_failed,
26907 omp_clause_operand_check_failed, omp_clause_range_check_failed):
26910 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26912 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
26913 flag_branch_probabilities.
26914 * ipa-inline.c (edge_badness): Likewise.
26915 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
26916 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
26917 * predict.c (maybe_hot_frequency_p): Likewise.
26918 (probably_never_executed): Likewise.
26919 * sched-ebb.c (schedule_ebbs): Likewise.
26920 * sched-rgn.c (find_single_block_region): Likewise.
26921 * tracer.c (tail_duplicate): Likewise.
26923 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26925 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
26926 longer requires x_flag_profile_use.
26928 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26930 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
26931 instead of flag_reorder_blocks_and_partition.
26932 * dbxout.c (dbxout_function_end): Likewise.
26933 * dwarf2out.c (gen_subprogram_die): Likewise.
26934 * haifa-sched.c (sched_create_recovery_edges): Likewise.
26935 * hw-doloop.c (reorg_loops): Likewise.
26936 * varasm.c (assemble_start_function,
26937 assemble_end_function): Likewise.
26938 (decide_function_section): Do not check for
26939 flag_reorder_blocks_and_partition.
26941 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
26943 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26945 (chkp_get_hard_register_fake_addr_expr): Ditto.
26946 (chkp_build_addr_expr): Add check for hard reg case.
26947 (chkp_parse_array_and_component_ref): Ditto.
26948 (chkp_find_bounds_1): Ditto.
26949 (chkp_process_stmt): Don't generate bounds store for
26952 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26954 * predict.c (maybe_hot_bb_p): Do not check profile status.
26955 (maybe_hot_edge_p): Likewise.
26956 (probably_never_executed): Check for zero counts even if profile
26958 (unlikely_executed_edge_p): New function.
26959 (unlikely_executed_stmt_p): New function.
26960 (unlikely_executed_bb_p): New function.
26961 (set_even_probabilities): Use unlikely predicates.
26962 (combine_predictions_for_bb): Likewise.
26963 (predict_paths_for_bb): Likewise.
26964 (predict_paths_leading_to_edge): Likewise.
26965 (determine_unlikely_bbs): New function.
26966 (estimate_bb_frequencies): Use it.
26967 (compute_function_frequency): Use zero counts even if profile is
26969 * profile-count.h: Fix typo.
26971 2017-08-08 Julia Koval <julia.koval@intel.com>
26973 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
26974 _mm512_mask_cvtsepi16_storeu_epi8,
26975 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
26976 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
26977 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
26978 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
26979 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
26980 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
26981 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
26982 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
26983 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
26984 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
26985 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
26986 __builtin_ia32_pmovuswb256mem_mask,
26987 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
26988 __builtin_ia32_pmovwb512mem_mask): New builtins.
26990 2017-08-08 Julia Koval <julia.koval@intel.com>
26992 PR target/73350,80862
26993 * config/i386/subst.md (round): Fix round pattern.
26994 * config/i386/i386.c (ix86_erase_embedded_rounding):
26995 Fix erasing rounding for the fixed pattern.
26997 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26999 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
27001 2017-06-08 Martin Liska <mliska@suse.cz>
27003 PR gcov-profile/80911
27004 * gcov.c (block_info::block_info): New constructor.
27006 2017-06-07 Carl Love <cel@us.ibm.com>
27008 * config/rs6000/rs6000-c: The return type of the following
27009 built-in functions was implemented as int not long long. Fix sign
27010 of return value for the unsigned version of vec_mulo and vec_mule.
27011 vector unsigned long long vec_bperm (vector unsigned long long,
27012 vector unsigned char)
27013 vector signed long long vec_mule (vector signed int,
27015 vector unsigned long long vec_mule (vector unsigned int,
27016 vector unsigned int)
27017 vector signed long long vec_mulo (vector signed int,
27019 vector unsigned long long vec_mulo (vector unsigned int,
27020 vector unsigned int)
27021 * doc/extend.texi: Fix the documentation for the built-in
27024 2017-06-07 Carl Love <cel@us.ibm.com>
27027 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
27030 2017-06-07 Carl Love <cel@us.ibm.com>
27032 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
27033 support, Generate doublehv for signed int/float for BE case only.
27035 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
27037 * doc/invoke.texi (mcx16): Rewrite.
27039 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27041 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
27042 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
27043 *mov<mode>_softfloat, and an anonymous splitter): Use
27044 nonimmediate_operand instead of rs6000_nonimmediate_operand.
27046 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27048 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
27050 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27051 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
27052 (rs6000_debug_reg_global): Adjust.
27053 (rs6000_init_hard_regno_mode_ok): Adjust.
27054 (rs6000_dbx_register_number): Adjust.
27055 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
27056 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27057 Remove SPE_ACC and SPEFSCR.
27058 (REG_ALLOC_ORDER): Ditto.
27059 (FRAME_POINTER_REGNUM): Change to 111.
27060 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
27061 (REG_CLASS_NAMES): Ditto.
27062 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
27063 (REGISTER_NAMES): Ditto.
27064 (ADDITIONAL_REG_NAMES): Ditto.
27065 (rs6000_reg_names): Ditto.
27066 * config/rs6000/rs6000.md: Renumber some register number
27069 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27071 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
27073 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
27074 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
27076 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
27077 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
27078 Delete the SPE high registers.
27079 (REG_ALLOC_ORDER): Ditto.
27080 (enum reg_class): Remove SPE_HIGH_REGS.
27081 (REG_CLASS_NAMES): Ditto.
27082 (REG_CLASS_CONTENTS): Delete the SPE high registers.
27083 (REGISTER_NAMES): Ditto.
27084 (rs6000_reg_names): Ditto.
27085 * doc/tm.texi.in: Remove SPE as example.
27086 * doc/tm.texi: Regenerate.
27088 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27090 * config/rs6000/8540.md (ppc8540_brinc): Delete.
27091 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
27092 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
27093 * config/rs6000/rs6000.md (type): Remove "brinc".
27095 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27097 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
27098 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
27099 * config/rs6000/linuxspe.h: Delete file.
27100 * config/rs6000/rs6000.md: Don't include spe.md.
27101 * config/rs6000/spe.h: Delete file.
27102 * config/rs6000/spe.md: Delete file.
27103 * config/rs6000/t-rs6000: Remove spe.md.
27105 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27107 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
27108 (reg_or_none500mem_operand): Delete.
27109 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
27110 instead of reg_or_none500mem_operand.
27112 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27114 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27115 handling of SPE flags.
27116 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
27118 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27120 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
27122 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
27123 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
27124 paired_divv2sf3): Similar.
27125 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
27126 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
27127 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
27129 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
27130 Rename the paired_* instruction patterns.
27131 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
27133 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
27134 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
27135 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
27136 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
27137 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
27138 PAIRED_VECTOR_MODE.
27139 (struct machine_function): Delete field spe_insn_chain_scanned_p.
27140 (spe_func_has_64bit_regs_p): Delete.
27141 (spe_expand_predicate_builtin): Delete.
27142 (spe_expand_evsel_builtin): Delete.
27143 (TARGET_DWARF_REGISTER_SPAN): Do not define.
27144 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
27145 (invalid_e500_subreg): Delete.
27146 (rs6000_legitimize_address): Always force_reg op2 as well, for
27147 paired single memory accesses.
27148 (rs6000_member_type_forces_blk): Delete.
27149 (rs6000_spe_function_arg): Delete.
27150 (rs6000_expand_unop_builtin): Delete SPE handling.
27151 (rs6000_expand_binop_builtin): Ditto.
27152 (spe_expand_stv_builtin): Delete.
27153 (bdesc_2arg_spe): Delete.
27154 (spe_expand_builtin): Delete.
27155 (spe_expand_predicate_builtin): Delete.
27156 (spe_expand_evsel_builtin): Delete.
27157 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
27158 (spe_init_builtins): Delete.
27159 (spe_func_has_64bit_regs_p): Delete.
27160 (savres_routine_name): Delete "info" parameter. Adjust callers.
27161 (rs6000_emit_stack_reset): Ditto.
27162 (rs6000_dwarf_register_span): Delete.
27163 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
27164 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
27165 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
27167 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
27169 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
27170 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
27171 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
27172 mulv2sf3, divv2sf3): Delete expanders.
27174 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27176 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
27178 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27180 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
27181 * config/rs6000/rs6000.c: Ditto.
27183 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27185 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
27186 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
27187 comparison_operator.
27189 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27191 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
27192 * config/rs6000/rs6000.opt: Ditto.
27193 * config/rs6000/t-rtems: Ditto.
27195 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27197 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
27198 TARGET_E500_SINGLE by 0, simplify.
27199 * config/rs6000/rs6000.c: Ditto.
27200 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
27201 (spe_build_register_parallel): Delete.
27202 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
27203 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
27204 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
27205 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
27206 (E500_CONVERT): Delete.
27207 * config/rs6000/spe.md: Remove many patterns and all define_constants.
27209 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
27211 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
27212 * config/rs6000/dfp.md: Ditto.
27213 (negdd2, *negdd2_fpr): Merge.
27214 (absdd2, *absdd2_fpr): Merge.
27215 (negtd2, *negtd2_fpr): Merge.
27216 (abstd2, *abstd2_fpr): Merge.
27217 * config/rs6000/e500.h: Delete file.
27218 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
27219 TARGET_FPRS by 1 and simplify.
27220 * config/rs6000/rs6000-c.c: Ditto.
27221 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
27222 TARGET_DF_SPE by 0.
27223 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
27225 * config/rs6000/rs6000.md: Ditto.
27226 (floatdidf2, *floatdidf2_fpr): Merge.
27227 (move_from_CR_gt_bit): Delete.
27228 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
27229 (E500_CR_IOR_COMPARE): Delete.
27230 (All patterns that require !TARGET_FPRS): Delete.
27231 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
27233 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27235 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
27237 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27239 * graphds.c (add_edge): Intitialize edge's attached data.
27240 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
27241 pointer parameter. Call pointed function on each edge during
27242 graph traversing. Skip traversing the edge when the function
27244 (graphds_dfs, graphds_scc): Ditto.
27245 (for_each_edge): New parameter. Pass the new parameter to callback
27247 * graphds.h (skip_edge_callback): New function pointer type.
27248 (graphds_dfs, graphds_scc): New function pointer parameter.
27249 (graphds_edge_callback, for_each_edge): New parameter.
27251 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27253 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
27254 out code checking if runtime alias check is possible to below ...
27255 Call the new function.
27256 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
27257 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
27259 2017-06-07 Marek Polacek <polacek@redhat.com>
27262 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
27263 TYPE_OVERFLOW_WRAPS check.
27265 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27267 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
27268 if versioning is required.
27269 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
27270 peeling with the check for versioning.
27272 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27274 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
27275 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
27276 Set true to new parameter if new ssa variable is defined.
27277 (vect_gen_vector_loop_niters): Refactor. Set range information
27278 for the new vector loop bound variable.
27279 (vect_do_peeling): Ditto.
27281 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27283 * tree-affine.c (ssa.h): Include header file.
27284 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
27285 has wrapping overflow behavior.
27287 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27289 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
27291 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27293 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
27294 (tree_to_aff_combination): ... here.
27296 2017-06-07 Bin Cheng <bin.cheng@arm.com>
27298 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
27299 reg_pressure model function.
27300 (ivopts_global_cost_for_size): Delete.
27301 (determine_set_costs, iv_ca_recount_cost): Call new model function
27302 ivopts_estimate_reg_pressure.
27304 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27306 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
27307 expensive than udiv. Remove floating point cases from mod.
27309 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27311 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
27312 Increase idiv cost.
27314 2017-06-07 Tamar Christina <tamar.christina@arm.com>
27316 * config/aarch64/aarch64.md
27317 (copysignsf3): Fix mask generation.
27319 2017-06-07 Jakub Jelinek <jakub@redhat.com>
27321 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
27323 (class dump_manager): Add register_dumps method.
27324 * dumpfile.c: Include langhooks.h.
27325 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
27326 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
27327 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
27328 (dump_manager::dump_register): Start with 512 entries instead of 32.
27329 (dump_manager::register_dumps): New method.
27330 * toplev.c (general_init): Instead of invoking register_dumps
27331 langhook, invoke register_dumps method on the dump manager.
27332 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
27335 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
27337 * doc/md.texi: Clarify the restrictions on a define_insn condition.
27338 Say that # requires an associated define_split to exist, and that
27339 the define_split must be suitable for use after register allocation.
27341 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27343 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
27344 (compute_outgoing_frequencies): Also initialize zero counts.
27345 (find_many_sub_basic_blocks): Do not produce uninitialized profile
27346 around loops; preserve more of profile when nothing changes.
27348 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
27350 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
27352 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
27353 * config/arm/arm-cpu-cdata.h: Regenerate.
27354 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
27355 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
27356 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
27357 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
27358 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
27360 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
27362 * config/arm/t-rmprofile: Likewise.
27363 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
27365 2017-06-06 David S. Miller <davem@davemloft.net>
27368 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
27369 blockage if function uses alloca.
27371 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27373 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
27374 New "uid" fields to hold pretty-print IDs of group and ref.
27375 Memory references are now identified as <group_id>:<ref_id>
27376 instead of using [random] addresses.
27377 (dump_mem_details): Simplify, no functional change.
27378 (dump_mem_ref): Simplify and make output more concise.
27379 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
27380 (find_or_create_group): Initialize group uid.
27381 (record_ref): Initialize ref uid. Improve debug output.
27382 (prune_group_by_reuse, should_issue_prefetch_p,)
27383 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
27384 (mark_nontemporal_store, determine_loop_nest_reuse):
27385 Improve debug output.
27387 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
27389 * dbgcnt.def (prefetch): New debug counter.
27390 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
27391 (schedule_prefetches): Stop issueing prefetches if debug counter
27394 2017-06-06 Tom de Vries <tom@codesourcery.com>
27396 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
27397 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
27399 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27401 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
27402 Use aarch64_reg_or_zero predicate for operand 4.
27403 (aarch64_compare_and_swap<mode> define_insn_and_split):
27404 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
27405 (aarch64_store_exclusive<mode>): Likewise for operand 2.
27407 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
27409 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
27410 (arm_compute_save_core_reg_mask): This.
27411 (thumb1_compute_save_reg_mask): Rename into ...
27412 (thumb1_compute_save_core_reg_mask): This.
27413 (arm_compute_save_reg0_reg12_mask): Adapt comment.
27414 (arm_compute_frame_layout): Likewise.
27416 2017-06-06 Richard Biener <rguenther@suse.de>
27418 PR tree-optimization/80974
27419 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
27420 keep or clear leaders SSA info.
27422 2017-06-06 Tom de Vries <tom@codesourcery.com>
27424 * config/nvptx/nvptx.c (split_mode_p): New function.
27425 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
27427 2017-06-06 Tom de Vries <tom@codesourcery.com>
27429 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
27431 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27434 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
27437 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27439 * shrink-wrap.c (handle_simple_exit): Update profile.
27440 (try_shrink_wrapping): Upate profile.
27442 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27444 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
27445 (tree_guess_outgoing_edge_probabilities): New.
27446 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
27447 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
27449 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27451 * ipa-split.c (split_function): Initialize return bb profile.
27453 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
27455 * profile.c (compute_branch_probabilities): Also initialize
27456 EXIT_BLOCK profile.
27458 2017-06-06 Richard Biener <rguenther@suse.de>
27460 PR tree-optimization/80928
27461 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
27462 (vect_analyze_loop_operations): Properly guard analysis for
27464 (vect_transform_loop): Likewise.
27465 (vect_analyze_loop_2): Also reset SLP type on PHIs.
27466 (vect_model_induction_cost): Do not cost for pure SLP.
27467 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
27468 of induction in inner loop vectorization.
27469 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
27470 (vect_get_and_check_slp_defs): Handle vect_induction_def.
27471 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
27473 (vect_analyze_slp_cost_1): Cost induction.
27474 (vect_detect_hybrid_slp_stmts): Handle PHIs.
27475 (vect_get_slp_vect_defs): Likewise.
27476 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
27477 (vect_transform_stmt): Handle SLP reductions.
27478 * tree-vectorizer.h (vectorizable_induction): Adjust.
27480 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27482 * config/rs6000/rs6000.c (make_resolver_func): Update
27483 init_lowered_empty_function call.
27485 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
27487 * doc/invoke.texi: Document the -fprofile-abs-path option.
27488 * common.opt (fprofile-abs-path): New option.
27489 * gcov-io.h (gcov_write_filename): Declare.
27490 * gcov-io.c (gcov_write_filename): New function.
27491 * coverage.c (coverage_begin_function): Use gcov_write_filename.
27492 * profile.c (output_location): Likewise.
27494 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27496 * shring-wrap.c: Revert accidental commit.
27498 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
27500 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
27502 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27504 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
27506 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
27507 profile in callgraph edge.
27508 * profile-count.h (apply_probability): If THIS is 0, then result is 0
27509 (apply_scale): Likewise.
27510 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
27511 Also scale profile when inlining function with zero profile.
27512 (initialize_cfun): Update exit block profile even when it is zero.
27513 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
27514 when profile is read.
27516 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
27518 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
27519 (CLONE_*): New constants to define the processors we can generate
27520 code for with the target_clone attribute.
27521 (rs6000_clone_map): New array to identify which clone processors
27522 the current program is running on.
27523 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
27524 target_clone attribute.
27525 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
27526 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
27527 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
27528 (cpu_expand_builtin): Add support for target_clone attribute.
27529 (rs6000_valid_attribute_p): Allow "default" attribute.
27530 (get_decl_name): New debug function to simplify printing the
27531 current function name in debugging statements.
27532 (rs6000_clone_priority): New functions to support the target_clone
27533 attribute, and be able to generate code to switch between ISA 2.05
27534 through ISA 3.0 (power6 through power9).
27535 (rs6000_compare_version_priority): Likewise.
27536 (rs6000_get_function_versions_dispatcher): Likewise.
27537 (make_resolver_func): Likewise.
27538 (add_condition_to_bb): Likewise.
27539 (dispatch_function_versions): Likewise.
27540 (rs6000_generate_version_dispatcher_body): Likewise.
27541 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
27542 (fusion_gpr_load_p): Fix a spacing issue.
27543 * doc/extend.texi (Common Function Attributes): Document that the
27544 PowerPC supports the target_clone attribute.
27546 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
27548 * config/arm/arm.h: explain F symbol found in description of ARM
27549 register allocation in its legend.
27551 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27553 * config/mips/frame-header-opt.c: Include profile-count.h.
27554 * config/riscv/riscv.c: Include profile-count.h
27556 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
27558 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
27560 (sm_set_flag_if_changed): Add bbs field.
27561 (execute_sm_if_changed_flag_set): Pass BBS.
27562 (execute_sm): Update.
27564 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27566 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
27569 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27571 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
27572 (peephole2): New peephole2 to emit the above.
27573 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
27575 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27577 * config/aarch64/aarch64.c (define_peephole2 above
27578 *sub_<shift>_<mode>): New peephole.
27580 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27582 * config/i386/i386.c (make_resolver_func): Update.
27583 * Makefile.in: Add profile-count.h and profile-count.o
27584 * auto-profile.c (afdo_indirect_call): Update to new API.
27585 (afdo_set_bb_count): Update.
27586 (afdo_propagate_edge): Update.
27587 (afdo_propagate_circuit): Update.
27588 (afdo_calculate_branch_prob): Update.
27589 (afdo_annotate_cfg): Update.
27590 * basic-block.h: Include profile-count.h
27591 (struct edge_def): Turn count to profile_count.
27592 (struct basic_block_def): Likewie.
27593 (REG_BR_PROB_BASE): Move to profile-count.h
27594 (RDIV): Move to profile-count.h
27595 * bb-reorder.c (max_entry_count): Turn to profile_count.
27596 (find_traces): Update.
27597 (rotate_loop):Update.
27598 (connect_traces):Update.
27599 (sanitize_hot_paths):Update.
27600 * bt-load.c (migrate_btr_defs): Update.
27601 * cfg.c (RDIV): Remove.
27602 (init_flow): Use alloc_block.
27603 (alloc_block): Uninitialize count.
27604 (unchecked_make_edge): Uninitialize count.
27605 (check_bb_profile): Update.
27606 (dump_edge_info): Update.
27607 (dump_bb_info): Update.
27608 (update_bb_profile_for_threading): Update.
27609 (scale_bbs_frequencies_int): Update.
27610 (scale_bbs_frequencies_gcov_type): Update.
27611 (scale_bbs_frequencies_profile_count): New.
27612 * cfg.h (update_bb_profile_for_threading): Update.
27613 (scale_bbs_frequencies_profile_count): Declare.
27614 * cfgbuild.c (compute_outgoing_frequencies): Update.
27615 (find_many_sub_basic_blocks): Update.
27616 * cfgcleanup.c (try_forward_edges): Update.
27617 (try_crossjump_to_edge): Update.
27618 * cfgexpand.c (expand_gimple_tailcall): Update.
27619 (construct_exit_block): Update.
27620 * cfghooks.c (verify_flow_info): Update.
27621 (dump_bb_for_graph): Update.
27622 (split_edge): Update.
27623 (make_forwarder_block): Update.
27624 (duplicate_block): Update.
27625 (account_profile_record): Update.
27626 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
27627 (get_estimated_loop_iterations): Update.
27628 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
27629 (single_likely_exit): Update.
27630 * cfgloopmanip.c (scale_loop_profile): Update.
27632 (set_zero_probability): Update.
27633 (lv_adjust_loop_entry_edge): Update.
27634 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
27635 (purge_dead_edges): Update.
27636 (rtl_account_profile_record): Update.
27637 * cgraph.c (cgraph_node::create): Uninitialize count.
27638 (symbol_table::create_edge): Uninitialize count.
27639 (cgraph_update_edges_for_call_stmt_node): Update.
27640 (cgraph_edge::dump_edge_flags): Update.
27641 (cgraph_node::dump): Update.
27642 (cgraph_edge::maybe_hot_p): Update.
27643 * cgraph.h: Include profile-count.h
27644 (create_clone), create_edge, create_indirect_edge): Update.
27645 (cgraph_node): Turn count to profile_count.
27646 (cgraph_edge0: Likewise.
27647 (make_speculative, clone): Update.
27648 (create_edge): Update.
27649 (init_lowered_empty_function): Update.
27650 * cgraphclones.c (cgraph_edge::clone): Update.
27651 (duplicate_thunk_for_node): Update.
27652 (cgraph_node::create_clone): Update.
27653 * cgraphunit.c (cgraph_node::analyze): Update.
27654 (cgraph_node::expand_thunk): Update.
27655 * final.c (dump_basic_block_info): Update.
27656 * gimple-streamer-in.c (input_bb): Update.
27657 * gimple-streamer-out.c (output_bb): Update.
27658 * graphite.c (print_global_statistics): Update.
27659 (print_graphite_scop_statistics): Update.
27660 * hsa-brig.c: Include basic-block.h.
27661 * hsa-dump.c: Include basic-block.h.
27662 * hsa-gen.c (T sum_slice): Update.
27663 (convert_switch_statements):Update.
27664 * hsa-regalloc.c: Include basic-block.h.
27665 * ipa-chkp.c (chkp_produce_thunks): Update.
27666 * ipa-cp.c (struct caller_statistics): Update.
27667 (init_caller_stats): Update.
27668 (gather_caller_stats): Update.
27669 (ipcp_cloning_candidate_p): Update.
27670 (good_cloning_opportunity_p): Update.
27671 (get_info_about_necessary_edges): Update.
27672 (dump_profile_updates): Update.
27673 (update_profiling_info): Update.
27674 (update_specialized_profile): Update.
27675 (perhaps_add_new_callers): Update.
27676 (decide_about_value): Update.
27677 (ipa_cp_c_finalize): Update.
27678 * ipa-devirt.c (struct odr_type_warn_count): Update.
27679 (struct decl_warn_count): Update.
27680 (struct final_warning_record): Update.
27681 (possible_polymorphic_call_targets): Update.
27682 (ipa_devirt): Update.
27683 * ipa-fnsummary.c (redirect_to_unreachable): Update.
27684 * ipa-icf.c (sem_function::merge): Update.
27685 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
27686 * ipa-inline.c (compute_uninlined_call_time): Update.
27687 (compute_inlined_call_time): Update.
27688 (want_inline_small_function_p): Update.
27689 (want_inline_self_recursive_call_p): Update.
27690 (edge_badness): Update.
27691 (lookup_recursive_calls): Update.
27692 (recursive_inlining): Update.
27693 (inline_small_functions): Update.
27694 (dump_overall_stats): Update.
27695 (dump_inline_stats): Update.
27696 * ipa-profile.c (ipa_profile_generate_summary): Update.
27697 (ipa_propagate_frequency): Update.
27698 (ipa_profile): Update.
27699 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27700 * ipa-utils.c (ipa_merge_profiles): Update.
27701 * loop-doloop.c (doloop_modify): Update.
27702 * loop-unroll.c (report_unroll): Update.
27703 (unroll_loop_runtime_iterations): Update.
27704 * lto-cgraph.c (lto_output_edge): Update.
27705 (lto_output_node): Update.
27706 (input_node): Update.
27707 (input_edge): Update.
27708 (merge_profile_summaries): Update.
27709 * lto-streamer-in.c (input_cfg): Update.
27710 * lto-streamer-out.c (output_cfg): Update.
27711 * mcf.c (create_fixup_graph): Update.
27712 (adjust_cfg_counts): Update.
27713 (sum_edge_counts): Update.
27714 * modulo-sched.c (sms_schedule): Update.
27715 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
27716 * predict.c (maybe_hot_count_p): Update.
27717 (probably_never_executed): Update.
27718 (dump_prediction): Update.
27719 (combine_predictions_for_bb): Update.
27720 (propagate_freq): Update.
27721 (handle_missing_profiles): Update.
27722 (counts_to_freqs): Update.
27723 (rebuild_frequencies): Update.
27724 (force_edge_cold): Update.
27725 * predict.h: Include profile-count.h
27726 (maybe_hot_count_p, counts_to_freqs): UPdate.
27727 * print-rtl-function.c: Do not include cfg.h
27728 * print-rtl.c: Include basic-block.h
27729 * profile-count.c: New file.
27730 * profile-count.h: New file.
27731 * profile.c (is_edge_inconsistent): Update.
27732 (correct_negative_edge_counts): Update.
27733 (is_inconsistent): Update.
27734 (set_bb_counts): Update.
27735 (read_profile_edge_counts): Update.
27736 (compute_frequency_overlap): Update.
27737 (compute_branch_probabilities): Update; Initialize and deinitialize
27739 (branch_prob): Update.
27740 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
27741 (edge_gcov_count): New.
27742 (bb_gcov_count): New.
27743 * shrink-wrap.c (try_shrink_wrapping): Update.
27744 * tracer.c (better_p): Update.
27745 * trans-mem.c (expand_transaction): Update.
27746 (ipa_tm_insert_irr_call): Update.
27747 (ipa_tm_insert_gettmclone_call): Update.
27748 * tree-call-cdce.c: Update.
27749 * tree-cfg.c (gimple_duplicate_sese_region): Update.
27750 (gimple_duplicate_sese_tail): Update.
27751 (gimple_account_profile_record): Update.
27752 (execute_fixup_cfg): Update.
27753 * tree-inline.c (copy_bb): Update.
27754 (copy_edges_for_bb): Update.
27755 (initialize_cfun): Update.
27756 (freqs_to_counts): Update.
27757 (copy_cfg_body): Update.
27758 (expand_call_inline): Update.
27759 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
27760 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
27761 (try_unroll_loop_completely): Update.
27762 (try_peel_loop): Update.
27763 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
27764 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
27765 * tree-ssa-loop-split.c (connect_loops): Update.
27766 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
27767 * tree-ssa-reassoc.c (branch_fixup): Update.
27768 * tree-ssa-tail-merge.c (replace_block_by): Update.
27769 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
27770 (compute_path_counts): Update.
27771 (update_profile): Update.
27772 (recompute_probabilities): Update.
27773 (update_joiner_offpath_counts): Update.
27774 (estimated_freqs_path): Update.
27775 (freqs_to_counts_path): Update.
27776 (clear_counts_path): Update.
27777 (ssa_fix_duplicate_block_edges): Update.
27778 (duplicate_thread_path): Update.
27779 * tree-switch-conversion.c (case_bit_test_cmp): Update.
27780 (struct switch_conv_info): Update.
27781 * tree-tailcall.c (decrease_profile): Update.
27782 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
27783 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
27784 * value-prof.c (check_counter): Update.
27785 (gimple_divmod_fixed_value): Update.
27786 (gimple_mod_pow2): Update.
27787 (gimple_mod_subtract): Update.
27788 (gimple_ic_transform): Update.
27789 (gimple_stringop_fixed_value): Update.
27790 * value-prof.h (gimple_ic): Update.
27792 2017-06-02 Carl Love <cel@us.ibm.com>
27794 * config/rs6000/rs6000-c: Add support for built-in functions
27795 vector double vec_doublee (vector signed int);
27796 vector double vec_doublee (vector unsigned int);
27797 vector double vec_doublee (vector float);
27798 vector double vec_doubleh (vector signed int);
27799 vector double vec_doubleh (vector unsigned int);
27800 vector double vec_doubleh (vector float);
27801 vector double vec_doublel (vector signed int);
27802 vector double vec_doublel (vector unsigned int);
27803 vector double vec_doublel (vector float);
27804 vector double vec_doubleo (vector signed int);
27805 vector double vec_doubleo (vector unsigned int);
27806 vector double vec_doubleo (vector float);.
27807 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
27808 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
27810 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
27811 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
27812 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
27814 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
27815 vec_doublel, vec_doubleh.
27816 * doc/extend.texi: Update the built-in documentation file for the
27817 new built-in functions.
27819 2017-06-02 David Malcolm <dmalcolm@redhat.com>
27822 * ipa-inline-analysis.c (free_growth_caches): Set
27823 edge_removal_hook_holder to NULL after removing it.
27825 2017-06-02 Sudakshina Das <sudi.das@arm.com>
27827 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
27828 comparision with zero.
27830 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27831 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27832 for early expansion of vec_min and vec_max builtins.
27833 (builtin_function_type): Add min/max unsigned variants to those
27834 identified as having unsigned arguments.
27836 2017-06-02 Olivier Hainque <hainque@adacore.com>
27838 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
27840 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27842 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
27843 Use VALL_F16 iterator rather than VALL.
27845 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27847 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
27848 Emit CBNZ inside loop when doing a strong exchange and comparing
27849 against zero. Generate the CC flags after the loop.
27851 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
27853 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
27854 (dl_section_ref): New.
27855 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
27856 On AIX, append an expression to subtract the size of the
27857 section length to dl_section_ref.
27859 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27861 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27862 for early expansion of vector absolute builtins.
27864 2017-06-02 Richard Biener <rguenther@suse.de>
27866 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
27867 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
27869 2017-06-02 Richard Biener <rguenther@suse.de>
27871 PR tree-optimization/80948
27872 * tree-tailcall.c (find_tail_calls): Track stmts to move in
27873 stmt order as well.
27875 2017-06-02 Richard Biener <rguenther@suse.de>
27877 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
27879 * tree-vect-stmts.c (process_use): Do not mark backedge defs
27880 for inductions as relevant.
27882 2017-06-02 Richard Biener <rguenther@suse.de>
27884 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
27885 (vectorizable_induction): ... this. Remove dead code.
27887 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
27889 * builtins. (expand_builtin_alloca): Remove second parameter and
27890 infer its value from the first parameter instead.
27891 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
27893 2017-06-02 Jakub Jelinek <jakub@redhat.com>
27895 PR rtl-optimization/80903
27896 * loop-doloop.c (add_test): Unshare sequence.
27898 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
27900 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
27902 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
27904 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
27906 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
27907 xlogue_layout::get_instance, logue_layout::xlogue_layout,
27908 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
27909 (xlogue_layout::get_stub_rtx): Make static.
27910 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
27911 (xlogue_layout::compute_stub_managed_regs): Rename to...
27912 (xlogue_layout::count_stub_managed_regs): ...this.
27913 (xlogue_layout::is_stub_managed_reg): New function.
27914 (xlogue_layout::m_stub_names): Rename to...
27915 (xlogue_layout::s_stub_names): ...this, make static.
27916 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
27917 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
27918 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
27919 xlogue_layout::s_stub_names): Instantiate statics.
27920 (stub_managed_regs): Remove.
27921 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
27922 (disable_call_ms2sysv_xlogues): Rename to...
27923 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
27924 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
27926 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
27927 change after reload_completed.
27928 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
27930 (ix86_expand_prologue): Likewise.
27931 (ix86_expand_epilogue): Likewise.
27932 (ix86_expand_split_stack_prologue): Likewise.
27933 (ix86_compute_frame_layout): Remove frame parameter ...
27934 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
27935 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
27937 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
27938 (ix86_frame): Move from here ...
27939 * config/i386/i386.h (ix86_frame): ... to here.
27940 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
27941 complete ix86_frame data structure instead. Remove some_ld_name.
27943 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
27945 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
27946 symbols that hold a DECL_VALUE_EXPR.
27948 2017-06-01 Martin Jambor <mjambor@suse.cz>
27950 PR tree-optimization/80898
27951 * tree-sra.c (process_subtree_disqualification): Removed.
27952 (disqualify_candidate): Do not acll
27953 process_subtree_disqualification.
27954 (subtree_mark_written_and_enqueue): New function.
27955 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
27956 RHS has been disqualified and re-queue LHS if necessary. Apart
27957 from that, ignore disqualified RHS.
27959 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27961 * config/s390/s390.c (s390_emit_epilogue): Disable early return
27962 address fetch for z10 or later.
27964 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27966 * config/arc/arc.md (tst_movb): Add guard when splitting.
27968 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27970 * config/arc/arc.c (arc_can_eliminate): Test against
27971 arc_frame_pointer_needed.
27973 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27975 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
27976 to prevent store reordering.
27977 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
27978 (type): Add block type.
27979 (stack_tie): Define special instruction to be used in
27982 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27984 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
27985 constraint. It is not valid for the pattern.
27986 (noncommutative_binary_comparison): Likewise.
27988 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27990 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
27993 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27995 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
27996 be used by the reg-alloc.
27998 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28000 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
28001 reg-alloc when having mul64 or mul32x16 instructions.
28002 (mulsidi3): Likewise.
28003 (umulsidi3): Likewise.
28004 (mulsi32x16): New pattern.
28005 (mulsi64): Likewise.
28006 (mulsidi64): Likewise.
28007 (umulsidi64): Likewise.
28008 (MUL32x16_REG): Define.
28009 (mul64_600): Use MUL32x16_REG.
28010 (mac64_600): Likewise.
28011 (umul64_600): Likewise.
28012 (umac64_600): Likewise.
28014 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
28016 * config/arc/arc.md (mulsi3_700): Make it commutative.
28018 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
28020 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
28022 (*sign_extendsidi2_insn): Likewise for movstosw.
28024 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
28026 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
28027 the type of the input discriminant value. Convert the
28028 discriminant value of signedness vary.
28030 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
28032 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
28033 Add to -Wall section.
28035 2017-06-01 Richard Biener <rguenther@suse.de>
28037 PR middle-end/66313
28038 * fold-const.c (fold_plusminus_mult_expr): If the factored
28039 factor may be zero use a wrapping type for the inner operation.
28040 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
28041 and handle moved defs.
28042 (process_assignment): Properly guard the unary op case. Return a
28043 tri-state indicating that moving the stmt before the call may allow
28044 to continue. Pass through to_move.
28045 (find_tail_calls): Handle moving unrelated defs before
28048 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
28051 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
28052 splitter result in the canonical way.
28054 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
28056 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
28057 also for 32bit target. Update insn attributes.
28058 (zero-extendsidi2 splitter): Allow all registers for operand 1.
28060 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
28062 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
28063 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
28064 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
28065 (_mm_maskz_min_ss): New intrinsics.
28067 2017-05-31 Martin Liska <mliska@suse.cz>
28069 * tree-vect-loop.c (vect_create_epilog_for_reduction):
28070 Change comment style to one we normally use.
28071 (vectorizable_reduction): Likewise.
28072 (vectorizable_induction): Likewise.
28073 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
28074 (vectorizable_call): Likewise.
28075 (vectorizable_simd_clone_call): Likewise.
28076 (vectorizable_conversion): Likewise.
28077 (vectorizable_assignment): Likewise.
28078 (vectorizable_shift): Likewise.
28079 (vectorizable_operation): Likewise.
28080 (vectorizable_store): Likewise.
28081 (vectorizable_load): Likewise.
28082 * tree-vectorizer.h: Likewise.
28084 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
28086 * passes.c (emergency_dump_function): New.
28087 * tree-pass.h (emergency_dump_function): Declare.
28088 * plugin.c (plugins_internal_error_function): Remove.
28089 * plugin.h (plugins_internal_error_function): Remove declaration.
28090 * toplev.c (internal_error_function): New static function. Use it...
28091 (general_init): ...here.
28093 2017-05-31 Graham Markall <graham.markall@embecosm.com>
28095 * config/arc/arc.c (arc_print_operand): Handle constant operands.
28096 (arc_rtx_costs): Add costs for new patterns.
28097 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
28098 * config/arc/predicates.md: Add _1_2_3_operand predicate.
28100 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28102 * tree-ssa-strlen.c (get_next_strinfo): New function.
28103 (get_stridx_plus_constant): Use it.
28104 (zero_length_string): Likewise.
28105 (adjust_related_strinfos): Likewise.
28106 (adjust_last_stmt): Likewise.
28108 2017-05-31 Richard Biener <rguenther@suse.de>
28111 * config/i386/i386.c (ix86_expand_builtin): Remove assert
28112 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
28114 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
28116 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
28117 loop_vinfo argument and use of dependence distance vectors.
28118 Check instead whether the two references differ only in their
28119 initial value and assume that they have the same alignment if the
28120 difference is a multiple of the vector alignment.
28121 (vect_analyze_data_refs_alignment): Update call accordingly.
28123 2017-05-31 Martin Liska <mliska@suse.cz>
28126 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
28128 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28130 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
28131 (create_intersect_range_checks): Move from ...
28132 * tree-data-ref.c (create_intersect_range_checks_index)
28133 (create_intersect_range_checks): ... to here.
28134 (create_runtime_alias_checks): New function factored from ...
28135 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
28136 here. Call above function.
28137 * tree-data-ref.h (create_runtime_alias_checks): New function.
28139 2017-05-31 Bin Cheng <bin.cheng@arm.com>
28141 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
28142 segment length for dr_b and compute it in wide_int.
28144 2017-05-31 Richard Biener <rguenther@suse.de>
28146 PR tree-optimization/80906
28147 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
28148 and pass through iv_map.
28149 (copy_bb_and_scalar_dependences): Adjust.
28150 (translate_pending_phi_nodes): Likewise.
28151 (copy_loop_close_phi_args): Handle code-generating IVs instead
28154 2017-05-30 David Malcolm <dmalcolm@redhat.com>
28156 * diagnostic-color.c (color_dict): Add "type-diff".
28157 (parse_gcc_colors): Update comment.
28158 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
28159 -fdiagnostics-show-template-tree and -fno-elide-type.
28160 (GCC_COLORS): Add type-diff to example.
28162 (-fdiagnostics-show-template-tree): New.
28163 (-fno-elide-type): New.
28164 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
28165 the pp_format_decoder callback. Call any m_format_postprocessor's
28167 (pretty_printer::pretty_printer): Initialize
28168 m_format_postprocessor.
28169 (pretty_printer::~pretty_printer): Delete any
28170 m_format_postprocessor.
28171 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
28172 (class format_postprocessor): New class.
28173 (struct pretty_printer::format_decoder): Document the new parameters.
28174 (struct pretty_printer::m_format_postprocessor): New field.
28175 * tree-diagnostic.c (default_tree_printer): Update for new
28176 bool and const char ** params.
28177 * tree-diagnostic.h (default_tree_printer): Likewise.
28179 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
28181 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
28182 (lwa_operand): Delete rs6000_gen_cell_microcode test.
28183 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28184 rs6000_gen_cell_microcode code.
28185 (rs6000_final_prescan_insn): Delete.
28186 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
28187 "warn-cell-microcode" entries.
28188 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
28189 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
28190 throughout. Change cc_reg_not_micro_cr0_operand to
28191 cc_reg_not_cr0_operand throughout.
28192 (*extendhi<mode>2_noload): Delete.
28193 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
28194 (mwarn-cell-microcode): Delete.
28195 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
28196 -mgen-cell-microcode and -mwarn-cell-microcode.
28198 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
28201 * config/i386/constraints.md (Yd): New constraint.
28203 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
28204 and (?Yd, r) alternatives. Update insn attributes.
28205 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
28206 and (?*Yd, r) alternatives. Update insn attributes.
28207 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
28209 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
28211 * gimplify.c (gimplify_modify_expr): Don't create a
28212 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
28215 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
28217 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
28219 2017-05-30 Richard Biener <rguenther@suse.de>
28221 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
28222 and reduc_def fields.
28223 (STMT_VINFO_REDUC_TYPE): New define.
28224 (STMT_VINFO_REDUC_DEF): Likewise.
28225 (vect_force_simple_reduction): Adjust prototype.
28226 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
28227 (vect_is_simple_reduction): Remove check_reduction argument.
28228 (vect_force_simple_reduction): Adjust and set
28229 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28230 (vectorizable_reduction): Do not re-do reduction analysis
28231 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
28232 * tree-parloops.c (gather_scalar_reductions): Adjust.
28234 2017-05-30 Richard Biener <rguenther@suse.de>
28236 PR middle-end/80901
28237 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
28240 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
28242 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
28243 Introduce unknown_misalignment parameter and remove vf.
28244 (vect_peeling_hash_get_lowest_cost):
28245 Pass unknown_misalignment parameter.
28246 (vect_enhance_data_refs_alignment):
28247 Fix unsupportable data ref treatment.
28249 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28251 * tree-vect-data-refs.c (vect_get_data_access_cost):
28252 Workaround for SLP handling.
28253 (vect_enhance_data_refs_alignment):
28254 Compute costs for doing no peeling at all, compare to the best
28255 peeling costs so far and avoid peeling if cheaper.
28257 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28259 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
28260 Return peeling info and set costs to zero for unlimited cost
28262 (vect_enhance_data_refs_alignment): Also inspect all datarefs
28263 with unknown misalignment. Compute and costs for unknown
28264 misalignment, compare them to the costs for known misalignment
28265 and choose the cheapest for peeling.
28267 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28269 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
28270 (vect_get_peeling_costs_all_drs): Create function.
28271 (vect_peeling_hash_get_lowest_cost):
28272 Use vect_get_peeling_costs_all_drs.
28273 (vect_peeling_supportable): Create function.
28274 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
28276 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
28278 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
28279 DR_HAS_NEGATIVE_STEP.
28280 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
28281 (vect_enhance_data_refs_alignment): Use.
28282 (vect_duplicate_ssa_name_ptr_info): Use.
28283 * tree-vectorizer.h (dr_misalignment): Use.
28284 (known_alignment_for_access_p): Use.
28286 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28289 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
28291 (has_section_name): New function.
28293 2017-05-30 Martin Liska <mliska@suse.cz>
28296 * auto-profile.c (get_function_decl_from_block): Fix
28299 2017-05-30 Richard Biener <rguenther@suse.de>
28301 PR middle-end/80876
28302 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
28304 2017-05-30 Martin Liska <mliska@suse.cz>
28306 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
28307 * dumpfile.h (struct dump_file_info): Remove ctors.
28309 2017-05-30 Martin Liska <mliska@suse.cz>
28311 * predict.def: Fix GNU coding style.
28313 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28315 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
28316 Mark 'to' argument with ATTRIBUTE_UNUSED.
28318 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
28320 * config/xtensa/xtensa.c (xtensa_emit_call): Use
28321 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
28322 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
28325 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
28327 * doc/install.texi (Options specification): Restore entry of
28328 --enable-sjlj-exceptions.
28330 2017-05-27 Michael Eager <eager@eagercon.com>
28333 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
28335 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
28337 * config/microblaze/microblaze.h
28338 (FIXED_REGISTERS): Update in macro.
28339 (CALL_USED_REGISTERS): Update in macro.
28341 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
28343 * doc/install.texi: Add links to macOS binary distributions.
28345 2017-05-27 Jakub Jelinek <jakub@redhat.com>
28349 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28351 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28353 2017-05-26 Martin Liska <mliska@suse.cz>
28355 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
28357 2017-05-26 Martin Liska <mliska@suse.cz>
28359 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
28360 always leading ';; '.
28361 (dump_bb_info): Likewise.
28362 (brief_dump_cfg): Likewise.
28363 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
28364 * dumpfile.c: Remove usage of TDF_VERBOSE.
28365 * dumpfile.h (enum dump_kind): Likewise.
28366 (dump_gimple_bb_header): Do not use TDF_COMMENT.
28367 * print-tree.c (debug_verbose): Remove.
28368 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
28369 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
28370 * tree-diagnostic.c (default_tree_printer): Replace
28371 TDF_DIAGNOSTIC with TDF_SLIM.
28373 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28375 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
28376 in parameter loop, rather than loop_vinfo.
28377 (create_intersect_range_checks): Ditto.
28378 (vect_create_cond_for_alias_checks): Update call to above functions.
28380 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28382 PR tree-optimization/80815
28383 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
28384 for merging runtime alias checks. Handle negative DR_STEPs.
28386 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28388 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
28390 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
28391 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
28392 out code pruning runtime alias checks.
28393 * tree-data-ref.c (prune_runtime_alias_test_list): New function
28394 factored out from above.
28395 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28397 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
28399 (prune_runtime_alias_test_list): New decalaration.
28401 2017-05-26 Bin Cheng <bin.cheng@arm.com>
28403 * tree-vect-data-refs.c (compare_tree): Rename and move ...
28404 * tree-data-ref.c (data_ref_compare_tree): ... to here.
28405 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
28406 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
28407 (operator==, comp_dr_with_seg_len_pair): Ditto.
28408 (vect_prune_runtime_alias_test_list): Ditto.
28410 2017-05-26 Martin Liska <mliska@suse.cz>
28413 * params.def: Bound partial-inlining-entry-probability param.
28415 2017-05-26 Marek Polacek <polacek@redhat.com>
28418 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
28421 2017-05-26 Richard Biener <rguenther@suse.de>
28423 PR tree-optimization/80842
28424 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
28427 2017-05-26 Richard Biener <rguenther@suse.de>
28429 PR tree-optimization/80844
28430 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
28432 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
28434 * doc/md.texi (Machine Constraints): Update x86 family
28435 machine constraints section to match 'config/i386/constraints.md'.
28437 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
28439 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
28441 2017-05-25 Nathan Sidwell <nathan@acm.org>
28443 * doc/invoke.texi (--enable-languages): Update documentation.
28445 2017-05-25 Martin Liska <mliska@suse.cz>
28447 * dumpfile.c: Add TDF_FOLDING.
28448 * dumpfile.h (enum dump_kind): Likewise.
28449 * genmatch.c (dt_simplify::gen_1): Use it.
28451 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28453 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
28455 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28457 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
28458 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
28460 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
28462 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
28463 * match.pd (X == C): Rewrite it here.
28464 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
28465 with_certain_nonzero_bits2): New predicates.
28466 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
28468 2017-05-24 Nathan Sidwell <nathan@acm.org>
28470 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
28473 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
28476 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28478 * config/powerpcspe: New port. Files are copied from the rs6000
28479 port, with "rs6000" in filenames replaced by "powerpcspe".
28481 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
28483 PR rtl-optimization/80754
28484 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
28486 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
28488 * config/sparc/sparc.md (length): Return the correct value for -mflat
28489 sibcalls to match output_sibcall.
28491 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
28495 * config/rs6000/rs6000.c (struct machine_function): Add new field
28497 (rs6000_get_separate_components): Init that field, use it.
28498 (rs6000_components_for_bb): Use the field.
28500 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
28502 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
28504 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
28506 PR middle-end/80823
28507 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
28509 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28512 * config/s390/s390.c (s390_check_qrst_address): Check incoming
28513 address against address_operand predicate.
28514 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
28516 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
28518 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
28519 parameters passed indirectly.
28521 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
28523 * config/i386/i386.md (*movdi_internal): Remove SSE4
28524 alternative 18 (?r, *v). Update insn attributes.
28525 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
28526 Update insn attributes.
28527 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
28528 Update insn attributes.
28529 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
28530 alternative 1 (r, v). Remove isa attribute.
28531 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
28532 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
28533 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
28535 2017-05-23 Tom de Vries <tom@codesourcery.com>
28537 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
28540 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28542 * cgraphunit.c (symbol_table::process_new_functions): Update.
28543 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
28544 (inline_generate_summary): Rename to ...
28545 (ipa_fn_summary_generate): ... this one.
28546 (inline_read_summary): Rename to ...
28547 (ipa_fn_summary_read): ... this one.
28548 (inline_write_summary): Rename to ...
28549 (ipa_fn_summary_write): ... this one.
28550 (inline_free_summary): Rename to ...
28551 (ipa_free_fn_summary): ... this one.
28552 (pass_data_local_fn_summary, pass_local_fn_summary,
28553 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
28554 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
28555 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
28556 make_pass_ipa_fn_summary): New.
28557 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
28558 inline_write_summary, inline_free_summary): Remove.
28559 (ipa_free_fn_summary) : New.
28560 * ipa-inline.c (ipa_inline): Update.
28561 (pass_ipa_inline): Do not generate summaries.
28562 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
28564 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
28565 and add pass_ipa_fn_summary.
28566 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
28568 (make_pass_inline_parameters): Remove.
28570 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
28572 * omp-low.c (struct omp_context): Remove "default_kind" member.
28575 * omp-offload.c (execute_oacc_device_lower): Remove the
28576 parallelism dimensions function attributes for unparallelized
28577 OpenACC kernels constructs.
28579 2017-05-23 Martin Liska <mliska@suse.cz>
28581 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
28583 (cgraph_edge::make_speculative): Likewise.
28584 (cgraph_edge::resolve_speculation): Likewise.
28585 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
28586 (cgraph_node::dump): Likewise.
28587 * cgraph.h: Likewise.
28588 * cgraphunit.c (analyze_functions): Likewise.
28589 (symbol_table::compile): Likewise.
28590 * ipa-cp.c (print_all_lattices): Likewise.
28591 (determine_versionability): Likewise.
28592 (initialize_node_lattices): Likewise.
28593 (ipcp_verify_propagated_values): Likewise.
28594 (estimate_local_effects): Likewise.
28595 (update_profiling_info): Likewise.
28596 (create_specialized_node): Likewise.
28597 (perhaps_add_new_callers): Likewise.
28598 (decide_about_value): Likewise.
28599 (decide_whether_version_node): Likewise.
28600 (identify_dead_nodes): Likewise.
28601 (ipcp_store_bits_results): Likewise.
28602 * ipa-devirt.c (dump_targets): Likewise.
28603 (ipa_devirt): Likewise.
28604 * ipa-icf.c (sem_item::dump): Likewise.
28605 (sem_function::equals): Likewise.
28606 (sem_variable::equals): Likewise.
28607 (sem_item_optimizer::read_section): Likewise.
28608 (sem_item_optimizer::execute): Likewise.
28609 (congruence_class::dump): Likewise.
28610 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
28611 (dump_inline_summary): Likewise.
28612 (estimate_node_size_and_time): Likewise.
28613 (inline_analyze_function): Likewise.
28614 * ipa-inline-transform.c (inline_call): Likewise.
28615 * ipa-inline.c (report_inline_failed_reason): Likewise.
28616 (want_early_inline_function_p): Likewise.
28617 (edge_badness): Likewise.
28618 (update_edge_key): Likewise.
28619 (inline_small_functions): Likewise.
28620 * ipa-profile.c (ipa_profile): Likewise.
28621 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
28622 (ipa_make_edge_direct_to_target): Likewise.
28623 (remove_described_reference): Likewise.
28624 (ipa_impossible_devirt_target): Likewise.
28625 (propagate_controlled_uses): Likewise.
28626 (ipa_print_node_params): Likewise.
28627 (ipcp_transform_function): Likewise.
28628 * ipa-pure-const.c (pure_const_read_summary): Likewise.
28629 (propagate_pure_const): Likewise.
28630 * ipa-reference.c (generate_summary): Likewise.
28631 (read_write_all_from_decl): Likewise.
28632 (propagate): Likewise.
28633 (ipa_reference_read_optimization_summary): Likewise.
28634 * ipa-utils.c (ipa_merge_profiles): Likewise.
28635 * ipa.c (walk_polymorphic_call_targets): Likewise.
28636 (symbol_table::remove_unreachable_nodes): Likewise.
28637 (ipa_single_use): Likewise.
28638 * passes.c (execute_todo): Likewise.
28639 * predict.c (drop_profile): Likewise.
28640 * symtab.c (symtab_node::get_dump_name): New function.
28641 (symtab_node::dump_name): Likewise.
28642 (symtab_node::dump_asm_name): Likewise.
28643 (symtab_node::dump_references): Likewise.
28644 (symtab_node::dump_referring): Likewise.
28645 (symtab_node::dump_base): Likewise.
28646 (symtab_node::debug_symtab): Likewise.
28647 * tree-sra.c (convert_callers_for_node): Likewise.
28648 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
28649 * value-prof.c (init_node_map): Likewise.
28651 2017-05-23 Martin Liska <mliska@suse.cz>
28653 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
28654 and symtab_node::debug_symtab to symbol_table::debug.
28655 * cgraphunit.c (analyze_functions): Use the renamed function.
28656 (symbol_table::compile): Likewise.
28657 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
28658 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
28659 * passes.c (execute_todo): Likewise.
28660 * symtab.c (symbol_table::dump): New function.
28661 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
28663 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28665 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
28666 that nonconst implies exec.
28668 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
28670 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
28671 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
28672 (inline_edge_summary_vec): Turn into ...
28673 (ipa_call_summaries): ... this one.
28674 (redirect_to_unreachable, edge_set_predicate,
28675 evaluate_properties_for_edge, inline_summary_alloc,
28676 reset_ipa_call_summary, reset_inline_summary,
28677 inline_summary_t::duplicate): Update.
28678 (inline_edge_duplication_hook): Turn to ...
28679 (ipa_call_summary_t::duplicate): ... this one.
28680 (inline_edge_removal_hook): Turn to ...
28681 (ipa_call_summary_t::remove): ... this one.
28682 (dump_inline_edge_summary): Turn to ...
28683 (dump_ipa_call_summary): ... this one.
28684 (estimate_function_body_sizes): Update.
28685 (inline_update_callee_summaries): Update.
28686 (remap_edge_change_prob): Update.
28687 (remap_edge_summaries): Update.
28688 (inline_merge_summary): Update.
28689 (do_estimate_edge_time): Update.
28690 (inline_generate_summary): Update.
28691 (inline_read_section): Update.
28692 (inline_read_summary): Update.
28693 (inline_free_summary): Update.
28694 * ipa-inline.c (can_inline_edge_p): Update.
28695 (compute_inlined_call_time): Update.
28696 (want_inline_small_function_p): Update.
28697 (edge_badness): Update.
28698 (early_inliner): Update.
28699 * ipa-inline.h (inline_edge_summary): Turn to ...
28700 (ipa_call_summary): ... this one.
28701 (ipa_call_summary_t): New class.
28702 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
28703 (ipa_call_summaries): New.
28704 (inline_edge_summary): Remove.
28705 (estimate_edge_growth): Update.
28706 * ipa-profile.c (ipa_propagate_frequency_1): Update.
28707 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28708 * ipa-split.c (execute_split_functions): Update.
28709 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
28711 2017-05-23 Tom de Vries <tom@codesourcery.com>
28713 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
28714 attributes): Document rdrand effective target.
28716 2017-05-23 Tom de Vries <tom@codesourcery.com>
28718 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
28719 attributes): Sort alphabetically.
28721 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
28723 * config/avr/genmultilib.awk: Use gsub instead of gensub.
28725 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
28728 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
28729 V2DF/V2DI splat into two separate patterns, one that handles
28730 registers, and the other that only handles memory. Drop support
28731 for splatting from a GPR on ISA 2.07 and then splitting the
28732 splat into direct move and splat.
28733 (vsx_splat_<mode>_reg): Likewise.
28734 (vsx_splat_<mode>_mem): Likewise.
28736 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
28738 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
28740 2017-05-22 Jakub Jelinek <jakub@redhat.com>
28742 PR middle-end/80809
28743 * omp-low.c (finish_taskreg_remap): New function.
28744 (finish_taskreg_scan): If unit size of ctx->record_type
28745 is non-constant, unshare the size expression and replace
28746 decls in it with possible outer var refs.
28748 PR middle-end/80809
28749 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
28750 GOVD_SHARED rather than GOVD_PRIVATE with it.
28751 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
28752 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
28754 PR middle-end/80853
28755 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
28756 as last argument to build_outer_var_ref for pointer bases of array
28757 section reductions.
28759 2017-05-19 Martin Sebor <msebor@redhat.com>
28761 * print-tree.c (print_node): Print DECL_READ_P flag.
28763 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28765 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
28766 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
28767 * cgraph.c: Likewise.
28768 * cgraphunit.c: Likewise.
28769 * gengtype.c: Likewise.
28770 * ipa-cp.c: Likewise.
28771 * ipa-devirt.c: Likewise.
28772 * ipa-icf.c: Likewise.
28773 * ipa-predicate.c: Likewise.
28774 * ipa-profile.c: Likewise.
28775 * ipa-prop.c: Likewise.
28776 * ipa-split.c: Likewise.
28778 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
28779 edge_predicate_pool, dump_inline_hints,
28780 inline_summary::account_size_time, redirect_to_unreachable,
28781 edge_set_predicate, set_hint_predicate,
28782 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
28783 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
28784 inline_summary_t::remove, remap_hint_predicate_after_duplication,
28785 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
28786 ipa_call_summary_t::remove, initialize_growth_caches,
28787 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
28788 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
28789 mark_modified, unmodified_parm_1, unmodified_parm,
28790 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
28791 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
28792 compute_bb_predicates, will_be_nonconstant_expr_predicate,
28793 will_be_nonconstant_predicate, record_modified_bb_info,
28794 get_minimal_bb, record_modified, param_change_prob,
28795 phi_result_unknown_predicate, predicate_for_phi_result,
28796 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
28797 estimate_function_body_sizes, compute_inline_parameters,
28798 compute_inline_parameters_for_curren, pass_data_inline_parameters,
28799 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
28800 inline_update_callee_summaries, remap_edge_change_prob,
28801 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
28802 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
28803 inline_analyze_function, inline_summary_t::insert,
28804 inline_generate_summary, read_ipa_call_summary, inline_read_section,
28805 inline_read_summary, write_ipa_call_summary, inline_write_summary,
28806 inline_free_summary): Move to ipa-fnsummary.h
28807 (predicate_t): Remove.
28808 * ipa-fnsummary.c: New file.
28809 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
28810 (enum inline_hints_vals, inline_hints, agg_position_info,
28811 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
28812 inline_summaries, ipa_call_summary, ipa_call_summary_t,
28813 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
28814 dump_inline_summary, dump_inline_hints, inline_generate_summary,
28815 inline_read_summary, inline_write_summary, inline_free_summary,
28816 inline_analyze_function, initialize_inline_failed,
28817 inline_merge_summary, inline_update_overall_summary,
28818 compute_inline_parameters): Move to ipa-fnsummary.h
28819 * ipa-fnsummary.h: New file.
28820 * ipa-inline-transform.h: Include ipa-inline.h.
28821 * ipa-inline.c: LIkewise.
28823 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28825 * ipa-inline.c (edge_badness): Use inlined_time instead of
28826 inline_summaries->get.
28828 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28830 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
28832 2017-05-22 Nathan Sidwell <nathan@acm.org>
28834 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
28835 (fdump-lang): Document 'raw' option.
28836 * dumpfile.h (TDI_tu): Delete.
28837 * dumpfile.c (dump_files): Remove translation-unit.
28838 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
28840 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
28842 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
28843 command option from $(AWK) call.
28844 * config/avr/genmultilib.awk: Simplify and rewrite so that it
28845 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
28846 [FORMAT]: Remove handling of variable.
28847 * config/avr/t-multilib: Regenerate.
28849 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28851 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
28853 (dump_inline_summary): Do not print self_time.
28854 (estimate_function_body_sizes): Do not set self_time.
28855 (compute_inline_parameters): Likewise.
28856 (inline_read_section, inline_write_summary): Do not stream self_time.
28857 * ipa-inline.h (inline_summary): Drop self_time.
28859 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28861 * ipa-inline-analysis.c (account_size_time): Rename to ...
28862 (inline_summary::account_size_time): ... this one.
28863 (reset_ipa_call_summary): Turn to ...
28864 (ipa_call_summary::reset): ... this one.
28865 (reset_inline_summary): Turn to ...
28866 (inline_summary::reset): ... this one.
28867 (inline_summary_t::remove): Update.
28868 (inline_summary_t::duplicate): Update.
28869 (ipa_call_summary_t::remove): Update.
28870 (dump_inline_summary): Update.
28871 (estimate_function_body_sizes): Update.
28872 (compute_inline_parameters): Update.
28873 (estimate_node_size_and_time): Update.
28874 (inline_merge_summary): Update.
28875 (inline_update_overall_summary): Update.
28876 (inline_read_section): Update.
28877 (inline_write_summary): Update.
28878 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
28879 add account_size_time and reset member functions.
28880 (ipa_call_summary): Add reset function.
28881 * ipa-predicate.h (predicate::operator &): Constify.
28883 2017-05-22 Richard Biener <rguenther@suse.de>
28885 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
28887 2017-05-19 Jason Merrill <jason@redhat.com>
28889 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
28891 2017-05-19 Marek Polacek <polacek@redhat.com>
28894 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
28895 TYPE_OVERFLOW_WRAPS checks.
28897 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
28899 * tree-core.h (enum omp_clause_default_kind): Add
28900 "OMP_CLAUSE_DEFAULT_PRESENT".
28901 * tree-pretty-print.c (dump_omp_clause): Handle it.
28902 * gimplify.c (enum gimplify_omp_var_data): Add
28903 "GOVD_MAP_FORCE_PRESENT".
28904 (gimplify_adjust_omp_clauses_1): Map it to
28905 "GOMP_MAP_FORCE_PRESENT".
28906 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
28908 * gimplify.c (oacc_default_clause): Clarify.
28910 2017-05-19 Nathan Sidwell <nathan@acm.org>
28912 LANG_HOOK_REGISTER_DUMPS
28913 * toplev.c (general_init): Call register dump lang hook.
28914 * doc/invoke.texi: Document -fdump-lang option family.
28915 * dumpfile.c (dump_files): Remove class dump here.
28916 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
28917 * dumpfile.h (tree_dump_index): Remove TDI_class.
28918 * langhooks-def.h (lhd_register_dumps): Declare.
28919 (LANG_HOOKS_REGISTER_DUMPS): Define.
28920 (LANG_HOOKS_INITIALIZER): Add it.
28921 * langhooks.c (lhd_register_dumps): Define.
28922 * langhooks.h (struct lang_hooks): Add register_dumps.
28924 2017-05-19 Nathan Sidwell <nathan@acm.org>
28926 * context.h (context::set_passes): New.
28927 * context.c (context::context): Do not create pass manager.
28928 * toplev.c (general_init): Create pass manager here.
28930 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
28932 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
28933 use this splitter if two add or or instructions would also work for
28934 the constant we want to generate.
28936 2017-05-19 Richard Biener <rguenther@suse.de>
28939 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
28940 predicate evaluation.
28942 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
28944 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
28946 * ipa-inline.c (want_inline_small_function_p): Do not cast to
28949 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
28951 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
28952 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
28953 (inline_edge_summary_vec): Turn into ...
28954 (ipa_call_summaries): ... this one.
28955 (redirect_to_unreachable, edge_set_predicate,
28956 evaluate_properties_for_edge, inline_summary_alloc,
28957 reset_ipa_call_summary, reset_inline_summary,
28958 inline_summary_t::duplicate): Update.
28959 (inline_edge_duplication_hook): Turn to ...
28960 (ipa_call_summary_t::duplicate): ... this one.
28961 (inline_edge_removal_hook): Turn to ...
28962 (ipa_call_summary_t::remove): ... this one.
28963 (dump_inline_edge_summary): Turn to ...
28964 (dump_ipa_call_summary): ... this one.
28965 (estimate_function_body_sizes): Update.
28966 (inline_update_callee_summaries): Update.
28967 (remap_edge_change_prob): Update.
28968 (remap_edge_summaries): Update.
28969 (inline_merge_summary): Update.
28970 (do_estimate_edge_time): Update.
28971 (inline_generate_summary): Update.
28972 (inline_read_section): Update.
28973 (inline_read_summary): Update.
28974 (inline_free_summary): Update.
28975 * ipa-inline.c (can_inline_edge_p): Update.
28976 (compute_inlined_call_time): Update.
28977 (want_inline_small_function_p): Update.
28978 (edge_badness): Update.
28979 (early_inliner): Update.
28980 * ipa-inline.h (inline_edge_summary): Turn to ...
28981 (ipa_call_summary): ... this one.
28982 (ipa_call_summary_t): New class.
28983 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
28984 (ipa_call_summaries): New.
28985 (inline_edge_summary): Remove.
28986 (estimate_edge_growth): Update.
28987 * ipa-profile.c (ipa_propagate_frequency_1): Update.
28988 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28989 * ipa-split.c (execute_split_functions): Update.
28990 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
28992 2017-05-19 Richard Biener <rguenther@suse.de>
28994 PR middle-end/80764
28995 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
28997 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
28999 * config/rs6000/rs6000.c (struct machine_function): Add field
29000 fpr_is_wrapped_separately.
29001 (rs6000_get_separate_components): Use 64 components. Handle the
29002 new FPR components.
29003 (rs6000_components_for_bb): Handle the FPR components.
29004 (rs6000_emit_prologue_components): Handle the FPR components.
29005 (rs6000_emit_epilogue_components): Handle the FPR components.
29006 (rs6000_set_handled_components): Handle the FPR components.
29007 (rs6000_emit_prologue): Don't output prologue code for those FPRs
29008 that are already separately shrink-wrapped.
29009 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
29010 that are already separately shrink-wrapped.
29012 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
29015 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
29018 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
29019 (define_peephole2 for Altivec d-form load): Add peepholes to catch
29020 cases where the register allocator uses a move and an offsettable
29021 memory operation to/from a FPR register on ISA 2.06/2.07.
29022 (define_peephole2 for Altivec d-form store): Likewise.
29024 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
29027 * config/i386/mmx.md (*mov<mode>_internal): Enable
29028 alternatives 11, 12, 13 and 14 also for 32bit targets.
29029 Remove alternatives 15, 16, 17 and 18.
29030 * config/i386/sse.md (vec_concatv2di): Change
29031 alternative (!x, *y) to (x, ?!*Yn).
29033 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
29035 * dumpfile.h (enum dump_kind): Remove stray comma.
29037 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29039 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
29040 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
29041 predicate::num_conditions
29042 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
29043 (CHANGED): turn into predicate::changed.
29044 (agg_position_info): Move to ipa-predicate.h
29045 (add_condition, predicate::add_clause, predicate::operator &=,
29046 predicate::or_with, predicate::evaluate, predicate::probability,
29047 dump_condition, dump_clause, predicate::dump,
29048 predicate::remap_after_duplication, predicate::remap_after_inlining,
29049 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
29050 (evaluate_conditions_for_known_args): Update.
29051 (set_cond_stmt_execution_predicate): Update.
29052 * ipa-inline.h: Include ipa-predicate.h
29053 (condition, inline_param_summary, conditions, agg_position_info,
29054 predicate): Move to ipa-predicate.h
29055 * ipa-predicate.c: New file.
29056 * ipa-predicate.h: New file.
29058 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
29060 * final.c (leaf_function_p): Check we are not in a sequence.
29062 2017-05-18 Martin Liska <mliska@suse.cz>
29064 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
29065 * dumpfile.c (dump_register): Use new enum dump_kind.
29066 (get_dump_file_name): Likewise.
29067 (dump_enable_all): Likewise.
29068 (dump_switch_p_1): Likewise.
29069 (enable_rtl_dump_file): Remove usage of TDF_RTL.
29070 * dumpfile.h (enum dump_kind): New enum type.
29071 (struct dump_file_info): Create constructor and
29072 format fields and comments.
29073 * passes.c (pass_manager::register_one_dump_file):
29075 * statistics.c (statistics_early_init): Likewise.
29076 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
29077 TDF_TREE with TDF_SLIM.
29078 (gather_memory_references_ref): Likewise.
29080 2017-05-18 Martin Liska <mliska@suse.cz>
29082 * vec.h (struct vnull): Use it.
29084 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
29086 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
29087 (true_predicate, false_predicate, true_predicate_p,
29088 false_predicate_p): Remove.
29089 (single_cond_predicate, not_inlined_predicate): Turn to member function
29091 (add_condition): Update.
29092 (add_clause): Turn to...
29093 (predicate::add_clause): ... this one; update; allow passing NULL
29095 (and_predicates): Turn to ...
29096 (predicate::operator &=): ... this one.
29097 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
29098 (or_predicates): Turn to ...
29099 (predicate::or_with): ... this one.
29100 (evaluate_predicate): Turn to ...
29101 (predicate::evaluate): ... this one.
29102 (predicate_probability): Turn to ...
29103 (predicate::probability): ... this one.
29104 (dump_condition): Update.
29105 (dump_predicate): Turn to ...
29106 (predicate::dump): ... this one.
29107 (account_size_time): Update.
29108 (edge_set_predicate): Update.
29109 (set_hint_predicate): UPdate.
29110 (evaluate_conditions_for_known_args): Update.
29111 (evaluate_properties_for_edge): Update.
29112 (remap_predicate_after_duplication): Turn to...
29113 (predicate::remap_after_duplication): ... this one.
29114 (remap_hint_predicate_after_duplication): Update.
29115 (inline_summary_t::duplicate): UPdate.
29116 (dump_inline_edge_summary): Update.
29117 (dump_inline_summary): Update.
29118 (set_cond_stmt_execution_predicate): Update.
29119 (set_switch_stmt_execution_predicate): Update.
29120 (compute_bb_predicates): Update.
29121 (will_be_nonconstant_expr_predicate): Update.
29122 (will_be_nonconstant_predicate): Update.
29123 (phi_result_unknown_predicate): Update.
29124 (predicate_for_phi_result): Update.
29125 (array_index_predicate): Update.
29126 (estimate_function_body_sizes): Update.
29127 (estimate_node_size_and_time): Update.
29128 (estimate_ipcp_clone_size_and_time): Update.
29129 (remap_predicate): Rename to ...
29130 (predicate::remap_after_inlining): ... this one.
29131 (remap_hint_predicate): Update.
29132 (inline_merge_summary): Update.
29133 (inline_update_overall_summary): Update.
29134 (estimate_size_after_inlining): Update.
29135 (read_predicate): Rename to ...
29136 (predicate::stream_in): ... this one.
29137 (read_inline_edge_summary): Update.
29138 (write_predicate): Rename to ...
29139 (predicate::stream_out): ... this one.
29140 (write_inline_edge_summary): Update.
29141 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
29142 (clause_t): Turn to uint32_t
29143 (predicate): Turn to class; implement constructor and operators
29145 (size_time_entry): Update.
29146 (inline_summary): Update.
29147 (inline_edge_summary): Update.
29149 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
29151 * fold-const.c (fold_binary_loc): Move transformation...
29152 * match.pd (C - X CMP X): ... here.
29154 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
29156 * config/sparc/sparc.c (sparc_option_override): Set function
29157 alignment for -mcpu=niagara7 to 64 to match the I$ line.
29158 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
29160 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
29162 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
29164 2017-05-18 Marek Polacek <polacek@redhat.com>
29167 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
29168 (pass_ubsan::execute): Call gimple_assign_single_p instead of
29169 gimple_assign_load_p.
29171 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
29173 PR middle-end/80692
29174 * real.c (do_compare): Give decimal_do_compare preference over
29175 comparing just the signs.
29177 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
29179 * doc/md.texi (Canonicalization of Instructions): Describe the
29180 canonical form of instructions that inherently set a condition
29183 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
29185 PR middle-end/80775
29186 * tree-cfg.c: Move deletion of unreachable case statements to after
29187 the merging of consecutive case labels.
29189 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
29191 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
29192 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
29193 restoring of callee-saved registers.
29195 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
29197 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
29198 * config/visium/visium.c (single_set_and_flags): Likewise.
29199 * config/visium/visium.md (Substitutions): Likewise.
29201 2017-05-17 Martin Liska <mliska@suse.cz>
29203 * cfg.c: Introduce dump_flags_t type and
29204 use it instead of int type.
29206 * cfghooks.c: Likewise.
29207 * cfghooks.h (struct cfg_hooks): Likewise.
29208 * cfgrtl.c: Likewise.
29209 * cfgrtl.h: Likewise.
29210 * cgraph.c (cgraph_node::get_body): Likewise.
29211 * coretypes.h: Likewise.
29212 * domwalk.c: Likewise.
29213 * domwalk.h: Likewise.
29214 * dumpfile.c (struct dump_option_value_info): Likewise.
29215 (dump_enable_all): Likewise.
29216 (dump_switch_p_1): Likewise.
29217 (opt_info_switch_p): Likewise.
29218 * dumpfile.h (enum tree_dump_index): Likewise.
29219 (struct dump_file_info): Likewise.
29220 * genemit.c: Likewise.
29221 * generic-match-head.c: Likewise.
29222 * gengtype.c (open_base_files): Likewise.
29223 * gimple-pretty-print.c: Likewise.
29224 * gimple-pretty-print.h: Likewise.
29225 * graph.c (print_graph_cfg): Likewise.
29226 * graphite-scop-detection.c (dot_all_sese): Likewise.
29227 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
29228 * loop-unroll.c (report_unroll): Likewise.
29229 * passes.c (pass_manager::register_one_dump_file): Likewise.
29230 * print-tree.c: Likewise.
29231 * statistics.c: Likewise.
29232 * tree-cfg.c: Likewise.
29233 * tree-cfg.h: Likewise.
29234 * tree-dfa.c: Likewise.
29235 * tree-dfa.h: Likewise.
29236 * tree-dump.c (dump_function): Likewise.
29237 * tree-dump.h (struct dump_info): Likewise.
29238 * tree-pretty-print.c: Likewise.
29239 * tree-pretty-print.h: Likewise.
29240 * tree-ssa-live.c: Likewise.
29241 * tree-ssa-live.h: Likewise.
29242 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
29243 * tree-vect-loop.c: Likewise.
29244 * tree-vect-slp.c: Likewise.
29246 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
29247 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29249 PR tree-optimization/80457
29250 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
29251 of all arguments to a statement as scalar_to_vec operations.
29252 (vectorizable_call): Adjust call to vect_model_simple_cost for
29254 (vectorizable_conversion): Likewise.
29255 (vectorizable_assignment): Likewise.
29256 (vectorizable_shift): Likewise.
29257 (vectorizable_operation): Likewise.
29258 (vectorizable_comparison): Likewise.
29259 (vect_is_simple_cond): Record the def types for operands.
29260 (vectorizable_condition): Likewise, call vect_model_simple_cost.
29261 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
29262 for statement argument count.
29264 2017-05-16 Carl Love <cel@us.ibm.com>
29266 * config/rs6000/rs6000-c: Add support for built-in functions
29267 vector unsigned long long vec_bperm (vector unsigned long long,
29268 vector unsigned char)
29269 vector signed long long vec_mule (vector signed int,
29271 vector unsigned long long vec_mule (vector unsigned int,
29272 vector unsigned int)
29273 vector signed long long vec_mulo (vector signed int,
29275 vector unsigned long long vec_mulo (vector unsigned int,
29276 vector unsigned int)
29277 vector signed char vec_sldw (vector signed char,
29278 vector signed char,
29280 vector unsigned char vec_sldw (vector unsigned char,
29281 vector unsigned char,
29283 vector signed short vec_sldw (vector signed short,
29284 vector signed short,
29286 vector unsigned short vec_sldw (vector unsigned short,
29287 vector unsigned short,
29289 vector signed int vec_sldw (vector signed int,
29292 vector unsigned int vec_sldw (vector unsigned int,
29293 vector unsigned int,
29295 vector signed long long vec_sldw (vector signed long long,
29296 vector signed long long,
29298 vector unsigned long long vec_sldw (vector unsigned long long,
29299 vector unsigned long long,
29301 * config/rs6000/rs6000-c: Add support for built-in functions
29302 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
29303 * config/rs6000/altivec.h: Add defintion for vec_sldw.
29304 * doc/extend.texi: Update the built-in documentation for the
29305 new built-in functions.
29307 2017-05-16 Marek Polacek <polacek@redhat.com>
29311 * tree.c (save_expr): Don't fold the expression.
29313 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
29315 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
29316 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
29317 and (?*y,m). Update insn attributes.
29319 2017-05-16 Martin Liska <mliska@suse.cz>
29321 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
29322 flags argument of print_gimple_stmt, print_gimple_expr,
29323 print_generic_stmt and print_generic_expr.
29324 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
29325 * coretypes.h: Likewise.
29326 * except.c (dump_eh_tree): Likewise.
29327 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
29328 * gimple-pretty-print.h: Likewise.
29329 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
29330 (backprop::push_to_worklist): Likewise.
29331 (backprop::pop_from_worklist): Likewise.
29332 (backprop::process_use): Likewise.
29333 (backprop::intersect_uses): Likewise.
29334 (note_replacement): Likewise.
29335 * gimple-ssa-store-merging.c
29336 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
29337 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
29338 (pass_store_merging::execute): Likewise.
29339 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
29340 (ssa_base_cand_dump_callback): Likewise.
29341 (dump_incr_vec): Likewise.
29342 (replace_refs): Likewise.
29343 (replace_mult_candidate): Likewise.
29344 (create_add_on_incoming_edge): Likewise.
29345 (create_phi_basis): Likewise.
29346 (insert_initializers): Likewise.
29347 (all_phi_incrs_profitable): Likewise.
29348 (introduce_cast_before_cand): Likewise.
29349 (replace_one_candidate): Likewise.
29350 * gimplify.c (gimplify_expr): Likewise.
29351 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
29352 (set_rename): Likewise.
29353 (rename_uses): Likewise.
29354 (copy_loop_phi_nodes): Likewise.
29355 (add_close_phis_to_merge_points): Likewise.
29356 (copy_loop_close_phi_args): Likewise.
29357 (copy_cond_phi_args): Likewise.
29358 (graphite_copy_stmts_from_block): Likewise.
29359 (translate_pending_phi_nodes): Likewise.
29360 * graphite-poly.c (print_pdr): Likewise.
29361 (dump_gbb_cases): Likewise.
29362 (dump_gbb_conditions): Likewise.
29363 (print_scop_params): Likewise.
29364 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
29365 (build_cross_bb_scalars_use): Likewise.
29366 (gather_bbs::before_dom_children): Likewise.
29367 * hsa-dump.c (dump_hsa_immed): Likewise.
29368 * ipa-cp.c (print_ipcp_constant_value): Likewise.
29369 (get_replacement_map): Likewise.
29370 * ipa-inline-analysis.c (dump_condition): Likewise.
29371 (estimate_function_body_sizes): Likewise.
29372 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
29373 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
29374 * ipa-prop.c (ipa_dump_param): Likewise.
29375 (ipa_print_node_jump_functions_for_edge): Likewise.
29376 (ipa_modify_call_arguments): Likewise.
29377 (ipa_modify_expr): Likewise.
29378 (ipa_dump_param_adjustments): Likewise.
29379 (ipa_dump_agg_replacement_values): Likewise.
29380 (ipcp_modif_dom_walker::before_dom_children): Likewise.
29381 * ipa-pure-const.c (check_stmt): Likewise.
29382 (pass_nothrow::execute): Likewise.
29383 * ipa-split.c (execute_split_functions): Likewise.
29384 * omp-offload.c (dump_oacc_loop_part): Likewise.
29385 (dump_oacc_loop): Likewise.
29386 * trans-mem.c (tm_log_emit): Likewise.
29387 (tm_memopt_accumulate_memops): Likewise.
29388 (dump_tm_memopt_set): Likewise.
29389 (dump_tm_memopt_transform): Likewise.
29390 * tree-cfg.c (gimple_verify_flow_info): Likewise.
29391 (print_loop): Likewise.
29392 * tree-chkp-opt.c (chkp_print_addr): Likewise.
29393 (chkp_gather_checks_info): Likewise.
29394 (chkp_get_check_result): Likewise.
29395 (chkp_remove_check_if_pass): Likewise.
29396 (chkp_use_outer_bounds_if_possible): Likewise.
29397 (chkp_reduce_bounds_lifetime): Likewise.
29398 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
29399 (chkp_mark_completed_bounds): Likewise.
29400 (chkp_register_incomplete_bounds): Likewise.
29401 (chkp_mark_invalid_bounds): Likewise.
29402 (chkp_maybe_copy_and_register_bounds): Likewise.
29403 (chkp_build_returned_bound): Likewise.
29404 (chkp_get_bound_for_parm): Likewise.
29405 (chkp_build_bndldx): Likewise.
29406 (chkp_get_bounds_by_definition): Likewise.
29407 (chkp_generate_extern_var_bounds): Likewise.
29408 (chkp_get_bounds_for_decl_addr): Likewise.
29409 * tree-chrec.c (chrec_apply): Likewise.
29410 * tree-data-ref.c (dump_data_reference): Likewise.
29411 (dump_subscript): Likewise.
29412 (dump_data_dependence_relation): Likewise.
29413 (analyze_overlapping_iterations): Likewise.
29414 * tree-inline.c (expand_call_inline): Likewise.
29415 (tree_function_versioning): Likewise.
29416 * tree-into-ssa.c (dump_defs_stack): Likewise.
29417 (dump_currdefs): Likewise.
29418 (dump_names_replaced_by): Likewise.
29419 (dump_update_ssa): Likewise.
29420 (update_ssa): Likewise.
29421 * tree-object-size.c (pass_object_sizes::execute): Likewise.
29422 * tree-parloops.c (build_new_reduction): Likewise.
29423 (try_create_reduction_list): Likewise.
29424 (ref_conflicts_with_region): Likewise.
29425 (oacc_entry_exit_ok_1): Likewise.
29426 (oacc_entry_exit_single_gang): Likewise.
29427 * tree-pretty-print.h: Likewise.
29428 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
29429 (get_scalar_evolution): Likewise.
29430 (add_to_evolution): Likewise.
29431 (get_loop_exit_condition): Likewise.
29432 (analyze_evolution_in_loop): Likewise.
29433 (analyze_initial_condition): Likewise.
29434 (analyze_scalar_evolution): Likewise.
29435 (instantiate_scev): Likewise.
29436 (number_of_latch_executions): Likewise.
29437 (gather_chrec_stats): Likewise.
29438 (final_value_replacement_loop): Likewise.
29439 (scev_const_prop): Likewise.
29440 * tree-sra.c (dump_access): Likewise.
29441 (disqualify_candidate): Likewise.
29442 (create_access): Likewise.
29443 (reject): Likewise.
29444 (maybe_add_sra_candidate): Likewise.
29445 (create_access_replacement): Likewise.
29446 (analyze_access_subtree): Likewise.
29447 (analyze_all_variable_accesses): Likewise.
29448 (sra_modify_assign): Likewise.
29449 (initialize_constant_pool_replacements): Likewise.
29450 (find_param_candidates): Likewise.
29451 (decide_one_param_reduction): Likewise.
29452 (replace_removed_params_ssa_names): Likewise.
29453 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
29454 * tree-ssa-copy.c (dump_copy_of): Likewise.
29455 (copy_prop_visit_cond_stmt): Likewise.
29456 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
29457 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
29458 (record_equivalences_from_stmt): Likewise.
29459 * tree-ssa-dse.c (compute_trims): Likewise.
29460 (delete_dead_call): Likewise.
29461 (delete_dead_assignment): Likewise.
29462 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
29463 (forward_propagate_into_cond): Likewise.
29464 (pass_forwprop::execute): Likewise.
29465 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
29466 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
29468 (move_computations_worker): Likewise.
29469 (execute_sm): Likewise.
29470 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
29471 (remove_exits_and_undefined_stmts): Likewise.
29472 (remove_redundant_iv_tests): Likewise.
29473 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
29474 (adjust_iv_update_pos): Likewise.
29475 * tree-ssa-math-opts.c (bswap_replace): Likewise.
29476 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
29477 (value_replacement): Likewise.
29478 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
29479 * tree-ssa-pre.c (print_pre_expr): Likewise.
29480 (get_representative_for): Likewise.
29481 (create_expression_by_pieces): Likewise.
29482 (insert_into_preds_of_block): Likewise.
29483 (eliminate_insert): Likewise.
29484 (eliminate_dom_walker::before_dom_children): Likewise.
29485 (eliminate): Likewise.
29486 (remove_dead_inserted_code): Likewise.
29487 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
29488 * tree-ssa-reassoc.c (get_rank): Likewise.
29489 (eliminate_duplicate_pair): Likewise.
29490 (eliminate_plus_minus_pair): Likewise.
29491 (eliminate_not_pairs): Likewise.
29492 (undistribute_ops_list): Likewise.
29493 (eliminate_redundant_comparison): Likewise.
29494 (update_range_test): Likewise.
29495 (optimize_range_tests_var_bound): Likewise.
29496 (optimize_vec_cond_expr): Likewise.
29497 (rewrite_expr_tree): Likewise.
29498 (rewrite_expr_tree_parallel): Likewise.
29499 (linearize_expr): Likewise.
29500 (break_up_subtract): Likewise.
29501 (linearize_expr_tree): Likewise.
29502 (attempt_builtin_powi): Likewise.
29503 (attempt_builtin_copysign): Likewise.
29504 (transform_stmt_to_copy): Likewise.
29505 (transform_stmt_to_multiply): Likewise.
29506 (dump_ops_vector): Likewise.
29507 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
29508 (print_scc): Likewise.
29509 (set_ssa_val_to): Likewise.
29510 (visit_reference_op_store): Likewise.
29511 (visit_use): Likewise.
29512 (sccvn_dom_walker::before_dom_children): Likewise.
29513 (run_scc_vn): Likewise.
29514 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
29516 (expr_hash_elt::print): Likewise.
29517 (const_and_copies::pop_to_marker): Likewise.
29518 (const_and_copies::record_const_or_copy_raw): Likewise.
29519 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
29520 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
29521 (dump_predicates): Likewise.
29522 (find_uninit_use): Likewise.
29523 (warn_uninitialized_phi): Likewise.
29524 (pass_late_warn_uninitialized::execute): Likewise.
29525 * tree-ssa.c (verify_vssa): Likewise.
29526 (verify_ssa): Likewise.
29527 (maybe_optimize_var): Likewise.
29528 * tree-vrp.c (dump_value_range): Likewise.
29529 (dump_all_value_ranges): Likewise.
29530 (dump_asserts_for): Likewise.
29531 (register_edge_assert_for_2): Likewise.
29532 (vrp_visit_cond_stmt): Likewise.
29533 (vrp_visit_switch_stmt): Likewise.
29534 (vrp_visit_stmt): Likewise.
29535 (vrp_visit_phi_node): Likewise.
29536 (simplify_cond_using_ranges_1): Likewise.
29537 (fold_predicate_in): Likewise.
29538 (evrp_dom_walker::before_dom_children): Likewise.
29539 (evrp_dom_walker::push_value_range): Likewise.
29540 (evrp_dom_walker::pop_value_range): Likewise.
29541 (execute_early_vrp): Likewise.
29543 2017-05-16 Richard Biener <rguenther@suse.de>
29545 * dwarf2out.c (loc_list_from_tree_1): Do not create
29546 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
29548 2017-05-16 Richard Biener <rguenther@suse.de>
29550 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
29552 (note_variable_value_in_expr): If we resolved the decl ref
29553 do not push to the stack.
29555 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
29557 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
29558 operations in fast-math mode.
29559 (vaddq_f16): Likewise.
29560 (vmul_f16): Likewise.
29561 (vmulq_f16): Likewise.
29562 (vsub_f16): Likewise.
29563 (vsubq_f16): Likewise.
29564 * config/arm/neon.md (add<mode>3): New.
29566 (fma:<VH:mode>3): New. Also remove outdated comment.
29569 2017-05-16 Martin Liska <mliska@suse.cz>
29573 * ipa-devirt.c (warn_types_mismatch): Fix typo.
29574 (odr_types_equivalent_p): Likewise.
29576 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
29578 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
29580 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
29583 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
29584 non-interunit SSE move alternatives with '?'.
29585 (zero-extendsidi peephole2): New peephole to skip intermediate
29586 general register in SSE zero-extend sequence.
29588 2017-05-15 Jeff Law <law@redhat.com>
29590 * reorg.c (relax_delay_slots): Create a new variable to hold
29591 the temporary target rather than clobbering TARGET_LABEL.
29593 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
29594 missing argument to extract_bit_field call.
29595 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
29597 2017-05-15 Martin Liska <mliska@suse.cz>
29600 * gcc.c (process_command): Do not allow empty argument of -o option.
29602 2017-05-15 Renlin Li <renlin.li@arm.com>
29604 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
29605 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
29606 * config/aarch64/constraints.md (Usf): Add long call check.
29607 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
29608 (call_value): Likewise.
29609 (sibcall): Likewise.
29610 (sibcall_value): Likewise.
29612 (call_value_insn): New.
29613 (sibcall_insn): Update rtx pattern.
29614 (sibcall_value_insn): Likewise.
29615 (call_internal): Remove.
29616 (call_value_internal): Likewise.
29617 (sibcall_internal): Likewise.
29618 (sibcall_value_internal): Likewise.
29619 (call_reg): Likewise.
29620 (call_symbol): Likewise.
29621 (call_value_reg): Likewise.
29622 (call_value_symbol): Likewise.
29624 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
29627 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
29629 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
29631 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
29632 compatible with CCGOCmode and with CCZmode.
29634 2017-05-14 Martin Sebor <msebor@redhat.com>
29636 PR middle-end/77671
29637 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
29638 (gimple_fold_builtin_snprintf): Same.
29639 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
29640 (gimple_fold_builtin_snprintf): Same.
29641 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
29642 of character types.
29643 (is_call_safe): New function.
29644 (try_substitute_return_value): Call it.
29645 (try_simplify_call): New function.
29646 (pass_sprintf_length::handle_gimple_call): Call it.
29648 2017-05-14 Martin Sebor <msebor@redhat.com>
29650 PR middle-end/80669
29651 * builtins.c (expand_builtin_stpncpy): Simplify.
29653 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
29655 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
29656 * config/i386/i386.h
29657 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
29658 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
29659 (struct machine_function): Add new members call_ms2sysv,
29660 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
29661 (struct machine_frame_state): New fields sp_realigned and
29662 sp_realigned_offset.
29663 * config/i386/i386.c
29664 (enum xlogue_stub): New enum.
29665 (enum xlogue_stub_sets): New enum.
29666 (class xlogue_layout): New class.
29667 (struct ix86_frame): New fields stack_realign_allocate_offset,
29668 stack_realign_offset and outlined_save_offset. Modify comments to
29669 detail stack layout when using out-of-line stubs.
29670 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
29671 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
29672 -mcall-ms2sysv-xlogues.
29673 (stub_managed_regs): New static variable.
29674 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
29675 registers managed by out-of-line stub.
29676 (disable_call_ms2sysv_xlogues): New function.
29677 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
29678 m->call_ms2sysv when appropriate and compute frame layout for
29680 (sp_valid_at, fp_valid_at): New inline functions.
29681 (choose_basereg): New function.
29682 (choose_baseaddr): Add align parameter, use choose_basereg and modify
29684 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
29685 Use align parameter of choose_baseaddr to generated aligned SSE movs
29687 (pro_epilogue_adjust_stack): Modify to track
29688 machine_frame_state::sp_realigned.
29689 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
29690 (ix86_nsaved_sseregs): Likewise.
29691 (ix86_emit_save_regs): Likewise.
29692 (ix86_emit_save_regs_using_mov): Likewise.
29693 (ix86_emit_save_sse_regs_using_mov): Likewise.
29694 (get_scratch_register_on_entry): Likewise.
29695 (gen_frame_set): New function.
29696 (gen_frame_load): Likewise.
29697 (gen_frame_store): Likewise.
29698 (emit_outlined_ms2sysv_save): Likewise.
29699 (emit_outlined_ms2sysv_restore): Likewise.
29700 (ix86_expand_prologue): Modify stack re-alignment code and call
29701 emit_outlined_ms2sysv_save when appropriate.
29702 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
29703 parameter rtx_insn *insn, which allows the function to be used to only
29704 generate the notes.
29705 (ix86_expand_epilogue): Modify validity checks of frame and stack
29706 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
29707 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
29708 * config/i386/predicates.md
29709 (save_multiple): New predicate.
29710 (restore_multiple): Likewise.
29711 * config/i386/sse.md
29712 (save_multiple<mode>): New pattern.
29713 (save_multiple_realign<mode>): Likewise.
29714 (restore_multiple<mode>): Likewise.
29715 (restore_multiple_and_return<mode>): Likewise.
29716 (restore_multiple_leave_return<mode>): Likewise.
29717 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
29719 2017-05-14 Julia Koval <julia.koval@intel.com>
29721 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
29722 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
29723 (__builtin_ia32_xsetbv): New builtins.
29724 * config/i386/i386.c (ix86_expand_special_args_builtin):
29726 (ix86_expand_builtin): Special expand for new intrinsics.
29727 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
29728 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
29729 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
29731 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29733 * cfganal.c (inverted_post_order_compute): Change argument type
29735 * cfganal.h (inverted_post_order_compute): Adjust prototype.
29736 * df-core.c (rest_of_handle_df_initialize): Adjust.
29737 (rest_of_handle_df_finish): Likewise.
29738 (df_analyze_1): Likewise.
29739 (df_analyze): Likewise.
29740 (loop_inverted_post_order_compute): Change argument to be a vec *.
29741 (df_analyze_loop): Adjust.
29742 (df_get_n_blocks): Likewise.
29743 (df_get_postorder): Likewise.
29744 * df.h (struct df_d): Change field to be a vec.
29745 * lcm.c (compute_laterin): Adjust.
29746 (compute_available): Likewise.
29747 * lra-lives.c (lra_create_live_ranges_1): Likewise.
29748 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
29749 * tree-ssa-pre.c (compute_antic): Likewise.
29751 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29753 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
29754 (depth_first_search::depth_first_search): Change structure init
29755 function to this constructor.
29756 (depth_first_search::add_bb): Rename function to this member.
29757 (depth_first_search::execute): Likewise.
29758 (flow_dfs_compute_reverse_finish): Adjust.
29760 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29762 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
29763 (longest_simple_path): Likewise.
29764 * shrink-wrap.c (spread_components): Likewise.
29765 (disqualify_problematic_components): Likewise.
29766 (emit_common_heads_for_components): Likewise.
29767 (emit_common_tails_for_components): Likewise.
29768 (insert_prologue_epilogue_for_components): Likewise.
29770 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29772 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
29775 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29777 * df-core.c (df_set_blocks): Start using auto_bitmap.
29778 (df_compact_blocks): Likewise.
29779 * df-problems.c (df_rd_confluence_n): Likewise.
29780 * df-scan.c (df_insn_rescan_all): Likewise.
29781 (df_process_deferred_rescans): Likewise.
29782 (df_update_entry_block_defs): Likewise.
29783 (df_update_exit_block_uses): Likewise.
29784 (df_entry_block_bitmap_verify): Likewise.
29785 (df_exit_block_bitmap_verify): Likewise.
29786 (df_scan_verify): Likewise.
29787 * lra-constraints.c (lra_constraints): Likewise.
29788 (undo_optional_reloads): Likewise.
29789 (lra_undo_inheritance): Likewise.
29790 * lra-remat.c (calculate_gen_cands): Likewise.
29791 (do_remat): Likewise.
29792 * lra-spills.c (assign_spill_hard_regs): Likewise.
29793 (spill_pseudos): Likewise.
29794 * tree-ssa-pre.c (bitmap_set_and): Likewise.
29795 (bitmap_set_subtract_values): Likewise.
29797 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29799 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
29800 management with auto_bitmap.
29801 (fix_inter_tick): Likewise.
29802 (fix_recovery_deps): Likewise.
29803 * ira.c (add_store_equivs): Likewise.
29804 (find_moveable_pseudos): Likewise.
29805 (split_live_ranges_for_shrink_wrap): Likewise.
29806 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
29807 (rtx_reuse_manager::seen_def_p): Likewise.
29808 (rtx_reuse_manager::set_seen_def): Likewise.
29809 * print-rtl.h (class rtx_reuse_manager): Likewise.
29811 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29813 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
29815 (migrate_btr_def): Likewise.
29816 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
29817 * df-core.c (loop_post_order_compute): Likewise.
29818 (loop_inverted_post_order_compute): Likewise.
29819 * hsa-common.h: Likewise.
29820 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
29821 * init-regs.c (initialize_uninitialized_regs): Likewise.
29822 * ipa-inline.c (resolve_noninline_speculation): Likewise.
29823 (inline_small_functions): Likewise.
29824 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
29825 * ira.c (combine_and_move_insns): Likewise.
29826 (build_insn_chain): Likewise.
29827 * loop-invariant.c (find_invariants): Likewise.
29828 * lower-subreg.c (propagate_pseudo_copies): Likewise.
29829 * predict.c (tree_predict_by_opcode): Likewise.
29830 (predict_paths_leading_to): Likewise.
29831 (predict_paths_leading_to_edge): Likewise.
29832 (estimate_loops_at_level): Likewise.
29833 (estimate_loops): Likewise.
29834 * shrink-wrap.c (try_shrink_wrapping): Likewise.
29835 (spread_components): Likewise.
29836 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
29837 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
29838 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
29839 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
29840 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
29841 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
29842 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
29843 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
29844 (mark_threaded_blocks): Likewise.
29845 (thread_through_all_blocks): Likewise.
29846 * tree-ssa.c (verify_ssa): Likewise.
29847 (execute_update_addresses_taken): Likewise.
29848 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
29850 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29852 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
29854 (post_order_compute): Likewise.
29855 (inverted_post_order_compute): Likewise.
29856 (pre_and_rev_post_order_compute_fn): Likewise.
29858 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29860 * genrecog.c (int_set::int_set): Explicitly construct our
29861 auto_vec base class.
29862 * vec.h (auto_vec::auto_vec): New constructor.
29864 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29866 * bitmap.h (class auto_bitmap): New constructor taking
29867 bitmap_obstack * argument.
29869 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29871 * bitmap.h (class auto_bitmap): Change type of m_bits to
29872 bitmap_head, and adjust ctor / dtor and member operators.
29874 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
29876 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
29877 when returned register mode doesn't match original mode.
29879 2017-05-12 Jeff Law <law@redhat.com>
29880 Jakub Jelinek <jakub@redhat.com>
29882 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
29883 we look for cc setter after the compare-elim changes.
29884 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
29885 within the vector to match what compare-elim now expects.
29886 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
29887 (xorsi3_flags, one_cmplsi2_flags): Likewise.
29889 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
29890 after the compare-elim changes.
29891 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
29892 the vector to match what compare-elim now expects.
29893 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
29894 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
29895 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
29896 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
29897 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
29899 * config/visium/visium.c (single_set_and_flags): Fix where
29900 we look for cc setter after the compare-elim changes.
29901 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
29902 with the vector to match what compare-elim now expects.
29903 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
29904 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
29905 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
29906 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
29907 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
29908 (neg<mode>2_insn_set_overflow): Likewise.
29910 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
29912 PR middle-end/79794
29913 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
29914 maybe_expand_insn call, set ops[0].target. If still set after call,
29915 set alt_rtl. Add extra arg to recursive calls.
29916 (extract_bit_field): Add alt_rtl argument. Pass to
29918 * expmed.h (extract_bit_field): Fix prototype.
29919 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
29920 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
29921 to extract_bit_field_calls.
29922 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
29923 Pass alt_rtl to extract_bit_field calls.
29924 * calls.c (store_unaligned_arguments_into_psuedos)
29925 load_register_parameters): Pass extra NULL to extract_bit_field calls.
29926 * optabs.c (maybe_legitimize_operand): Clear op->target when call
29928 * optabs.h (struct expand_operand): Add target bitfield.
29930 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
29932 * compare-elim.c (try_eliminate_compare): Canonicalize
29933 operation with embedded compare to
29934 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
29935 (set (reg) (operation)].
29937 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
29939 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
29942 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
29943 cost of adding a carry flag for ADC instruction.
29944 [case MINUS]: Ignore the cost of subtracting a carry flag
29945 for SBB instruction.
29947 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
29949 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
29951 * config/rs6000/bmiintrin.h: New file.
29952 * config/rs6000/bmi2intrin.h: New file.
29953 * config/rs6000/x86intrin.h: New file.
29955 2017-05-12 Jeff Law <law@redhat.com>
29957 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
29960 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
29962 PR middle-end/80707
29963 * tree-cfg.c: Remove cfg edges of unreachable case statements.
29965 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29967 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
29968 early expansion of vector divide builtins.
29969 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
29970 builtins identified as having unsigned arguments.
29972 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29974 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
29975 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
29976 expansion of vector logical operations (and, andc, or, xor,
29979 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29981 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
29982 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
29984 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29986 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
29987 early GIMPLE expansion of vector multiplies.
29989 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29991 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
29992 TARGET_HAVE_MOVT conditional.
29993 (movt splitter): Likewise.
29995 2017-05-12 Richard Biener <rguenther@suse.de>
29997 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
29998 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
29999 Fold all stmts not inplace.
30001 2017-05-12 Richard Biener <rguenther@suse.de>
30003 PR tree-optimization/80713
30004 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
30005 inserted_exprs bit for not removed stmts.
30007 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
30009 PR middle-end/69921
30010 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
30011 parallelized" attribute for parallelized OpenACC kernels.
30012 * omp-offload.c (execute_oacc_device_lower): Use it.
30014 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
30015 Set "oacc kernels" attribute.
30016 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
30017 parameter. Adjust all users.
30018 (oacc_fn_attrib_kernels_p): Remove function.
30019 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
30020 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
30021 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
30022 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
30023 assert "oacc kernels" attribute is set.
30025 2017-05-11 Carl Love <cel@us.ibm.com>
30027 * config/rs6000/rs6000-c: Add support for built-in functions
30028 vector unsigned char vec_popcnt (vector signed char)
30029 vector unsigned char vec_popcnt (vector unsigned char)
30030 vector unsigned short vec_popcnt (vector signed short)
30031 vector unsigned short vec_popcnt (vector unsigned short)
30032 vector unsigned int vec_popcnt (vector signed int)
30033 vector unsigned int vec_popcnt (vector unsigned int)
30034 vector unsigned long long vec_popcnt (vector signed long long)
30035 vector unsigned long long vec_popcnt (vector unsigned long long)
30036 vector signed long long vec_slo (vector signed long long,
30037 vector signed char)
30038 vector signed long long vec_slo (vector signed long long,
30039 vector unsigned char)
30040 vector unsigned long long vec_slo (vector unsigned long long,
30041 vector signed char)
30042 vector unsigned long long vec_slo (vector unsigned long long,
30043 vector unsigned char)
30044 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
30045 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
30046 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
30047 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
30048 * doc/extend.texi: Update the built-in documentation file for the
30049 new built-in functions.
30051 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
30053 * attribs.h (sorted_attr_string): Move machine independent
30054 functions for target clone support from the i386 port to common
30055 code. Rename ix86_function_versions to common_function_versions.
30056 Rename make_name to make_unique_name.
30057 (common_function_versions): Likewise.
30058 (make_unique_name): Likewise.
30059 (make_dispatcher_decl): Likewise.
30060 (is_function_default_version): Likewise.
30061 * attribs.c (attr_strcmp): Likewise.
30062 (sorted_attr_string): Likewise.
30063 (common_function_versions): Likewise.
30064 (make_unique_name): Likewise.
30065 (make_dispatcher_decl): Likewise.
30066 (is_function_default_version): Likewise.
30067 * config/i386/i386.c (attr_strcmp): Likewise.
30068 (sorted_attr_string): Likewise.
30069 (ix86_function_versions): Likewise.
30070 (make_name): Likewise.
30071 (make_dispatcher_decl): Likewise.
30072 (is_function_default_version): Likewise.
30073 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
30075 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30078 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
30079 Account for direct move costs for vec_construct of integer
30082 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
30085 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
30086 (UNSPEC_STX_ATOMIC): Ditto.
30087 (loaddi_via_sse): New insn.
30088 (storedi_via_sse): Ditto.
30089 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
30090 Update corresponding peephole2 patterns.
30091 (atomic_storedi_fpu): Ditto.
30093 2017-05-11 Julia Koval <julia.koval@intel.com>
30095 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
30096 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
30098 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
30099 (__builtin_ia32_rsqrt14ss_mask): New builtins.
30100 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
30102 2017-05-11 Nathan Sidwell <nathan@acm.org>
30104 * graphite-poly.c: Include dumpfile.h.
30106 * dumpfle.h (dump_function): Declare here ...
30107 * tree-dump.h (dump_function): ... not here.
30108 * dumpfile.c: #include tree-cfg.h.
30109 (dump_function): Move here from ...
30110 * tree-dump.c (dump_function): ... here.
30111 * gimplify.c: #include splay-tree.h, not tree-dump.h.
30112 * graphite-poly.c: Don't include tree-dump.h.
30113 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
30114 * print-tree.c: Likewise.
30115 * stor-layout.c: Likewise.
30116 * tree-nested.c: Likewise.
30118 * dumpfile.c (dump_start): Use TDF_FLAGS.
30119 (dump_enable_all): Fix TDF_KIND check thinko.
30121 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
30123 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
30124 array entries to represent two legal parameterizations of the
30125 overloaded __builtin_cmpb function, as represented by the
30126 P6_OV_BUILTIN_CMPB constant.
30127 (altivec_resolve_overloaded_builtin): Add special case handling
30128 for the __builtin_cmpb function, as represented by the
30129 P6_OV_BUILTIN_CMPB constant.
30130 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
30131 (BU_P6_64BIT_2): New macro.
30132 (BU_P6_OVERLOAD_2): New macro
30133 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
30134 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
30135 (CMPB): Add overload support to represent both 32-bit and 64-bit
30136 compare-bytes function.
30137 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
30138 support for TARGET_CMPB.
30139 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
30140 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
30141 documentation of the __builtin_cmpb overloaded built-in function.
30143 2017-05-11 Richard Biener <rguenther@suse.de>
30145 PR tree-optimization/80705
30146 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
30147 bases are not vectorizable.
30149 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30151 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
30152 when counting register pressure.
30154 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30156 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
30157 (iv_ca_more_deps): Renamed to ...
30158 (iv_ca_compare_deps): ... this.
30159 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
30161 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30163 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
30165 (determine_group_iv_costs): ... here.
30166 (find_inv_vars_cb): Record inv var if it's not recorded before.
30168 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30170 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
30171 (get_shiftadd_cost): Ditto.
30173 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30175 * tree-ssa-address.c: Include header file.
30176 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
30178 (add_to_parts): Refactor.
30179 (addr_to_parts): New parameter. Update use of move_hint_to_base.
30180 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
30183 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30185 PR tree-optimization/53090
30186 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
30188 (extract_cond_operands): Detect condition with IV on both sides
30189 and return COMP_IV_EXPR_2.
30190 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
30191 (rewrite_use_compare): Simplify by removing call to function
30192 extract_cond_operands.
30194 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30196 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
30197 (extract_cond_operands): Detect condition comparing against non-
30198 invariant bound and return appropriate enum value.
30199 (find_interesting_uses_cond): Update use of extract_cond_operands.
30200 Handle its return value accordingly.
30201 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
30203 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30205 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
30206 nonlinear iv_use computation in loop invariant sensitive way.
30208 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30210 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
30211 (find_iv_candidates): Call relate_compare_use_with_all_cands.
30213 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30215 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
30216 (dump_cand): Support iv_cand.inv_exprs.
30217 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
30219 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
30222 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30224 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
30226 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
30227 as local function. Include necessary header files.
30228 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
30230 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30232 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
30234 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30236 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
30237 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
30238 RSHIFT_EXPR and BIT_NOT_EXPR.
30240 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30242 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
30243 (adjust_setup_cost): New parameter supporting round up adjustment.
30244 (struct address_cost_data): Delete.
30245 (force_expr_to_var_cost): Don't bound cost with spill_cost.
30246 (split_address_cost, ptr_difference_cost): Delete.
30247 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
30248 (struct ainc_cost_data): New struct.
30249 (get_address_cost_ainc): New function.
30250 (get_address_cost, get_computation_cost): Reimplement.
30251 (determine_group_iv_cost_address): Record inv_expr for all uses of
30253 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
30254 (iv_ca_has_deps): Reimplemented to ...
30255 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
30257 (iv_ca_extend): Call iv_ca_more_deps.
30259 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30261 * tree-ssa-address.c (struct mem_address): Move to header file.
30262 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
30263 * tree-ssa-address.h (struct mem_address): Move from C file.
30264 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
30266 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30268 * tree-affine.h (aff_combination_type): New interface.
30269 (aff_combination_zero_p): Remove static.
30270 (aff_combination_const_p): New interface.
30271 (aff_combination_singleton_var_p): New interfaces.
30273 2017-05-11 Richard Biener <rguenther@suse.de>
30275 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
30276 Skip unreachable blocks and destinations.
30277 (eliminate): Move stmt removal and fixup ...
30278 (fini_eliminate): ... here. Skip inserted exprs.
30279 (pass_pre::execute): Move fini_pre after fini_eliminate.
30280 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
30281 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
30282 PRE to get rid of dead code that has invalid SSA form and
30283 split critical edges again.
30285 2017-05-11 Bin Cheng <bin.cheng@arm.com>
30287 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30289 2017-05-11 Richard Biener <rguenther@suse.de>
30291 * passes.c (execute_function_todo): Verify loops if they are
30292 said to be up-to-date.
30293 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
30294 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
30296 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
30299 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
30300 handle calling assemble_external ourself.
30303 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
30304 modes with zero size. Enhance comment.
30306 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30308 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
30309 built-ins for vec_xl and vec_xst with short and char pointer
30312 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30314 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
30315 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
30316 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
30317 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
30318 (_mm_maskz_min_round_ss): New intrinsics.
30319 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
30320 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
30321 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
30322 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
30323 (__builtin_ia32_minss_mask_round): New builtins.
30324 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30325 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30326 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
30328 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
30329 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
30331 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
30334 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
30336 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
30337 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
30338 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
30339 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
30340 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
30341 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
30342 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
30343 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30344 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30345 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
30346 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
30347 (__builtin_ia32_mulss_mask_round): New builtins.
30348 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30349 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30350 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
30352 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
30353 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30355 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30358 2017-05-10 Julia Koval <julia.koval@intel.com>
30360 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
30361 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
30362 (_mm256_setr_m128i): New intrinsics.
30364 2017-05-10 Julia Koval <julia.koval@intel.com>
30366 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
30367 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
30368 (_mm_maskz_rcp14_ss): New intrinsics.
30369 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
30370 (__builtin_ia32_rcp14ss_mask): New builtins.
30371 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
30373 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
30375 PR tree-optimization/51513
30376 * tree-cfg.c (gimple_seq_unreachable_p): New function.
30377 (assert_unreachable_fallthru_edge_p): Use it.
30378 (group_case_labels_stmt): Likewise.
30379 * tree-cfg.h: Prototype it.
30380 * stmt.c: Include cfghooks.h and tree-cfg.h.
30381 (emit_case_dispatch_table) <gap_label>: New local variable.
30382 Use it to fill dispatch table gaps.
30383 Test for default_label before updating probabilities.
30384 (expand_case) <default_label>: Remove unneeded initialization.
30385 Test for unreachable default case statement and remove its edge.
30386 Set default_label accordingly.
30387 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
30389 2017-05-10 Carl Love <cel@us.ibm.com>
30391 * config/rs6000/rs6000-c: Add support for built-in functions
30392 vector signed char vec_neg (vector signed char)
30393 vector signed short int vec_neg (vector short int)
30394 vector signed int vec_neg (vector signed int)
30395 vector signed long long vec_neg (vector signed long long)
30396 vector float vec_neg (vector float)
30397 vector double vec_neg (vector double)
30398 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
30400 * config/rs6000/altivec.h: Add define for vec_neg
30401 * doc/extend.texi: Update the built-in documentation for the
30402 new built-in functions.
30404 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30406 PR tree-optimization/77644
30407 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
30409 2017-05-10 Nathan Sidwell <nathan@acm.org>
30411 * dumpfile.h (TDI_lang_all): New.
30412 (TDF_KIND): New. Renumber others
30413 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
30415 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
30417 (get_dump_file_name): Adjust suffix generation.
30418 (dump_enable_all): Use TDF_KIND.
30419 * doc/invoke.texi (-fdump-lang-all): Document.
30421 * dumpfile.h: Tabify.
30423 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
30426 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
30427 Move member access before delete.
30429 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
30431 * tree-inline.c (expand_call_inline): Split block at stmt
30434 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
30437 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
30438 are now unused after splitting mov{sf,sd}_hardfloat.
30439 (f32_lr2): Likewise.
30440 (f32_lm): Likewise.
30441 (f32_lm2): Likewise.
30442 (f32_li): Likewise.
30443 (f32_li2): Likewise.
30444 (f32_lv): Likewise.
30445 (f32_sr): Likewise.
30446 (f32_sr2): Likewise.
30447 (f32_sm): Likewise.
30448 (f32_sm2): Likewise.
30449 (f32_si): Likewise.
30450 (f32_si2): Likewise.
30451 (f32_sv): Likewise.
30452 (f32_dm): Likewise.
30453 (f32_vsx): Likewise.
30454 (f32_av): Likewise.
30455 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
30456 For movsf, order stores so the VSX stores occur before the GPR
30457 store which encourages the register allocator to use a traditional
30458 FPR instead of a GPR. For movsd, order the stores so that the GPR
30459 store comes before the VSX stores to allow the power6 to work.
30460 This is due to the power6 not having a 32-bit integer store
30461 instruction from a FPR.
30462 (movsf_hardfloat): Likewise.
30463 (movsd_hardfloat): Likewise.
30465 2017-05-09 Martin Sebor <msebor@redhat.com>
30467 PR translation/80280
30468 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
30471 PR translation/80280
30472 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
30473 data member added in r247778.
30474 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
30476 2017-05-09 Nathan Sidwell <nathan@acm.org>
30478 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
30480 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
30483 2017-05-09 Marek Polacek <polacek@redhat.com>
30485 * doc/invoke.texi: Fix typo.
30487 2017-05-09 Richard Biener <rguenther@suse.de>
30489 * tree-vrp.c (vrp_val_is_max): Adjust comment.
30490 (vrp_val_is_min): Likewise.
30491 (set_value_range_to_value): Likewise.
30492 (set_value_range_to_nonnegative): Likewise.
30493 (gimple_assign_nonzero_p): Likewise.
30494 (gimple_stmt_nonzero_p): Likewise.
30495 (vrp_int_const_binop): Likewise. Remove unreachable case.
30496 (adjust_range_with_scev): Adjust comments.
30497 (compare_range_with_value): Likewise.
30498 (extract_range_from_phi_node): Likewise.
30499 (test_for_singularity): Likewise.
30501 2017-05-09 Richard Biener <rguenther@suse.de>
30503 * tree-vrp.c (get_single_symbol): Add assert that we don't
30504 get overflowed constants as invariant part.
30505 (compare_values_warnv): Add comment before the TREE_NO_WARNING
30506 checks. Use wi::cmp instead of recursing for integer constants.
30507 (compare_values): Just ignore whether we assumed undefined
30508 overflow instead of failing the compare.
30509 (extract_range_for_var_from_comparison_expr): Add comment before the
30510 TREE_NO_WARNING sets.
30511 (test_for_singularity): Likewise.
30512 (extract_range_from_comparison): Do not disable optimization
30513 when we assumed undefined overflow.
30514 (extract_range_basic): Remove init of unused var.
30516 2017-05-09 Richard Biener <rguenther@suse.de>
30518 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
30519 (extract_range_from_multiplicative_op_1): Adjust.
30520 (extract_range_from_binary_expr_1): Use int_const_binop.
30522 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
30525 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
30526 rs6000_store_data_bypass_p in seven define_bypass directives and
30527 in several comments.
30528 * config/rs6000/rs6000-protos.h: Add prototype for
30529 rs6000_store_data_bypass_p function.
30530 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
30531 function implements slightly different (rs6000-specific) semantics
30532 than store_data_bypass_p, returning false rather than aborting
30533 with assertion error when arguments do not satisfy the
30534 requirements of store data bypass.
30535 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
30536 rs6000_store_data_bypass_p.
30538 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
30540 * config/xtensa/xtensa-protos.h
30541 (xtensa_initial_elimination_offset): New declaration.
30542 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30543 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
30544 macro definition, add case for FRAME_POINTER_REGNUM when
30545 FRAME_GROWS_DOWNWARD.
30546 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
30547 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
30548 xtensa_initial_elimination_offset.
30550 2017-05-08 Nathan Sidwell <nathan@acm.org>
30552 * doc/invoke.texi: Alphabetize -fdump options.
30554 2017-05-08 Martin Sebor <msebor@redhat.com>
30556 PR translation/80280
30557 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
30559 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
30561 * target.def (compute_frame_layout): New optional target hook.
30562 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
30563 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
30564 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
30566 * reload1.c (verify_initial_elim_offsets): Likewise.
30567 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
30568 (use_simple_return_p): Call arm_compute_frame_layout if needed.
30569 (arm_get_frame_offsets): Split up into this ...
30570 (arm_compute_frame_layout): ... and this function.
30572 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
30574 * config/aarch64/constraints.md (Usa): New constraint.
30575 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
30577 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
30579 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
30580 with_multilib_list after it has been checked.
30582 2017-05-08 Richard Biener <rguenther@suse.de>
30584 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
30585 (bitmap_set_subtract_values): Likewise.
30587 2017-05-08 Richard Biener <rguenther@suse.de>
30589 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
30590 (gimple_assign_nonzero): ... this and remove strict_overflow_p
30592 (gimple_stmt_nonzero_warnv_p): Rename to ...
30593 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
30595 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
30596 (extract_range_basic): Adjust, do not disable propagation on
30597 strict overflow sensitive simplification.
30598 (vrp_visit_cond_stmt): Likewise.
30600 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
30602 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
30603 body size unconditionally.
30605 2017-05-07 Jeff Law <law@redhat.com>
30608 2017-05-06 Jeff Law <law@redhat.com>
30609 PR tree-optimization/78496
30610 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
30613 PR tree-optimization/78496
30614 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
30615 (simplify_stmt_using_ranges): Call it.
30616 (vrp_dom_walker::before_dom_children): Extract equivalences
30617 from an ASSERT_EXPR with an equality comparison against a
30620 2017-05-06 Jeff Law <law@redhat.com>
30622 PR tree-optimization/78496
30623 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
30626 PR tree-optimization/78496
30627 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
30628 (simplify_stmt_using_ranges): Call it.
30629 (vrp_dom_walker::before_dom_children): Extract equivalences
30630 from an ASSERT_EXPR with an equality comparison against a
30633 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
30635 * lra-constraints.c (lra_copy_reg_equiv): New function.
30636 (split_reg): Use it to copy equivalence information from the
30637 original register to the spill register.
30639 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
30641 PR rtl-optimization/75964
30642 * simplify-rtx.c (simplify_const_relational_operation): Remove
30643 invalid handling of comparisons of integer ABS.
30645 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
30647 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
30648 initialize to zero.
30649 (init_regs): Remove declaration.
30650 (function_arg_advance_32): Initialize error_p as boolean variable.
30652 2017-05-05 Nathan Sidwell <nathan@acm.org>
30654 * store-motion.c (remove_reachable_equiv_notes): Reformat long
30655 lines. Use for (;;).
30657 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30659 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
30660 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
30661 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
30662 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
30663 VF=2 that require versioning.
30665 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30667 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
30670 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30672 * diagnostic.h (diagnostic_override_option_index): Convert from
30673 macro to inline function.
30675 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30677 * diagnostic.c (last_module_changed_p): New function.
30678 (set_last_module): New function.
30679 (diagnostic_report_current_module): Convert macro usage to
30680 the above functions.
30681 * diagnostic.h (diagnostic_context::last_module): Strengthen
30682 from const line_map * to const line_map_ordinary *.
30683 (diagnostic_last_module_changed): Delete macro.
30684 (diagnostic_set_last_module): Delete macro.
30686 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30688 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
30689 with diagnostic_report_diagnostic.
30690 (diagnostic_n_impl_richloc): Likewise.
30691 * diagnostic.h (report_diagnostic): Delete macro.
30692 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
30693 with diagnostic_report_diagnostic.
30694 * substring-locations.c (format_warning_va): Likewise.
30696 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30698 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
30699 save/restor of format_spec. Move option-printing code to...
30700 (print_option_information): ...this new function, and
30701 reimplement by simply printing to the pretty_printer,
30702 rather than appending to the format string.
30704 2017-05-05 David Malcolm <dmalcolm@redhat.com>
30706 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
30707 handling logic into...
30708 (update_effective_level_from_pragmas): ...this new function.
30710 2017-05-04 Andrew Waterman <andrew@sifive.com>
30712 * config/riscv/riscv.opt (mstrict-align): New option.
30713 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
30714 (SLOW_UNALIGNED_ACCESS): Define.
30715 (riscv_slow_unaligned_access): Declare.
30716 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
30718 (riscv_slow_unaligned_access): New variable.
30719 (rocket_tune_info): Set slow_unaligned_access to true.
30720 (optimize_size_tune_info): Set slow_unaligned_access to false.
30721 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
30722 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
30723 (riscv_option_override): Set riscv_slow_unaligned_access.
30724 * doc/invoke.texi: Add -mstrict-align to RISC-V.
30726 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
30728 * config/riscv/riscv.md: Unify indentation.
30730 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
30735 * config/rs6000/rs6000.md (u code attribute): Add FIX and
30737 (extendsi<mode>2): Add support for doing sign extension via
30738 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
30739 don't have ISA 3.0 instructions.
30740 (extendsi<mode>2 splitter): Likewise.
30741 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
30742 generate the normal insns since SImode can now go in vector
30743 registers. Disallow the special UNSPECs needed for previous
30744 machines to hide SImode being used. Add new insns
30745 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
30746 (fix_trunc<mode>si2_stfiwx): Likewise.
30747 (fix_trunc<mode>si2_internal): Likewise.
30748 (fixuns_trunc<mode>si2): Likewise.
30749 (fixuns_trunc<mode>si2_stfiwx): Likewise.
30750 (fctiw<u>z_<mode>_smallint): Likewise.
30751 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
30752 of floating point to 32-bit integer from doing a direct move to
30753 the GPR registers to do a store.
30754 (fctiwz_<mode>): Break long line.
30756 2017-05-05 Bin Cheng <bin.cheng@arm.com>
30758 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
30759 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
30760 (addr_list, addr_offset_valid_p): New.
30761 (split_address_groups): Check offset validity with above function.
30762 (gt-tree-ssa-loop-ivopts.h): Include header file.
30764 2017-05-05 Nathan Sidwell <nathan@acm.org>
30766 * config.gcc (arm*-*-*): Add missing 'fi'.
30768 2017-05-05 Steve Ellcey <sellcey@cavium.com>
30770 * doc/invoke.texi (-fopt-info): Explicitly say order of options
30771 included in -fopt-info does not matter.
30772 * doc/optinfo.texi (-fopt-info): Fix description of default
30773 behavour. Explicitly say order of options included in -fopt-info
30776 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
30778 * config.gcc: Allow combinations of aprofile and rmprofile values for
30779 --with-multilib-list.
30780 * config/arm/t-multilib: New file.
30781 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
30782 variables. Remove setting of ISA and floating-point ABI in
30783 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
30784 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
30785 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
30786 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
30788 * config/arm/t-rmprofile: Likewise except for the matches changes.
30789 * doc/install.texi (--with-multilib-list): Document the combination of
30790 aprofile and rmprofile values and warn about pitfalls in doing that.
30792 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30794 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
30795 (movdi_aarch64): Likewise.
30797 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30799 PR tree-optimization/80632
30800 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
30802 (build_arrays): Initialize it for virtual phis.
30803 (fix_phi_nodes): Use it for virtual phis.
30805 PR tree-optimization/80558
30806 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
30807 [x, y] op z into [x op, y op z] for op & or | if conditions
30810 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
30811 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30814 * config/arm/arm.md (use_literal_pool): Remove.
30815 (64-bit immediate split): No longer takes cost into consideration
30816 if arm_disable_literal_pool is enabled.
30817 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
30818 used when arm_disable_literal_pool is enabled.
30819 (arm_max_const_double_inline_cost): Remove use of
30820 arm_disable_literal_pool.
30821 (push_minipool_fix): Add assert.
30822 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
30823 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
30824 (no_literal_pool_sf_immediate): New.
30826 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30828 PR tree-optimization/80613
30829 * tree-ssa-dce.c (propagate_necessity): Remove cases for
30830 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
30832 2017-05-05 Richard Biener <rguenther@suse.de>
30834 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
30836 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
30838 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
30839 of this flag from insn conditions due to removal from r247495.
30841 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30843 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
30845 (arm_early_store_addr_dep_ptr): Likewise.
30846 * config/arm/aarch-common-protos.h
30847 (arm_early_load_addr_dep_ptr): Add prototype.
30848 (arm_early_store_addr_dep_ptr): Likewise.
30849 * config/arm/cortex-a53.md: Add new bypasses.
30851 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30853 * tree.c (next_type_uid): Change type to unsigned.
30854 (type_hash_canon): Decrement back next_type_uid if
30855 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
30856 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
30859 2017-05-04 Martin Sebor <msebor@redhat.com>
30861 * builtins.c: Fix a trivial typo in a comment.
30863 PR middle-end/79234
30864 * builtins.c (check_sizes): Adjust to handle reading past the end.
30865 Avoid printing excessive upper bound of ranges. Use %E to print
30866 tree nodes instead of converting them to %wu.
30867 (expand_builtin_memchr): New function.
30868 (compute_dest_size): Rename...
30869 (compute_objsize): ...to this.
30870 (expand_builtin_memcpy): Adjust.
30871 (expand_builtin_mempcpy): Adjust.
30872 (expand_builtin_strcat): Adjust.
30873 (expand_builtin_strcpy): Adjust.
30874 (check_strncat_sizes): Adjust.
30875 (expand_builtin_strncat): Adjust.
30876 (expand_builtin_strncpy): Adjust and simplify.
30877 (expand_builtin_memset): Adjust.
30878 (expand_builtin_bzero): Adjust.
30879 (expand_builtin_memcmp): Adjust.
30880 (expand_builtin): Handle memcmp.
30881 (maybe_emit_chk_warning): Check strncat just once.
30883 2017-05-04 Martin Sebor <msebor@redhat.com>
30885 PR preprocessor/79214
30886 PR middle-end/79222
30887 PR middle-end/79223
30888 * builtins.c (check_sizes): Add inlining context and issue
30889 warnings even when -Wno-system-headers is set.
30890 (check_strncat_sizes): Same.
30891 (expand_builtin_strncat): Same.
30892 (expand_builtin_memmove): New function.
30893 (expand_builtin_stpncpy): Same.
30894 (expand_builtin): Handle memmove and stpncpy.
30896 2017-05-04 Bin Cheng <bin.cheng@arm.com>
30898 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
30899 which is not used any more.
30901 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30903 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
30905 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30907 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
30908 (cortexa53_tunings): Likewise.
30909 (cortexa57_tunings): Likewise.
30910 (cortexa72_tunings): Likewise.
30911 (cortexa73_tunings): Likewise.
30913 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30915 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
30916 Set loop alignment to 8.
30918 2017-05-04 Martin Sebor <msebor@redhat.com>
30920 PR translation/80280
30921 * builtins.c (expand_builtin_object_size): Add missing quoting to
30922 %D and like directives.
30923 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
30924 (hsa_type_for_tree_type): Same.
30925 (verify_function_arguments): Same.
30926 * symtab.c (symbol_table::change_decl_assembler_name): Same.
30927 * varasm.c (get_section): Same.
30930 2017-05-04 Martin Sebor <msebor@redhat.com>
30932 PR translation/80280
30933 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
30935 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30937 * config/aarch64/aarch64.c (generic_addrcost_table):
30938 Change HI/TI mode setting.
30940 2017-05-04 Martin Jambor <mjambor@suse.cz>
30942 PR tree-optimization/80622
30943 * tree-sra.c (comes_initialized_p): New function.
30944 (build_accesses_from_assign): Only set write lazily when
30945 comes_initialized_p is false.
30946 (analyze_access_subtree): Use comes_initialized_p.
30947 (propagate_subaccesses_across_link): Assert !comes_initialized_p
30948 instead of testing for PARM_DECL.
30950 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30952 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
30953 constraint on operand 0 to allow more general addressing modes.
30954 Adjust output template.
30955 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
30957 * config/aarch64/aarch64-protos.h
30958 (aarch64_address_valid_for_prefetch_p): Declare prototype.
30959 * config/aarch64/constraints.md (Dp): New address constraint.
30960 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
30963 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
30965 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
30966 update use of estimate_ipcp_clone_size_and_time.
30967 (estimate_local_effects): Update use of
30968 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
30969 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
30970 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
30971 Return nonspecialized time.
30973 2017-05-04 Richard Biener <rguenther@suse.de>
30975 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
30976 for the last VUSE which def dominates the PHI. Directly call
30978 (get_continuation_for_phi_1): Remove.
30980 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
30982 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
30983 to explain the use of truncating division. Cap the number of
30984 iterations to the maximum given by nb_iterations_upper_bound,
30987 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
30989 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
30990 * configure: Regenerate.
30991 * config.in: Regenerate.
30992 * config/i386/driver-mingw32.c: new file.
30993 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
30994 * config.host: Link driver-mingw32.o on MinGW host.
30995 * doc/install.texi: Document new --enable-mingw-wildcard configure
30998 2017-05-04 Marek Polacek <polacek@redhat.com>
31000 PR tree-optimization/80612
31001 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
31003 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31004 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
31006 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
31007 (movt splitter): Likewise.
31008 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
31009 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
31010 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
31011 block for Thumb-1 with MOVT.
31012 (thumb2_legitimate_address_p): Move code block ...
31013 (can_avoid_literal_pool_for_label_p): ... into this new function.
31014 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
31016 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
31017 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
31018 "M-profile targets with the MOVT instruction".
31020 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31022 * config/arm/arm-builtins.c (arm_init_builtins): Rename
31023 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
31024 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
31026 2017-05-04 Martin Liska <mliska@suse.cz>
31028 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
31029 variable cond_code.
31031 2017-05-04 Richard Biener <rguenther@suse.de>
31033 * tree.c (array_at_struct_end_p): Handle arrays at struct
31034 end with flexarrays more conservatively. Refactor and treat
31035 arrays of arrays or aggregates more strict. Fix
31036 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
31037 * tree.c (array_at_struct_end_p): Adjust prototype.
31038 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
31039 * gimple-fold.c (get_range_strlen): Likewise.
31040 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
31042 2017-05-04 Richard Biener <rguenther@suse.de>
31044 PR tree-optimization/31130
31045 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
31047 (supports_overflow_infinity): Likewise.
31048 (is_negative_overflow_infinity): Likewise.
31049 (is_positive_overflow_infinity): Likewise.
31050 (is_overflow_infinity): Likewise.
31051 (stmt_overflow_infinity): Likewise.
31052 (overflow_infinity_range_p): Likewise.
31053 (usable_range_p): Remove as always returning true.
31054 (make_overflow_infinity): Remove.
31055 (negative_overflow_infinity): Likewise.
31056 (positive_overflow_infinity): Likewise.
31057 (avoid_overflow_infinity): Likewise.
31058 (set_value_range): Adjust accordingly.
31059 (set_value_range_to_nonnegative): Likewise, remove now unused
31060 overflow_infinity arg.
31061 (vrp_operand_equal_p): Adjust.
31062 (update_value_range): Likewise.
31063 (range_int_cst_singleton_p): Likewise.
31064 (operand_less_p): Likewise.
31065 (compare_values_warnv): Likewise.
31066 (extract_range_for_var_from_comparison_expr): Likewise.
31067 (vrp_int_const_binop): Likewise.
31068 (zero_nonzero_bits_from_vr): Likewise.
31069 (extract_range_from_multiplicative_op_1): Likewise.
31070 (extract_range_from_binary_expr_1): Likewise.
31071 (extract_range_from_unary_expr): Likewise.
31072 (extract_range_from_comparison): Likewise.
31073 (extract_range_basic): Likewise.
31074 (adjust_range_with_scev): Likewise.
31075 (compare_ranges): Likewise.
31076 (compare_range_with_value): Likewise.
31077 (dump_value_range): Likewise.
31078 (test_for_singularity): Likewise, remove strict_overflow_p parameter
31080 (simplify_cond_using_ranges): Adjust.
31082 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31084 * brig-builtins.def: Added a builtin for class_f64.
31085 * builtin-types.def: Added a builtin type needed by class_f64.
31087 2017-05-03 Jason Merrill <jason@redhat.com>
31089 * timevar.def: Add TV_CONSTEXPR.
31091 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31093 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
31095 2017-05-03 Martin Jambor <mjambor@suse.cz>
31097 * ipa-prop.c (ipa_update_after_lto_read): Removed.
31098 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
31099 * ipa-cp.c (ipcp_propagate_stage): Do not call
31100 ipa_update_after_lto_read.
31101 * ipa-inline.c (ipa_inline): Likewise.
31103 2017-05-03 Martin Jambor <mjambor@suse.cz>
31105 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
31106 tag. Added a default constructor and a destructor.
31107 (ipa_edge_args_sum_t): New class;
31108 (ipa_edge_args_sum): Declare.
31109 (ipa_edge_args_vector): Remove declaration.
31110 (IPA_EDGE_REF): Use ipa_edge_args_sum.
31111 (ipa_free_edge_args_substructures): Remove declaration.
31112 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
31113 (ipa_edge_args_info_available_for_edge_p): Likewise.
31114 * ipa-prop.c (ipa_edge_args_vector): Removed.
31115 (edge_removal_hook_holder): Likewise.
31116 (edge_duplication_hook_holder): Likewise.
31117 (ipa_edge_args_sum): New variable.
31118 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
31119 ipa_edge_args_vector.
31120 (ipa_free_edge_args_substructures): Likewise.
31121 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
31122 ipa_edge_args_vector.
31123 (ipa_edge_removal_hook): Turned into method
31124 ipa_edge_args_sum_t::remove.
31125 (ipa_edge_duplication_hook): Turned into method
31126 ipa_edge_args_sum_t::duplicate.
31127 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
31128 registering edge hooks.
31129 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
31130 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
31131 ipa_edge_args_sum instead of ipa_edge_args_vector.
31132 * ipa-profile.c (ipa_profile): Likewise.
31134 2017-05-03 Martin Jambor <mjambor@suse.cz>
31136 * symbol-summary.h (function_summary): New method exists.
31137 (function_summary::symtab_removal): Deallocate through release.
31138 (call_summary): New class.
31139 (gt_ggc_mx): New overload.
31140 (gt_pch_nx): Likewise.
31141 (gt_pch_nx): Likewise.
31143 2017-05-03 Jeff Law <law@redhat.com>
31145 PR tree-optimization/78496
31146 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
31147 from simplify_cond_using_ranges. Split off code to walk
31148 backwards through casts into ...
31149 (simplify_cond_using_ranges_2): New function.
31150 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
31151 (execute_vrp): After identifying jump threads, call
31152 simplify_cond_using_ranges_2.
31154 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
31157 * ipa-inline.h (inline_summary): Add ctor.
31158 (create_ggc): Do not use ggc_cleared_alloc.
31160 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
31161 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31163 * gcc.c (handle_braces): Support escaping in switch matching
31165 * doc/invoke.texi (Spec Files): Document it.
31166 Remove superfluous @code markup in items.
31168 2017-05-03 David Malcolm <dmalcolm@redhat.com>
31170 * diagnostic-show-locus.c (struct column_range): New struct.
31171 (get_affected_columns): New function.
31172 (get_printed_columns): New function.
31173 (struct correction): New struct.
31174 (correction::ensure_capacity): New function.
31175 (correction::ensure_terminated): New function.
31176 (struct line_corrections): New struct.
31177 (line_corrections::~line_corrections): New dtor.
31178 (line_corrections::add_hint): New function.
31179 (layout::print_trailing_fixits): Reimplement in terms of the new
31181 (selftest::test_overlapped_fixit_printing): New function.
31182 (selftest::diagnostic_show_locus_c_tests): Call it.
31184 2017-05-03 Nathan Sidwell <nathan@acm.org>
31186 Canonicalize canonical type hashing
31187 * tree.h (type_hash_canon_hash): Declare.
31188 * tree.c (type_hash_list, attribute_hash_list): Move into
31189 type_hash_canon_hash.
31190 (build_type_attribute_qual_variant): Break out hash code calc into
31191 type_hash_canon_hash.
31192 (type_hash_canon_hash): New. Generic type hash computation.
31193 (build_range_type_1, build_array_type_1, build_function_type,
31194 build_method_type_directly, build_offset_type, build_complex_type,
31195 make_vector_type): Call it.
31197 2017-05-03 Richard Biener <rguenther@suse.de>
31199 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
31200 When all DRs have unknown misaligned do not always peel
31201 when there is a store but apply the same costing model as if
31202 there were only loads.
31204 2017-05-03 Richard Biener <rguenther@suse.de>
31207 PR tree-optimization/80492
31208 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31209 compare_base_decls returning dont-know properly.
31211 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
31213 * config/arm/iterators.md (CCSI): New mode iterator.
31214 (arch): New mode attribute.
31215 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
31216 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
31217 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
31218 code iterator for success result mode.
31219 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
31220 the corresponding new insn generators.
31222 2017-05-03 Bin Cheng <bin.cheng@arm.com>
31225 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31226 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31228 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
31230 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
31231 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
31232 (DDR_A): Wrap DDR argument in brackets.
31233 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
31234 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
31235 (DDR_REVERSED_P): Likewise.
31237 2017-05-03 Jakub Jelinek <jakub@redhat.com>
31239 PR tree-optimization/79472
31240 * tree-switch-conversion.c (struct switch_conv_info): Add
31241 contiguous_range and default_case_nonstandard fields.
31242 (collect_switch_conv_info): Compute contiguous_range and
31243 default_case_nonstandard fields, don't clear final_bb if
31244 contiguous_range and only the default case doesn't have the required
31246 (check_all_empty_except_final): Set default_case_nonstandard instead
31247 of failing if contiguous_range and the default case doesn't have empty
31249 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
31250 and only the default case doesn't have the required constants. Skip
31252 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
31253 if default_case_nonstandard.
31254 (build_constructors): Build constant 1 just once. Assert that default
31255 values aren't inserted in between cases if contiguous_range. Skip
31257 (build_arrays): Skip virtual phis.
31258 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
31259 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
31260 Handle virtual phis.
31261 (gen_inbound_check): Handle default_case_nonstandard case.
31262 (process_switch): Adjust check_final_bb caller. Call
31263 gather_default_values with the first non-default case instead of
31264 default case if default_case_nonstandard.
31266 2017-05-02 Nathan Sidwell <nathan@acm.org>
31268 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
31269 check. Fix formatting.
31271 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
31273 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
31274 errors when comparing specialized and unspecialized times.
31276 2017-05-02 David Malcolm <dmalcolm@redhat.com>
31278 * diagnostic-show-locus.c
31279 (layout::should_print_annotation_line_p): Make private.
31280 (layout::print_annotation_line): Make private.
31281 (layout::annotation_line_showed_range_p): Make private.
31282 (layout::show_ruler): Make private.
31283 (layout::print_source_line): Make private. Pass in line and
31284 line_width, rather than calling location_get_source_line. Drop
31286 (layout::print_leading_fixits): New method.
31287 (layout::print_any_fixits): Rename to...
31288 (layout::print_trailing_fixits): ...this, and make private.
31289 Don't print newline fixits.
31290 (diagnostic_show_locus): Move logic for printing one row into...
31291 (layout::print_line): ...this new function. Move the
31292 location_get_source_line call and error-handling from
31293 print_source_line to here. Call print_leading_fixits, and rename
31294 print_any_fixits to print_trailing_fixits.
31295 (selftest::test_fixit_insert_containing_newline): Update now that
31296 newlines are partially supported.
31297 (selftest::test_fixit_insert_containing_newline_2): New test.
31298 (selftest::test_fixit_replace_containing_newline): Update comments.
31299 (selftest::diagnostic_show_locus_c_tests): Call the new test.
31300 * edit-context.c (class added_line): New class.
31301 (class edited_line): Describe newline handling in comment.
31302 (edited_line::actually_edited_p): New method.
31303 (edited_line::print_content): Delete redundant decl.
31304 (edited_line::m_predecessors): New field.
31305 (edited_file::print_content): Call edited_line::print_content.
31306 (edited_file::print_diff): Update to support newlines.
31307 (edited_file::print_diff_hunk): Likewise.
31308 (edited_file::print_run_of_changed_lines): New function.
31309 (edited_file::print_diff_line): Convert to...
31310 (print_diff_line): ...this.
31311 (edited_file::get_effective_line_count): New function.
31312 (edited_line::edited_line): Initialize new field m_predecessors.
31313 (edited_line::~edited_line): Clean up m_predecessors.
31314 (edited_line::apply_fixit): Handle newlines.
31315 (edited_line::get_effective_line_count): New function.
31316 (edited_line::print_content): New function.
31317 (edited_line::print_diff_lines): New function.
31318 (selftest::test_applying_fixits_insert_containing_newline): New
31320 (selftest::test_applying_fixits_replace_containing_newline): New
31322 (selftest::insert_line): New function.
31323 (selftest::test_applying_fixits_multiple_lines): Add example of
31325 (selftest::edit_context_c_tests): Call the new tests.
31327 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31329 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
31330 parameter cand. Update dump information.
31331 (get_computation_cost): Update uses.
31333 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31335 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
31336 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
31337 (get_computation_at, rewrite_use_address): Update use of
31338 get_computation_aff.
31340 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31342 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
31343 (get_computation): Delete.
31344 (get_computation_cost): Implement like get_computation_cost_at.
31345 Use get_computation_at.
31346 (get_computation_cost_at): Delete.
31347 (rewrite_use_nonlinear_expr): Use get_computation_at.
31348 (rewrite_use_compare, remove_unused_ivs): Ditto.
31350 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31352 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
31354 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31356 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
31357 (ivopts_global_cost_for_size): Rename parameter and update uses.
31358 (iv_ca_recount_cost): Update uses.
31359 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
31360 candidates seperately in n_invs and n_cands.
31361 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
31363 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31365 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
31366 (find_inv_vars_cb): New.
31367 (find_depends): Renamed to ...
31368 (find_inv_vars): ... this.
31369 (add_candidate_1, force_var_cost): Call find_inv_vars.
31370 (split_address_cost, determine_group_iv_cost_cond): Ditto.
31372 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31374 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
31375 inv_vars. Add inv_exprs.
31376 (struct iv_cand): Rename depends_on to inv_vars.
31377 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
31378 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
31379 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
31380 (dump_cand): Dump inv_vars.
31381 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
31382 (record_invariant, find_depends, add_candidate_1): Ditto.
31383 (set_group_iv_cost, force_var_cost): Ditto.
31384 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
31385 (get_computation_cost_at, get_computation_cost): Ditto.
31386 (determine_group_iv_cost_generic): Ditto.
31387 (determine_group_iv_cost_address): Ditto.
31388 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
31389 (determine_group_iv_costs): Ditto.
31390 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
31391 (iv_ca_set_remove_invariants): Renamed to ...
31392 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
31393 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
31394 (iv_ca_set_add_invariants): Renamed to ...
31395 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
31396 (iv_ca_set_cp): Use iv_ca_set_add_invs.
31397 (iv_ca_has_deps): Support inv_vars and inv_exprs.
31398 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
31399 (create_new_ivs): Remove useless dump.
31401 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31403 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
31405 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
31406 (iv_ca_set_no_cp, create_new_iv): Ditto.
31408 2017-05-02 Bin Cheng <bin.cheng@arm.com>
31410 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
31412 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31414 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
31415 function tree_check2.
31417 2017-05-02 Martin Liska <mliska@suse.cz>
31419 * doc/gcov.texi: Add missing preposition.
31420 * gcov.c (function_info::function_info): Properly fill up
31421 all member variables.
31423 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31425 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
31427 2017-05-02 Tamar Christina <tamar.christina@arm.com>
31429 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
31431 2017-05-02 Martin Liska <mliska@suse.cz>
31434 * lto-streamer-in.c (lto_read_tree_1): Remove
31435 LTO_STREAMER_DEBUG.
31436 * lto-streamer.c (struct tree_hash_entry): Likewise.
31437 (struct tree_entry_hasher): Likewise.
31438 (tree_entry_hasher::hash): Likewise.
31439 (tree_entry_hasher::equal): Likewise.
31440 (lto_streamer_init): Likewise.
31441 (lto_orig_address_map): Likewise.
31442 (lto_orig_address_get): Likewise.
31443 (lto_orig_address_remove): Likewise.
31444 * lto-streamer.h: Likewise.
31445 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
31446 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
31448 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
31450 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
31451 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
31452 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
31453 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
31454 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
31455 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
31456 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
31457 (mm_maskz_sub_ss): New intrinsics.
31458 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31459 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
31460 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
31461 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
31462 (__builtin_ia32_subss_mask_round): New builtins.
31463 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
31464 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
31465 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
31467 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
31468 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
31470 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
31473 2017-05-02 Martin Jambor <mjambor@suse.cz>
31475 PR tree-optimization/78687
31476 * tree-sra.c (access): New field parent.
31477 (process_subtree_disqualification): New function.
31478 (disqualify_candidate): Call it.
31479 (build_accesses_from_assign): Reset write flag if creating an
31481 (build_access_subtree): Fill in parent field and also prpagate
31482 down grp_write flag.
31483 (create_artificial_child_access): New parameter set_grp_write, set
31484 grp_write to its value.
31485 (propagate_subaccesses_across_link): Also propagate grp_write flag
31487 (propagate_all_subaccesses): Push the closest parent back to work
31488 queue if add_access_to_work_queue returned true.
31490 2017-05-02 Richard Biener <rguenther@suse.de>
31492 * common.opt (fstrict-overflow): Alias negative to fwrapv.
31493 * doc/invoke.texi (fstrict-overflow): Remove all traces of
31494 -fstrict-overflow documentation.
31495 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
31496 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
31497 flag_strict_overflow.
31498 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
31499 * lto-opts.c (lto_write_options): Do not stream it.
31500 * lto-wrapper.c (merge_and_complain): Do not handle it.
31501 * opts.c (default_options_table): Do not set -fstrict-overflow.
31502 (finish_options): Likewise do not clear it when sanitizing.
31503 * simplify-rtx.c (simplify_const_relational_operation): Do not
31504 test flag_strict_overflow.
31506 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31508 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
31509 using enabled attribute.
31510 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
31511 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
31512 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
31513 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
31514 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
31515 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
31516 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
31517 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
31518 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
31519 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
31521 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
31523 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
31525 2017-05-02 Richard Biener <rguenther@suse.de>
31527 PR tree-optimization/80591
31529 2017-04-10 Richard Biener <rguenther@suse.de>
31531 * tree-ssa-structalias.c (find_func_aliases): Properly handle
31534 2017-05-02 Richard Biener <rguenther@suse.de>
31536 PR tree-optimization/80549
31537 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
31538 (cleanup_tree_cfg_noloop): Create forwarders to known loop
31539 headers if they do not have a preheader.
31541 2017-05-02 Martin Liska <mliska@suse.cz>
31544 * common.opt: Fix typo.
31545 * doc/invoke.texi: Likewise.
31547 2017-05-01 Jan Beulich <jbeulich@suse.com>
31549 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
31550 swapping, add (x,x,m,x,n) alternative.
31552 2017-05-01 Nathan Sidwell <nathan@acm.org>
31554 * calls.c (combine_pending_stack_adjustment_and_call): Remove
31555 unnecessary unadjusted_alignment check.
31557 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
31560 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
31562 * gimplify.c (gimplify_cilk_detach): New function.
31563 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
31564 * tree-core.h: Document EXPR_CILK_SPAWN.
31565 * tree.h (EXPR_CILK_SPAWN): Define.
31567 2017-05-01 David Malcolm <dmalcolm@redhat.com>
31569 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
31570 to use new fixit_hint representation, using the "replace" logic.
31571 (get_line_span_for_fixit_hint): Likewise.
31572 (layout::print_any_fixits): Likewise.
31573 (selftest::test_one_liner_many_fixits): Rename to...
31574 (selftest::test_one_liner_many_fixits_1): ...this, and update
31575 comment and expected output to reflect that the multiple fix-it
31576 hints are now consolidated into one insertion.
31577 (selftest::test_one_liner_many_fixits_2): New test.
31578 (selftest::test_diagnostic_show_locus_one_liner): Update for
31580 (selftest::test_fixit_consolidation): Update for fix-it API
31582 * diagnostic.c (print_parseable_fixits): Likewise.
31583 * edit-context.c (edited_line::m_line_events): Convert from
31584 auto_vec <line_event *> to auto_vec <line_event>.
31585 (class line_event): Convert from abstract base class to a concrete
31586 class, taking over the role of replace_event.
31587 (class insert_event): Delete.
31588 (class replace_event): Rename to class line_event. Convert to
31590 (edit_context::add_fixits): Reimplement.
31591 (edit_context::apply_insert): Delete.
31592 (edit_context::apply_replace): Rename to...
31593 (edit_context::apply_fixit): ...this. Convert to half-open range.
31594 (edited_file::apply_insert): Delete.
31595 (edited_file::apply_replace): Rename to...
31596 (edited_file::apply_fixit): ...this.
31597 (edited_line::~edited_line): Drop deletion of events.
31598 (edited_line::apply_insert): Delete.
31599 (edited_line::apply_replace): Rename to...
31600 (edited_line::apply_fixit): ...this. Convert to half-open range.
31601 Update for change to type of m_line_events.
31602 * edit-context.h (edit_context::apply_insert): Delete.
31603 (edit_context::apply_replace): Rename to...
31604 (edit_context::apply_fixit): ...this.
31606 2017-05-01 Martin Sebor <msebor@redhat.com>
31608 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
31611 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
31614 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
31615 __get_cpuid_max returns 0.
31616 (__get_cpuid_count): Ditto.
31618 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
31620 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
31621 replacement expression is another instance of one of its arguments.
31623 2017-05-01 Jakub Jelinek <jakub@redhat.com>
31626 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
31627 check for stack push/pop autoinc.
31628 * config/i386/i386.c (ix86_agi_dependent): Return false
31629 if the only reason why modified_in_p returned true is that
31630 addr is SP based and set_insn is a push or pop.
31632 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
31634 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
31637 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
31640 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
31641 (account_size_time): Use two predicates - exec_pred and
31643 (evaluate_conditions_for_known_args): Compute both clause and
31645 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
31646 (inline_summary_t::duplicate): Update.
31647 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
31649 (compute_inline_parameters): Likewise.
31650 (estimate_edge_size_and_time): Update caluclation of time.
31651 (estimate_node_size_and_time): Compute both time and nonspecialized
31653 (estimate_ipcp_clone_size_and_time): Update.
31654 (inline_merge_summary): Update.
31655 (do_estimate_edge_time): Update.
31656 (do_estimate_edge_size): Update.
31657 (do_estimate_edge_hints): Update.
31658 (inline_read_section, inline_write_summary): Stream both new predicates.
31659 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
31661 (compute_inlined_call_time): Cleanup.
31662 (big_speedup_p): Update.
31663 (edge_badness): Update.
31664 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
31665 (size_time_entry): Replace predicate by exec_predicate and
31666 nonconst_predicate.
31667 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
31668 (estimate_edge_time): Return also nonspec_time.
31669 (reset_edge_growth_cache): Update.
31671 2017-04-29 Jakub Jelinek <jakub@redhat.com>
31673 PR rtl-optimization/80491
31674 * ifcvt.c (noce_process_if_block): When looking for x setter
31675 with missing else_bb, don't check only the insn right before
31676 cond_earliest, but look for the last insn that x is modified in
31677 within the same bb.
31679 PR rtl-optimization/80491
31680 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
31682 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
31684 PR tree-optimization/80487
31685 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
31687 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31689 PR tree-optimization/79697
31690 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
31691 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
31692 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
31694 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
31695 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
31697 2017-04-28 Martin Sebor <msebor@redhat.com>
31699 PR tree-optimization/80523
31700 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
31701 (init_target_to_host_charmap, target_to_host, target_strtol10): New
31703 (maybe_warn, format_directive, parse_directive): Use new functions.
31704 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
31706 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
31708 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
31710 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
31712 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
31713 target_header_dir): Set correctly.
31714 * configure: Regenerated.
31715 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31716 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31717 instead of SYSTEM_HEADER_DIR.
31719 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
31721 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
31722 (estimate_local_effects): Likewise.
31723 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
31724 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
31725 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
31726 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
31727 do_estimate_edge_time, estimate_edge_time): Likewise.
31728 * ipa-inline-analysis.c (estimate_node_size_and_time,
31729 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
31730 (estimate_time_after_inlining): Remove.
31732 2017-04-28 Martin Liska <mliska@suse.cz>
31734 * doc/gcov.texi: Enhance documentation of gcov.
31736 2017-04-28 Martin Liska <mliska@suse.cz>
31738 * doc/gcov.texi: Sort options in alphabetic order.
31739 * doc/gcov-dump.texi: Likewise.
31740 * doc/gcov-tool.texi: Likewise.
31741 * gcov.c (print_usage): Likewise.
31742 * gcov-dump.c (print_usage): Likewise.
31743 * gcov-tool.c (print_merge_usage_message): Likewise.
31744 (print_rewrite_usage_message): Likewise.
31745 (print_overlap_usage_message): Likewise.
31747 2017-04-28 Martin Liska <mliska@suse.cz>
31749 PR gcov-profile/53915
31750 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
31752 2017-04-28 Martin Liska <mliska@suse.cz>
31754 PR gcov-profile/79891
31755 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
31756 is marked by compiler as living on a line.
31757 (get_cycles_count): Remove usage of the union.
31758 (output_intermediate_file): Likewise.
31759 (find_source): Fix GNU coding style.
31760 (accumulate_line_counts): Remove old non-all block mode.
31761 (output_lines): Remove usage of the union.
31762 * profile.c (output_location): Include all BBs, even if
31763 belonging to a same line (and file) as a previous BB.
31765 2017-04-28 Martin Liska <mliska@suse.cz>
31767 * gcov.c (process_args): Handle new argument 'w'.
31768 (read_graph_file): Assign ID to BBs.
31769 (output_branch_count): Display BB # if verbose flag is set.
31770 (output_lines): Likewise for arcs.
31771 (print_usage): Add '--verbose' option help.
31772 * doc/gcov.texi: Document --verbose (-w) option.
31774 2017-04-28 Martin Liska <mliska@suse.cz>
31776 * gcov.c (struct block_location_info): New struct.
31777 (process_file): Fill up the new structure.
31778 (read_graph_file): Replace usage of encoding by the newly added
31780 (add_line_counts): Likewise.
31781 (accumulate_line_counts): Remove usage of the union.
31782 (function_info::function_info): New function.
31783 (function_info::~function_info): Likewise.
31784 (process_file): Call delete instead of release_function.
31785 (release_function): Release the function.
31786 (release_structures): Call delete instead of release_function.
31787 (solve_flow_graph): Replace usage of num_blocks.
31788 (find_exception_blocks): Likewise.
31789 (output_lines): Fix GNU coding style.
31791 2017-04-28 Martin Liska <mliska@suse.cz>
31794 * coverage.c (coverage_remove_note_file): New function.
31795 * coverage.h: Declare the function.
31796 * toplev.c (finalize): Clean if an error has been seen.
31798 2017-04-28 Martin Liska <mliska@suse.cz>
31800 PR gcov-profile/80031
31801 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
31802 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
31803 * gcov.c (read_graph_file): Read just number of blocks.
31804 * profile.c (branch_prob): Do not stream 0 flags per a basic
31807 2017-04-28 Martin Liska <mliska@suse.cz>
31809 * gcov-dump.c (tag_*): Add new argument to declarations.
31810 (dump_gcov_file): Likewise.
31811 (tag_blocks): Add and use new argument depth.
31812 (tag_arcs): Likewise.
31813 (tag_lines): Likewise.
31814 (tag_counters): Likewise.
31815 (tag_summary): Likewise.
31816 (dump_working_sets): Use depth to do a proper indentation.
31818 2017-04-28 Jakub Jelinek <jakub@redhat.com>
31821 * cgraph.h (symtab_node::debug_symtab): No longer inline.
31822 * symtab.c (symtab_node::debug_symtab): Move definition here.
31824 2017-04-28 Richard Biener <rguenther@suse.de>
31826 * lto-streamer.h (LTO_major_version): Bump to 7.
31828 2017-04-28 Richard Biener <rguenther@suse.de>
31830 * tree-vrp.c (assert_info): New struct.
31831 (add_assert_info): New helper.
31832 (register_edge_assert_for_2): Refactor to add asserts to a vector
31834 (register_edge_assert_for_1): Likewise.
31835 (register_edge_assert_for): Likewise.
31836 (finish_register_edge_assert_for): New helper actually registering
31837 asserts where live on edge.
31838 (find_conditional_asserts): Adjust.
31839 (find_switch_asserts): Likewise.
31840 (evrp_dom_walker::try_find_new_range): Generalize.
31841 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
31843 2017-04-27 Marek Polacek <polacek@redhat.com>
31846 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
31847 arg10 and arg11 to itype.
31849 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
31851 * doc/extend.texi (Object Size Checking): Improve grammar.
31853 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
31856 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
31857 that the logic for permitting reciprocal estimates matches that
31860 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31863 * tree.c (type_cache_hasher::equal): Only compare
31864 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
31865 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
31866 non-aggregate element types.
31867 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
31868 about the flag on ARRAY_TYPEs in the comment, formatting fix.
31870 2017-04-27 Richard Biener <rguenther@suse.de>
31872 PR middle-end/80533
31873 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
31874 stripping ARRAY_REFs from MEM_EXPR make sure we're not
31875 keeping a reference to a trailing array.
31877 2017-04-27 Richard Biener <rguenther@suse.de>
31879 PR middle-end/80539
31880 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
31881 being in loop-closed SSA form conservatively.
31882 (chrec_fold_multiply_poly_poly): Likewise.
31884 2017-04-27 Tamar Christina <tamar.christina@arm.com>
31886 PR middle-end/79665
31887 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
31888 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
31890 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31893 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
31894 (aarch64_function_arg_alignment): Return unsigned int again, but still
31895 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
31896 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
31897 Don't emit -Wpsabi note.
31898 (aarch64_function_arg_boundary): Likewise.
31899 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
31902 2017-04-26 Nathan Sidwell <nathan@acm.org>
31904 * tree.h (crc32_unsigned_n): Declare.
31905 (crc32_unsigned, crc32_unsigned): Make inline.
31906 * tree.c (crc32_unsigned_bits): Replace with ...
31907 (crc32_unsigned_n): ... this.
31908 (crc32_unsigned, crc32_byte): Remove.
31909 (crc32_string): Remove unnecessary braces.
31911 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
31913 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
31914 * ipa-inline-analysis.c (MAX_TIME): Remove.
31915 (account_size_time): Use sreal for time.
31916 (dump_inline_summary): Update.
31917 (estimate_function_body_sizes): Update.
31918 (estimate_edge_size_and_time): Update.
31919 (estimate_calls_size_and_time): Update.
31920 (estimate_node_size_and_time): Update.
31921 (inline_merge_summary): Update.
31922 (inline_update_overall_summary): Update.
31923 (estimate_time_after_inlining): Update.
31924 (inline_read_section): Update.
31925 (inline_write_summary): Update.
31926 * ipa-inline.c (compute_uninlined_call_time): Update.
31927 (compute_inlined_call_time): Update.
31928 (recursive_inlining): Update.
31929 (inline_small_functions): Update.
31930 (dump_overall_stats): Update.
31931 * ipa-inline.h: Include sreal.h.
31932 (size_time_entry): Turn time to sreal.
31933 (inline_summary): Turn self_time nad time to sreal.
31935 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
31937 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
31939 (sreal::stream_out, sreal::stream_in): New.
31940 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
31942 2017-04-25 Jakub Jelinek <jakub@redhat.com>
31944 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
31947 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
31950 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
31952 (dimode_scalar_chain::compute_convert_gain): Ditto.
31953 (dimode_scalar_chain::make_vector_copies): Ditto.
31954 (dimode_scalar_chain::convert_reg): Ditto.
31955 (dimode_scalar_chain::convert_insn): Ditto.
31956 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
31957 (VI248_AVX512BW_1): New mode iterator.
31958 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
31959 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
31962 2017-04-25 Martin Sebor <msebor@redhat.com>
31964 PR tree-optimization/80497
31965 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
31966 constants are representable in HOST_WIDE_INT.
31967 (parse_directive): Ditto.
31969 2017-04-25 Martin Sebor <msebor@redhat.com>
31972 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
31973 (new_zero_array): Adjust signature.
31974 (dom_info::dom_init): Used unsigned rather that size_t.
31975 (dom_info::dom_info): Same.
31977 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
31978 Jakub Jelinek <jakub@redhat.com>
31981 * config/arm/arm.c: Include gimple.h.
31982 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
31983 returns negative, increment ncrn only if it returned positive.
31984 (arm_needs_doubleword_align): Return int instead of bool,
31985 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
31986 members, but if there is any such non-FIELD_DECL
31987 > PARM_BOUNDARY aligned decl, return -1 instead of false.
31988 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
31989 returns negative, increment nregs only if it returned positive.
31990 (arm_setup_incoming_varargs): Likewise.
31991 (arm_function_arg_boundary): Emit -Wpsabi note if
31992 arm_needs_doubleword_align returns negative, return
31993 DOUBLEWORD_ALIGNMENT only if it returned positive.
31995 2017-04-25 Marek Polacek <polacek@redhat.com>
31998 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
31999 first argument to type.
32001 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
32004 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
32005 type checks to test for compatibility instead of equality.
32007 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
32008 Jakub Jelinek <jakub@redhat.com>
32011 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
32013 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
32014 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
32015 the alignment computation, but return their maximum in warn_alignment.
32016 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
32017 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
32019 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
32020 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
32023 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32025 * config/arc/simdext.md (dmpyh): Fix typo.
32027 2017-04-25 Richard Biener <rguenther@suse.de>
32029 PR tree-optimization/80492
32030 * alias.c (compare_base_decls): Handle registers with asm
32031 specification conservatively.
32032 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
32033 compare_base_decls returning dont-know properly.
32035 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32037 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
32038 (legitimate_offset_address_p): New function.
32039 (arc_legitimate_address_p): Use above function.
32041 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32043 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
32045 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32047 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
32048 ACCH registers whenever they are available.
32050 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32052 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
32053 double regs fix when not used.
32055 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32057 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
32059 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
32060 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
32062 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32064 * config/arc/arc.c (arc_output_addsi): Check for h-register class
32065 when emitting short ADD instructions.
32067 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
32069 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
32071 (cmpsi_cc_c_insn): Likewise.
32072 (cbranchsi4_scratch): Compute proper instruction length using
32073 compact_hreg_operand.
32074 * config/arc/predicates.md (compact_hreg_operand): New predicate.
32076 2017-04-25 Richard Biener <rguenther@suse.de>
32078 PR middle-end/80509
32079 * passes.c (pass_manager::pass_manager): Initialize
32080 m_name_to_pass_map.
32082 2017-04-25 Richard Biener <rguenther@suse.de>
32084 PR tree-optimization/79201
32085 * tree-ssa-sink.c (statement_sink_location): Handle calls.
32087 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32090 * config/s390/vector.md: Split MEM->GPR vector moves for
32091 non-s_operand addresses.
32093 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32096 * config/s390/predicates.md (reload_const_wide_int_operand): New
32098 * config/s390/s390.md ("movti"): Remove d/P alternative.
32099 ("movti_bigconst"): New pattern definition.
32101 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32104 * s390-protos.h (s390_expand_cs_hqi): Removed.
32105 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
32106 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
32107 modes as well as CCZ1mode and CCZmode.
32108 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
32109 signature of s390_emit_compare_and_swap.
32110 (s390_expand_cs_hqi): Likewise, make static.
32111 (s390_expand_cs_tdsi): Generate an explicit compare before trying
32112 compare-and-swap, in some cases.
32113 (s390_expand_cs): Wrapper function.
32114 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
32116 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
32117 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
32118 patterns for small and large integers. Forbid symref memory operands.
32119 Move expander to s390.c. Require cc register.
32120 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
32121 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
32122 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
32123 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
32124 symref memory operands. Remove CC mode and call s390_match_ccmode
32126 ("atomic_exchange<mode>"): Allow and implement all integer modes.
32128 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32130 * config/s390/s390.md (define_peephole2): New peephole to help
32131 combining the load-and-test pattern with volatile memory.
32133 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
32135 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
32136 with CCZmode for TARGET_Z196.
32138 2017-04-25 Jakub Jelinek <jakub@redhat.com>
32140 PR rtl-optimization/80501
32141 * combine.c (make_compound_operation_int): Set subreg_code to SET
32142 even for AND with mask of the sign bit of mode.
32144 PR rtl-optimization/80500
32145 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
32146 sum's initial value.
32148 2017-04-25 Julian Brown <julian@codesourcery.com>
32149 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32151 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
32153 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
32155 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
32157 2017-04-25 Julian Brown <julian@codesourcery.com>
32158 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32160 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
32161 (thunderx2t99_sha): New Reservation.
32163 2017-04-25 Julian Brown <julian@codesourcery.com>
32164 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32166 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
32167 type for 1-element load.
32169 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
32171 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
32173 2017-04-24 Martin Jambor <mjambor@suse.cz>
32175 PR tree-optimization/80293
32176 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
32177 char arrays not totally scalarizable if it is false.
32178 (analyze_all_variable_accesses): Pass correct value in the new
32179 parameter. Add a statistics counter.
32181 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
32183 PR middle-end/79931
32184 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
32186 2017-04-24 Richard Biener <rguenther@suse.de>
32188 PR tree-optimization/80494
32189 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
32190 out for complex types.
32192 2017-04-24 Richard Biener <rguenther@suse.de>
32194 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
32195 * tree-ssa-sccvn.c (print_scc): Print SCC size.
32196 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
32197 (DFS): Adjust and never fail.
32198 (sccvn_dom_walker::fail): Remove.
32199 (sccvn_dom_walker::before_dom_children): Adjust.
32200 (run_scc_vn): Likewise and never fail.
32201 * tree-ssa-pre.c (pass_pre::execute): Adjust.
32202 (pass_fre::execute): Likewise.
32204 2017-04-24 Richard Biener <rguenther@suse.de>
32206 PR tree-optimization/79725
32207 * tree-ssa-sink.c (statement_sink_location): Return whether
32208 failure reason was zero uses. Move that check later.
32209 (sink_code_in_bb): Deal with zero uses by removing the stmt
32212 2017-04-24 Richard Biener <rguenther@suse.de>
32215 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
32216 pointer-based references.
32218 2017-04-24 Richard Biener <rguenther@suse.de>
32221 * pass_manager.h (pass_manager::operator new): Remove.
32222 (pass_manager::operator delete): Likewise.
32223 * passes.c (pass_manager::operator new): Remove.
32224 (pass_manager::operator delete): Likewise.
32225 (pass_manager::pass_manager): Zero individual pass members.
32227 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
32230 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
32231 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
32232 Check "XEXP (src, 1)" operand here.
32233 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
32234 Check "XEXP (src, 1)" operand here.
32235 (dimode_scalar_chain::make_vector_copies): Detect count register
32236 of a shift instruction. Zero extend count register from QImode
32237 to DImode to satisfy vector shift pattern count operand predicate.
32238 Substitute vector shift count operand with a DImode copy.
32239 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
32242 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
32244 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
32245 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32246 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
32247 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
32248 (UNSPEC_NOREX_MEM): Remove definition.
32250 2017-04-21 Richard Biener <rguenther@suse.de>
32252 PR tree-optimization/79547
32253 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
32254 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
32255 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
32256 without any constraints.
32258 2017-04-21 Richard Biener <rguenther@suse.de>
32260 PR tree-optimization/78847
32261 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
32263 2017-04-21 Richard Biener <rguenther@suse.de>
32265 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
32266 (build_distinct_type_copy): Likewise.
32267 (build_variant_type_copy): Likewise.
32268 * tree.c (build_qualified_type): Pass down mem-stat info.
32269 (build_distinct_type_copy): Likewise.
32270 (build_variant_type_copy): Likewise.
32272 2017-04-21 Richard Biener <rguenther@suse.de>
32274 PR tree-optimization/80237
32275 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
32277 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
32278 for a simplified result.
32280 2016-04-21 Richard Biener <rguenther@suse.de>
32282 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
32283 sth as strict as a simple_iv but a chrec without symbols and an
32284 operand defined in the loop we are peeling (and not some subloop).
32285 (propagate_constants_for_unrolling): Propagate all constants.
32287 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32290 * config/i386/i386.c (print_reg): Remove assert for disalowed
32291 regno values, call output_operand_lossage instead.
32293 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
32296 * config/i386/constraints.md (Yc): New register constraint.
32297 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
32298 Use Yc constraint for alternative 2 of operand 0. Remove
32299 preferred_for_speed attribute.
32301 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32303 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
32304 lastprivate clauses in SIMT case.
32306 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
32308 * doc/invoke.texi (-Wextra-semi): Document new warning option.
32310 2017-04-20 Richard Biener <rguenther@suse.de>
32312 PR tree-optimization/57796
32313 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
32314 as N scalar stores.
32315 (vect_model_load_cost): Cost gathers as N scalar loads.
32317 2017-04-20 Richard Biener <rguenther@suse.de>
32319 * ggc-page.c (ggc_allocated_p): Rename to ...
32320 (safe_lookup_page_table_entry): ... this and return the lookup
32322 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
32324 2017-04-20 Richard Biener <rguenther@suse.de>
32326 PR tree-optimization/80453
32327 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
32328 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
32329 from the conditions.
32330 (vn_phi_eq): Pass them down.
32331 (vn_phi_lookup): Record them.
32332 (vn_phi_insert): Likewise.
32334 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
32336 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
32337 uninitialized variable warning to avoid buffer overrun.
32339 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
32342 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
32343 is suppressed for '{ 0 }' in C.
32345 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32347 * BASE-VER: Set to 8.0.0.
32349 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
32351 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
32352 priority .init_array and .fini_array section with SECTION_NOTYPE
32355 2017-04-20 Jakub Jelinek <jakub@redhat.com>
32357 PR middle-end/80423
32358 * tree.h (build_array_type): Add typeless_storage default argument.
32359 * tree.c (type_cache_hasher::equal): Also compare
32360 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
32361 (build_array_type): Add typeless_storage argument, set
32362 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
32364 (build_nonshared_array_type): Adjust build_array_type_1 caller.
32365 (build_array_type): Likewise. Add typeless_storage argument.
32367 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32368 Jakub Jelinek <jakub@redhat.com>
32370 PR tree-optimization/80426
32371 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
32372 operation on symbolic operands, also compute the overflow for the
32373 invariant part when the operation degenerates into a negation.
32375 2017-04-19 Jakub Jelinek <jakub@redhat.com>
32378 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
32379 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
32382 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
32384 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
32387 * config/avr/avr.c (tree.h): Include it.
32388 (cgraph.h): Include it.
32389 (avr_encode_section_info): Don't warn for uninitialized progmem
32390 variable if it's just an alias.
32392 2017-04-19 Richard Biener <rguenther@suse.de>
32395 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
32396 when needed by AutoPGO.
32398 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
32401 * doc/lto.texi: Remove an extra 'that'.
32403 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
32405 PR rtl-optimization/80429
32406 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
32407 are only used in debug insns.
32409 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
32410 Vladimir Makarov <vmakarov@redhat.com>
32412 * config/sparc/predicates.md (input_operand): Add comment. Return
32413 true for any memory operand when LRA is in progress.
32414 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
32416 2017-04-18 Jeff Law <law@redhat.com>
32419 * mips.md ({return,simple_return}_internal): Do not overwrite
32422 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32424 PR tree-optimization/80443
32425 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
32426 instead of adding 1, subtract -1 and similarly instead of subtracting
32429 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
32431 PR rtl-optimization/80357
32432 * haifa-sched.c (tmp_bitmap): New variable.
32433 (model_recompute): Handle duplicate use records.
32434 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
32435 (free_global_sched_pressure_data): Free it.
32437 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
32440 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
32441 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
32442 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
32443 instead of SYSTEM_HEADER_DIR.
32445 2017-04-18 Jeff Law <law@redhat.com>
32447 PR middle-end/80422
32448 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
32449 predecessors after walking up the insn chain.
32451 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32454 * dwarf2out.c (modified_type_die): Try harder not to emit internal
32455 sizetype type into debug info.
32457 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32460 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
32461 unneeded test for TARGET_UPPER_REGS_SF.
32462 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
32464 2017-04-18 Jakub Jelinek <jakub@redhat.com>
32467 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
32468 instead of gsi_after_labels.
32470 2017-04-18 Jeff Law <law@redhat.com>
32472 * regcprop.c (maybe_mode_change): Avoid creating copies of the
32476 2017-04-13 Jeff Law <law@redhat.com>
32477 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32478 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32480 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
32483 * config/avr/avr.c (intl.h): Include it.
32484 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
32486 2017-04-18 Martin Liska <mliska@suse.cz>
32488 PR gcov-profile/78783
32489 * gcov-tool.c (gcov_output_files): Validate that destination
32490 file is either removed by the tool or by a user.
32492 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32493 Guy Benyei <guybe@mellanox.com>
32495 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
32496 block, and do not negate it, the stored id is already negative.
32498 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
32500 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
32502 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
32505 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
32506 masks of options that should be turned off if the VSX vector
32507 options are turned off.
32508 (OTHER_P8_VECTOR_MASKS): Likewise.
32509 (OTHER_VSX_VECTOR_MASKS): Likewise.
32510 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
32511 rs6000_disable_incompatible_switches to validate no type switches
32513 (rs6000_incompatible_switch): New function to disallow turning on
32514 other vector options if -mno-vsx, -mno-power8-vector, or
32515 -mno-power9-vector are specified.
32517 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32519 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
32521 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32523 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
32524 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
32525 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
32526 (ARG_POINTER_CFA_OFFSET): Likewise.
32528 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
32530 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
32531 conditions to take advantage of various optimizations.
32533 2017-04-13 Jeff Law <law@redhat.com>
32535 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
32536 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
32537 (zero_extendsidi2_dext): Likewise.
32539 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32542 * fold-const.c (fold_ternary_loc): Revert
32543 use op0 instead of fold_convert_loc (loc, type, arg0) part of
32546 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
32548 PR rtl-optimization/80343
32549 * lra-remat.c (update_scratch_ops): Assign original hard reg to
32550 new scratch pseudo.
32552 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
32555 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
32556 to ubsan_encode_value.
32558 2017-04-13 Jeff Law <law@redhat.com>
32560 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
32561 appearing in DEBUG_INSNs.
32563 2017-04-13 Martin Liska <mliska@suse.cz>
32565 PR gcov-profile/80413
32566 * gcov-io.c (gcov_write_string): Copy to buffer just when
32567 allocated size is greater than zero.
32569 2017-04-13 Jakub Jelinek <jakub@redhat.com>
32572 * dwarf2out.c (decls_for_scope): Ignore declarations of
32573 current_function_decl in BLOCK_NONLOCALIZED_VARS.
32575 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
32578 * ipa-visibility.c (non_local_p): Fix typos.
32579 (localize_node): When localizing symbol in same comdat group,
32580 dissolve the group only when we know external symbols are going
32582 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
32584 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32586 PR tree-optimization/79390
32587 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
32588 order does not result in usable sequence, retry with reversed operand
32594 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
32595 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
32596 op0 instead of fold_convert_loc (loc, type, arg0).
32598 2017-04-12 Jeff Law <law@redhat.com>
32600 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
32601 has a delay slot in the generated code.
32603 * config/cris/cris.md (cris_preferred_reload_class): Return
32604 GENNONACR_REGS rather than GENERAL_REGS.
32606 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32609 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
32610 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
32611 signedness of the result type.
32613 2017-04-12 Richard Biener <rguenther@suse.de>
32614 Jeff Law <law@redhat.com>
32616 PR tree-optimization/80359
32617 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
32618 trim stores to TARGET_MEM_REFs.
32620 2017-04-12 Richard Biener <rguenther@suse.de>
32622 PR tree-optimization/79390
32623 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
32624 threading case even more.
32626 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
32629 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
32630 for quad_address_p for TImode, instead of just not indexed_address.
32632 2017-04-12 Richard Biener <rguenther@suse.de>
32633 Bernd Edlinger <bernd.edlinger@hotmail.de>
32635 PR middle-end/79671
32636 * alias.c (component_uses_parent_alias_set_from): Handle
32637 TYPE_TYPELESS_STORAGE.
32638 (get_alias_set): Likewise.
32639 * tree-core.h (tree_type_common): Add typeless_storage flag.
32640 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
32641 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
32642 for types containing members with TYPE_TYPELESS_STORAGE.
32643 (place_field): Likewise.
32644 (layout_type): Likewise for ARRAY_TYPE.
32645 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
32646 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
32647 TYPE_TYPELESS_STORAGE.
32648 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
32650 2017-04-12 Jakub Jelinek <jakub@redhat.com>
32653 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
32654 first argument to type.
32656 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32660 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
32661 CONST0_RTX (mode) rather than const0_rtx where appropriate.
32662 (rs6000_expand_binop_builtin): Likewise.
32663 (rs6000_expand_ternop_builtin): Likewise; also add missing
32664 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
32665 vshasigma built-ins.
32666 * doc/extend.texi: Document that vec_xxpermdi's third argument
32667 must be a constant.
32669 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
32671 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
32672 Use shift_const cost parameter when calculating gain of STV shifts.
32674 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
32676 PR rtl-optimization/70478
32677 * lra-constraints.c (process_alt_operands): Check memory for
32678 disfavoring memory insn operand.
32680 2017-04-11 Jakub Jelinek <jakub@redhat.com>
32682 PR middle-end/80100
32683 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
32684 left shift in unsigned HOST_WIDE_INT type.
32686 PR rtl-optimization/80385
32687 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
32688 (not (neg X)) into (plus X -1) for complex or non-integral modes.
32691 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
32692 if they have any depend clauses.
32694 2017-04-11 Martin Liska <mliska@suse.cz>
32697 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
32698 * ipa-split.c (split_function): Create a local comdat symbol
32699 if caller is in a comdat group.
32701 2017-04-11 Martin Liska <mliska@suse.cz>
32704 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
32707 2017-04-11 Martin Sebor <msebor@redhat.com>
32709 PR middle-end/80364
32710 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
32711 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
32713 (directive::set_width, directive::set_precision, format_character):
32715 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
32718 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
32721 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
32722 conflict, set target->arch_name instead of target->cpu_name.
32724 2017-04-11 Richard Biener <rguenther@suse.de>
32726 PR tree-optimization/80374
32727 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
32728 build_zero_cst, remove fold_convertible_p check again.
32730 2017-04-11 Martin Liska <mliska@suse.cz>
32733 * ubsan.c (instrument_object_size): Do not instrument register
32736 2017-04-11 Jakub Jelinek <jakub@redhat.com>
32739 * config/i386/i386-builtin-types.def
32740 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
32741 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
32742 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
32743 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
32744 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
32745 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
32746 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
32747 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
32748 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
32749 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
32750 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
32751 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
32752 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
32753 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
32754 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
32755 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
32756 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
32757 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
32758 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
32759 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
32760 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
32761 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
32762 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
32763 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
32764 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
32765 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
32766 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
32767 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
32768 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
32769 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
32770 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
32771 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
32772 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
32773 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
32774 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
32775 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
32776 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
32777 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
32778 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
32779 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
32780 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
32781 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
32782 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
32783 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
32784 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
32785 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
32787 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
32788 flag to second_arg_count, handle 4 argument function type _COUNT
32789 aliases, handle second_arg_count on second argument rather than last.
32791 2017-04-10 Jeff Law <law@redhat.com>
32793 PR tree-optimization/80374
32794 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
32795 record anything if we can not convert integer_zero_node to the
32798 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
32801 * config/rs6000/rs6000.c (rs6000_option_override_internal):
32802 Enhance special handling given to the TARGET_P9_MINMAX option in
32803 relation to certain other options.
32805 2017-04-10 Bin Cheng <bin.cheng@arm.com>
32807 PR tree-optimization/80153
32808 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
32809 remove POINTER_PLUS_EXPR's base part directly, rather than through
32812 2017-04-10 Richard Biener <rguenther@suse.de>
32813 Bin Cheng <bin.cheng@arm.com>
32815 PR tree-optimization/80153
32816 * tree-affine.c (aff_combination_to_tree): Get base pointer from
32817 the first element of pointer type aff_tree. Build result expr in
32819 (add_elt_to_tree): Convert to type unconditionally. Remove other
32820 fold_convert calls.
32821 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
32822 (rewrite_use_nonlinear_expr): Check invariant using iv information.
32824 2017-04-10 Richard Biener <rguenther@suse.de>
32826 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32829 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
32831 PR rtl-optimization/70478
32832 * lra-constraints.c (curr_small_class_check): New.
32833 (update_and_check_small_class_inputs): New.
32834 (process_alt_operands): Update curr_small_class_check. Disfavor
32835 alternative insn memory operands. Check available regs for small
32838 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32841 * config/mips/mips.opt (-mvirt): Update description.
32842 * doc/invoke.texi (-mvirt): Likewise.
32844 2017-04-10 Richard Biener <rguenther@suse.de>
32846 PR middle-end/80362
32847 * fold-const.c (fold_binary_loc): Look at unstripped ops when
32848 looking for NEGATE_EXPR in -A / -B to A / B folding.
32850 2017-04-10 Martin Liska <mliska@suse.cz>
32852 PR gcov-profile/80224
32853 * gcov.c (print_usage): Fix usage string.
32854 (get_gcov_intermediate_filename): Remove.
32855 (output_gcov_file): Use both for normal and intermediate format.
32856 (generate_results): Do not initialize special file for
32857 intermediate format.
32859 2017-04-10 Richard Biener <rguenther@suse.de>
32861 PR tree-optimization/80304
32862 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
32865 2017-04-10 Nathan Sidwell <nathan@acm.org>
32868 * config/rs6000/rs6000.c (rs6000_vector_type): New.
32869 (rs6000_init_builtins): Use it.
32871 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32873 * config/arm/arm.md (<mrc>): Add mode to SET source.
32874 (<mrrc>): Likewise.
32876 2017-04-10 Richard Biener <rguenther@suse.de>
32878 PR middle-end/80344
32879 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
32881 2017-04-10 Jakub Jelinek <jakub@redhat.com>
32884 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
32885 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
32886 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
32887 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
32888 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
32889 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
32890 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
32891 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
32892 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
32893 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
32894 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
32895 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
32896 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
32897 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
32898 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
32899 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
32900 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
32901 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
32902 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
32903 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
32904 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
32905 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
32906 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
32908 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
32910 PR rtl-optimization/70478
32911 * lra-constraints.c: Reverse the last patch.
32913 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
32915 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
32916 Add comment for WCHAR_T.
32918 2017-04-08 Martin Liska <mliska@suse.cz>
32921 2017-04-07 Martin Liska <mliska@suse.cz>
32924 * ipa-split.c (split_function): Add function part to a same comdat
32927 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32930 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
32932 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
32934 * rs6000/rs6000.c (vec_load_pendulum): Rename...
32935 (vec_pairing): ...to this.
32936 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
32937 (rs6000_sched_init): Adjust for name change.
32938 (struct rs6000_sched_context): Likewise.
32939 (rs6000_init_sched_context): Likewise.
32940 (rs6000_set_sched_context): Likewise.
32942 2017-04-07 Jakub Jelinek <jakub@redhat.com>
32948 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
32950 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
32951 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
32952 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
32954 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
32956 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
32958 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
32960 PR rtl-optimization/70703
32961 * ira-color.c (update_conflict_hard_regno_costs): Use
32962 int64_t instead of HOST_WIDE_INT.
32964 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
32966 PR rtl-optimization/70478
32967 * lra-constraints.c (process_alt_operands): Disfavor alternative
32968 insn memory operands.
32970 2017-04-07 Jeff Law <law@redhat.com>
32972 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
32973 CALL and NOTE_INSN_CALL_ARG_LOCATION.
32975 2017-04-07 Martin Liska <mliska@suse.cz>
32978 * config/aarch64/aarch64.c (aarch64_process_target_attr):
32979 Show error message instead of an ICE.
32981 2017-04-07 Martin Liska <mliska@suse.cz>
32984 * ipa-split.c (split_function): Add function part to a same comdat
32987 2017-04-07 Richard Biener <rguenther@suse.de>
32989 PR middle-end/80341
32990 * tree.c (get_unwidened): Also handle ! for_type case for
32992 * convert.c (do_narrow): Split out from ...
32993 (convert_to_integer_1): ... here. Do not pass final truncation
32994 type to get_unwidened for TRUNC_DIV_EXPR.
32996 2017-04-07 Richard Biener <rguenther@suse.de>
32998 * tree-affine.c (wide_int_ext_for_comb): Take type rather
33000 (aff_combination_const): Adjust.
33001 (aff_combination_scale): Likewise.
33002 (aff_combination_add_elt): Likewise.
33003 (aff_combination_add_cst): Likewise.
33004 (aff_combination_convert): Likewise.
33005 (add_elt_to_tree): Likewise. Remove unused argument.
33006 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
33008 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
33010 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
33012 * config/arm/arm.c (arm_default_short_enums): Use
33013 ARM_DEFAULT_SHORT_ENUMS.
33014 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
33016 2017-04-06 Jakub Jelinek <jakub@redhat.com>
33019 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
33020 members with redundant out-of-class redeclaration.
33022 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33025 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
33026 * config/i386/i386.md (*zero_extendsidi2):
33027 Add (?*x,*x) and (?*v,*v) alternatives.
33029 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
33032 * config/i386/i386.c (ix86_expand_builtin)
33033 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
33034 mode from insn data. Convert operands to insn operand mode.
33035 Copy operands that don't satisfy insn predicate to a register.
33037 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
33039 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
33042 2017-04-06 Richard Biener <rguenther@suse.de>
33044 PR tree-optimization/80334
33045 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
33046 preserve alignment of accesses.
33048 2017-04-06 Richard Biener <rguenther@suse.de>
33050 PR tree-optimization/80262
33051 * tree-sra.c (build_ref_for_offset): Preserve address-space
33053 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
33054 Drop useless address-space information on MEM_REF offsets.
33056 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
33058 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
33060 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33062 PR rtl-optimization/70703
33063 * ira-color.c (update_conflict_hard_regno_costs): Use
33064 HOST_WIDE_INT instead of long.
33066 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
33069 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
33070 not defined for x86_64 target. Add -mmmx target option when __SSE2__
33072 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
33073 for x86_64 target. Handle -m3dnowa option.
33075 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
33077 PR rtl-optimization/70703
33078 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
33079 (update_conflict_hard_regno_costs): Use long instead of unsigned
33080 arithmetic for cost calculation.
33082 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33083 Bernd Edlinger <bernd.edlinger@hotmail.de>
33086 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
33089 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
33092 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
33093 ptr_mode with Pmode throughout.
33094 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
33095 into probe_stack_range and use DImode.
33097 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
33100 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
33101 call_eh_return is true.
33103 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33105 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
33106 Initialize last_match_fntype_index.
33108 2017-04-05 Jakub Jelinek <jakub@redhat.com>
33111 * tree-nvr.c: Include internal-fn.h.
33112 (pass_return_slot::execute): Ignore internal calls without
33115 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33116 Richard Biener <rguenther@suse.de>
33119 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
33120 captures used multiple times, except for the last use.
33121 * generic-match-head.c: Include gimplify.h.
33123 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33125 PR tree-optimization/79390
33126 * target.h (struct noce_if_info): Declare.
33127 * targhooks.h (default_noce_conversion_profitable_p): Declare.
33128 * target.def (noce_conversion_profitable_p): New target hook.
33129 * ifcvt.h (struct noce_if_info): New type, moved from ...
33130 * ifcvt.c (struct noce_if_info): ... here.
33131 (noce_conversion_profitable_p): Renamed to ...
33132 (default_noce_conversion_profitable_p): ... this. No longer
33134 (noce_try_store_flag_constants, noce_try_addcc,
33135 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
33136 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
33137 instead of noce_conversion_profitable_p.
33138 * config/i386/i386.c: Include ifcvt.h.
33139 (ix86_option_override_internal): Don't override
33140 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
33141 (ix86_noce_conversion_profitable_p): New function.
33142 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
33143 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
33144 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
33145 * doc/tm.texi: Regenerated.
33147 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33149 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
33152 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
33155 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
33156 instructions for small multiply cores.
33158 2017-04-04 Jeff Law <law@redhat.com>
33160 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
33162 (mips_expand_vec_perm_const): Initialize elements in orig_perm
33163 that are not set by the loop over the elements.
33165 2017-04-04 Jakub Jelinek <jakub@redhat.com>
33168 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
33169 int mode, convert_modes it to mode as unsigned, otherwise use
33170 lowpart_subreg to mode rather than SImode.
33171 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
33172 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
33173 Use DImode instead of SImode for the shift count operand.
33174 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
33177 2017-04-04 Richard Biener <rguenther@suse.de>
33179 PR middle-end/80281
33180 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
33181 arithmetic done for the negate or the plus. Simplify.
33182 (A - (-B) -> A + B): Likewise.
33183 * fold-const.c (split_tree): Make sure to not negate pointers.
33185 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
33187 PR rtl-optimization/60818
33188 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
33189 a compare of comparisons with the thing compared if this results
33190 in a different machine mode.
33192 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
33194 * alias.c (base_alias_check): Fix typo in comment.
33195 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
33196 * cgraphunit.c (symbol_table::compile): Likewise.
33197 * collect2.c (maybe_run_lto_and_relink): Likewise.
33198 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
33199 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
33200 * config/avr/avr.c (avr_map_op_t): Likewise.
33201 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
33202 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
33203 * config/epiphany/epiphany.md (movcc): Likewise.
33204 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
33205 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
33207 * config/mips/mips.c (mips_save_restore_reg): Likewise.
33208 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
33209 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
33210 * config/sh/sh.c (sh_rtx_costs): Likewise.
33211 * fold-const.c (fold_truth_andor): Likewise.
33212 * genautomata.c (collapse_flag): Likewise.
33213 * gengtype.h (struct type::u::s): Likewise.
33214 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
33215 * input.c (FORMAT_AMOUNT): Likewise.
33216 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
33217 (known_aggs_to_agg_replacement_list): Likewise.
33218 * ipa-inline-analysis.c: Likewise.
33219 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
33220 * ipa-polymorphic-call.c
33221 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
33222 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
33223 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
33225 * modulo-sched.c (apply_reg_moves): Likewise.
33226 * omp-expand.c (build_omp_regions_1): Likewise.
33227 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
33228 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
33229 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
33230 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
33231 * value-prof.c: Likewise.
33232 * var-tracking.c (val_reset): Likewise.
33234 2017-04-03 Richard Biener <rguenther@suse.de>
33236 PR tree-optimization/80275
33237 * fold-const.c (split_address_to_core_and_offset): Handle
33240 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
33242 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
33243 descriptors is at least equal to that of functions.
33245 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33247 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
33249 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
33252 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
33253 (mov<IMOD4:mode>): New expander.
33254 (*mov<IMOD4:mode>_internal): New insn and split pattern.
33256 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
33258 PR rtl-optimization/79405
33259 * fwprop.c (propagations_left): New variable.
33260 (forward_propagate_into): Decrement it.
33261 (fwprop_init): Initialize it.
33262 (fw_prop): If the variable has reached zero, stop propagating.
33263 (fwprop_addr): Ditto.
33265 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33268 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
33269 a FUNCTION_DECL, pass it as decl instead of origin to
33272 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
33274 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
33277 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
33280 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
33281 TARGET_VSX_SMALL_INTEGER.
33283 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33285 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
33286 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
33288 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
33290 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
33291 extraction from odd-numbered MSA register.
33293 2017-03-31 Jakub Jelinek <jakub@redhat.com>
33295 PR middle-end/80173
33296 * expmed.c (store_bit_field_1): Don't attempt to create
33297 a word subreg out of hard registers wider than word if they
33298 have HARD_REGNO_NREGS of 1 for their mode.
33300 PR middle-end/80163
33301 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
33302 conversions to integer types wider than word and pointer.
33305 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
33306 (rtx_equal_for_cselib_p): Pass 0 to it.
33307 * cselib.c (cselib_hasher::equal): Likewise.
33308 (rtx_equal_for_cselib_1): Add depth argument. If depth
33309 is 128, don't look up VALUE locs and punt. Increment
33310 depth in recursive calls when walking VALUE locs.
33312 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
33314 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
33315 (make_gcov_file_name): Use the canonical path name for generating
33317 (read_line): Fix handling of files with ascii null bytes.
33319 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
33321 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
33322 to initialise a vector register instead
33323 of using a const_int.
33325 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33327 PR translation/80189
33328 * gimplify.c (omp_default_clause): Use %qs instead of %s in
33329 diagnostic messages.
33331 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
33334 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
33335 (dfp_diex_<mode>): Update mode of operand 1.
33336 * doc/extend.texi (dxex, dxexq): Document change to return type.
33337 (diex, diexq): Document change to argument type.
33339 2017-03-30 Martin Jambor <mjambor@suse.cz>
33342 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
33343 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
33344 it reflects the signature changes performed at the callee side.
33345 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
33346 to cgraph_build_function_type_skip_args.
33347 (build_function_decl_skip_args): Adjust call to the above function.
33349 2017-03-30 Jakub Jelinek <jakub@redhat.com>
33352 * config/i386/sse.md
33353 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
33354 register as dest whenever it is a MEM not rtx_equal_p to the
33355 corresponding dup operand, and when forcing into reg move the
33356 reg into the memory afterwards.
33357 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
33358 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
33359 for the force_reg mode.
33360 (avx512vl_vextractf128<mode>): Use register as dest either
33361 always when a MEM, or when it is a MEM not rtx_equal_p to the
33362 corresponding dup operand, or even not when it is a CONST_VECTOR
33363 depending on the mode and lo vs. hi.
33364 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
33366 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
33367 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
33368 Likewise. Require that operands[2] is even.
33369 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
33370 Remove extraneous parens. Require that operands[2] is a multiple
33372 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
33373 operands[0] is a MEM if <mask_applied>, the predicates/constraints
33374 disallow memory then.
33376 2017-03-30 Richard Biener <rguenther@suse.de>
33378 PR tree-optimization/77498
33379 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
33380 to non-constants over backedges.
33382 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
33384 PR rtl-optimization/80233
33385 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
33386 as last_combined_insn. Do not test for BARRIER_P separately.
33388 2017-03-29 Andreas Schwab <schwab@suse.de>
33391 * calls.c (prepare_call_address): Convert funexp to Pmode before
33392 copying to temp reg.
33394 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33396 PR tree-optimization/80158
33397 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
33398 Handle possible future case of more than one alternate
33400 (replace_rhs_if_not_dup): Likewise.
33401 (replace_one_candidate): Likewise.
33403 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
33405 PR rtl-optimization/80193
33406 * ira.c (ira): Do not check allocation for LRA.
33408 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
33410 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
33411 (nvptx_output_simt_exit): Declare.
33412 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
33413 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
33414 (init_softstack_frame): Move initialization of crtl->is_leaf to...
33415 (nvptx_declare_function_name): ...here. Emit declaration of local
33416 memory space buffer for omp_simt_enter insn.
33417 (nvptx_output_unisimt_switch): New.
33418 (nvptx_output_softstack_switch): New.
33419 (nvptx_output_simt_enter): New.
33420 (nvptx_output_simt_exit): New.
33421 * config/nvptx/nvptx.h (struct machine_function): New fields
33422 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
33423 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
33424 (UNSPECV_SIMT_EXIT): Ditto.
33425 (omp_simt_enter_insn): New insn.
33426 (omp_simt_enter): New expansion.
33427 (omp_simt_exit): New insn.
33428 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
33430 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
33431 (expand_GOMP_SIMT_ENTER_ALLOC): New.
33432 (expand_GOMP_SIMT_EXIT): New.
33433 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
33434 (GOMP_SIMT_ENTER_ALLOC): Ditto.
33435 (GOMP_SIMT_EXIT): Ditto.
33436 * target-insns.def (omp_simt_enter): New insn.
33437 (omp_simt_exit): Ditto.
33438 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
33440 (lower_rec_simd_input_clauses): Implement SIMT privatization.
33441 (lower_rec_input_clauses): Likewise.
33442 (lower_lastprivate_clauses): Handle SIMT privatization.
33444 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
33445 (ompdevlow_adjust_simt_enter): New.
33446 (find_simtpriv_var_op): New.
33447 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
33448 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
33450 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
33451 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
33452 (copy_decl_for_dup_finish): Ditto.
33454 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
33456 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
33459 * config/i386/i386.c (ix86_option_override_internal): Always
33460 allow -mpreferred-stack-boundary=3 for 64-bit targets.
33462 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33464 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
33466 2017-03-28 Bin Cheng <bin.cheng@arm.com>
33468 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
33469 mark new edge's irreducible flag accordign to it.
33470 (vect_do_peeling): Check loop preheader edge's irreducible flag
33471 and pass it to function slpeel_add_loop_guard.
33473 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
33475 PR tree-optimization/80218
33476 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33477 Update block frequencies and counts.
33479 2017-03-28 Richard Biener <rguenther@suse.de>
33481 PR tree-optimization/78644
33482 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
33483 of a simplification result we may not use it at all.
33485 2017-03-28 Richard Biener <rguenther@suse.de>
33488 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
33489 without arguments, generate default definition of a SSA name.
33491 2017-03-28 Richard Biener <rguenther@suse.de>
33493 PR middle-end/80222
33494 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
33495 TYPE_REF_CAN_ALIAS_ALL references.
33496 * fold-const.c (fold_indirect_ref_1): Likewise.
33498 2017-03-28 Martin Liska <mliska@suse.cz>
33501 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
33502 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
33504 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33505 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
33507 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
33508 (EXTRA_SPECS): Define.
33509 (SUBTARGET_EXTRA_SPECS): Likewise.
33510 (SUBTARGET_CPP_SPEC): Likewise.
33511 * config/arc/elf.h (EXTRA_SPECS): Renamed to
33512 SUBTARGET_EXTRA_SPECS.
33513 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
33515 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
33517 * config/arc/simdext.md (vst64_insn): Update pattern.
33518 (vld32wh_insn): Likewise.
33519 (vld32wl_insn): Likewise.
33520 (vld64_insn): Likewise.
33521 (vld32_insn): Likewise.
33523 2017-03-28 Marek Polacek <polacek@redhat.com>
33526 * fold-const.c (fold_comparison): Use protected_set_expr_location
33527 instead of SET_EXPR_LOCATION.
33529 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
33531 * tree.c (add_expr): Avoid name lookup warning.
33533 2017-03-27 Jeff Law <law@redhat.com>
33535 PR tree-optimization/80216
33536 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
33537 function name. Limit recursion depth.
33538 (record_temporary_equivalences): Corresponding changes.
33540 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
33542 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
33545 2017-03-27 Jakub Jelinek <jakub@redhat.com>
33548 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
33550 * cfgcleanup.c (reg_note_cfa_p): New array.
33551 (insns_have_identical_cfa_notes): New function.
33552 (old_insns_match_p): Don't cross-jump in between /f
33553 and non-/f instructions. If both i1 and i2 are frame related,
33554 verify all CFA notes, their order and content.
33556 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
33559 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
33560 HImode and SImode with zero extend to DImode to one insn.
33561 (bswap<mode>2_extenddi): Likewise.
33562 (bswapsi2_extenddi): Likewise.
33563 (bswaphi2_extendsi): Likewise.
33564 (bswaphi2): Combine bswap HImode and SImode into one insn.
33565 Separate memory insns from swapping register.
33566 (bswapsi2): Likewise.
33567 (bswap<mode>2): Likewise.
33568 (bswaphi2_internal): Delete, no longer used.
33569 (bswapsi2_internal): Likewise.
33570 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
33571 store, and gpr<-gpr swap insns.
33572 (bswap<mode>2_store): Likewise.
33573 (bswaphi2_reg): Register only splitter, combine with the splitter.
33574 (bswaphi2 splitter): Likewise.
33575 (bswapsi2_reg): Likewise.
33576 (bswapsi2 splitter): Likewise.
33577 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
33578 the insns into load, store, and register/register insns.
33579 (bswapdi2_ldbrx): Likewise.
33580 (bswapdi2_load): Likewise.
33581 (bswapdi2_store): Likewise.
33582 (bswapdi2_reg): Likewise.
33584 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
33586 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
33587 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
33589 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
33592 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
33594 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33595 special handling for target option conflicts between dform
33596 options (-mpower9-dform, -mpower9-dform-vector,
33597 -mpower9-dform-scalar) and -mno-direct-move.
33599 2017-03-27 Richard Biener <rguenther@suse.de>
33601 PR tree-optimization/80181
33602 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
33604 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33606 * config/arc/predicates.md (move_double_src_operand): Replace the
33607 call to move_double_src_operand with a call to address_operand.
33609 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33611 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
33612 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
33613 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
33615 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33617 * config/arc/predicates.md (long_immediate_loadstore_operand):
33618 Consider scaled addresses cases.
33620 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
33622 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
33623 restored when in interrupt.
33624 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
33625 doesn't have delay slot.
33627 2017-03-27 Richard Biener <rguenther@suse.de>
33630 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
33631 inlined thunk clones.
33633 2017-03-27 Jakub Jelinek <jakub@redhat.com>
33636 * asan.c (instrument_derefs): Copy over last operand from
33637 original COMPONENT_REF to the new COMPONENT_REF with
33638 DECL_BIT_FIELD_REPRESENTATIVE.
33639 * ubsan.c (instrument_object_size): Likewise.
33641 2017-03-27 Richard Biener <rguenther@suse.de>
33643 PR tree-optimization/80170
33644 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
33645 sure DR/SCEV didnt fold in constants we do not see when looking
33646 at the reference base alignment.
33648 2017-03-27 Richard Biener <rguenther@suse.de>
33650 PR middle-end/80171
33651 * gimple-fold.c (fold_ctor_reference): Properly guard against
33652 NULL return value from canonicalize_constructor_val.
33654 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
33657 * config/i386/i386.c (ix86_expand_builtin)
33658 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
33659 flags reg setting and flags reg using instructions.
33660 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
33661 clobbering instructions to zero extend op2.
33663 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
33665 * doc/install.texi (Configuration) <--with-aix-soname>:
33666 Update link to AIX ld.
33668 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
33670 PR rtl-optimization/80160
33671 PR rtl-optimization/80159
33672 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
33673 reg_alternate_class into account.
33675 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
33678 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
33679 to consider in curr_insn_transform.
33681 2017-03-24 Jakub Jelinek <jakub@redhat.com>
33683 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
33684 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
33685 and emit_mode_inner.
33687 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33689 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
33690 argument to the overloaded builtin variants. Use the new flag to
33691 deprecate certain builtin variants.
33692 * config/s390/s390-builtin-types.def: Add new builtin types.
33693 * config/s390/s390-builtins.h: Support new flags field for
33694 overloaded builtins.
33695 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
33696 (s390_macro_to_expand): Enable vector float data type.
33697 (s390_cpu_cpp_builtins_internal): Indicate support of the new
33698 builtins by incrementing the __VEC__ version number.
33699 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
33701 (s390_resolve_overloaded_builtin): Emit error messages depending
33702 on the builtin flags.
33703 * config/s390/s390.c (s390_expand_builtin): Support additional
33704 flags argument. Change error message to match the messages
33705 emitted in s390-c.c.
33706 * config/s390/s390.md: New UNSPEC_* constants.
33707 (op_type): Add new instruction types.
33708 * config/s390/vecintrin.h: Add new builtins and test data class
33710 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
33711 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
33712 (VEC_INEXACT, VEC_NOINEXACT): New constants.
33713 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
33714 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
33715 ("vec_mergel<mode>"): V_HW -> VEC_HW.
33717 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
33718 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
33719 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
33720 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
33722 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
33723 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
33724 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
33725 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
33727 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
33728 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
33729 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
33730 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
33731 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
33732 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
33733 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
33735 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
33736 ("vec_scatter_element<V_HW_4:mode>_DI")
33737 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
33738 ("vec_fpint<mode>", "vflls")
33739 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
33740 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
33741 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
33742 ("*vec_cmphe<mode>_cc"): ... these.
33744 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
33745 mode constant instead of magic value.
33747 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33749 * config/s390/s390.c (s390_expand_vec_compare): Support other
33750 vector floating point modes than just V2DF.
33751 (s390_expand_vcond): Likewise.
33752 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
33753 (s390_cannot_change_mode_class): Prevent mode changes between TF
33754 and V1TF in vector registers.
33755 * config/s390/s390.md (DF, SF): New mode attributes.
33756 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
33757 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
33758 SFmode support for VRs.
33759 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
33761 (VFT, VF_HW): New mode iterators.
33762 (vw, sdx): New mode attributes.
33763 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
33764 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
33765 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
33766 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
33767 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
33768 also the new vector floating point modes. Renaming to ...
33770 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
33771 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
33772 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
33773 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
33774 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
33775 ("vec_unordered<mode>"): ... these.
33777 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
33778 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
33779 ("*vec_extendv2df"): New insn definitions.
33781 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33783 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
33784 ("mulditi3_2", "*muldi3_sign"): New patterns.
33785 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
33786 rename the pattern definition.
33788 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33790 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
33792 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
33794 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33796 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
33797 instruction if possible.
33798 * config/s390/vector.md (vec_halfnumelts): New mode
33800 ("*vec_vllezlf<mode>"): New pattern.
33802 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33804 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
33805 ("popcountv4si2", "popcountv2di2"): Rename to ...
33806 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
33807 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
33809 ("popcount<mode>2_vxe"): New pattern.
33811 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33813 * common/config/s390/s390-common.c (processor_flags_table): Add
33815 * config.gcc: Add arch12.
33816 * config/s390/driver-native.c (s390_host_detect_local_cpu):
33817 Default to arch12 for unknown CPU model numbers.
33818 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
33819 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
33820 PROCESSOR_max sanity check.
33821 * config/s390/s390-opts.h (enum processor_type): Add
33823 * config/s390/s390.c (processor_table): Add arch12.
33824 (s390_expand_builtin): Add check for B_VXE flag.
33825 (s390_issue_rate): Add PROCESSOR_ARCH12.
33826 (s390_get_sched_attrmask): Likewise.
33827 (s390_get_unit_mask): Likewise.
33828 (s390_sched_score): Enable z13 scheduling for arch12.
33829 (s390_sched_reorder): Likewise.
33830 (s390_sched_variable_issue): Likewise.
33831 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
33833 (s390_tune_attr): Use z13 scheduling also for arch12.
33834 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
33835 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
33836 (TARGET_VXE_P): New macros.
33837 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
33838 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
33839 * config/s390/s390.opt: Add arch12 as processor_type.
33841 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33843 * config/s390/s390.md
33844 ("fixuns_truncdddi2", "fixuns_trunctddi2")
33845 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
33846 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
33848 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
33849 Rename expanders to ...
33851 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
33852 ("fixuns_truncdddi2_emu"): ... these.
33854 ("fixuns_trunc<mode>si2_emu"): New expander.
33856 ("*fixuns_truncdfdi2_z13"): Rename to ...
33857 ("*fixuns_truncdfdi2_vx"): ... this.
33859 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33861 * config/s390/2964.md: Remove the single element vector compare
33862 instructions which are no longer used.
33863 * config/s390/s390.c (s390_select_ccmode): Remove handling of
33865 (s390_canonicalize_comparison): Remove handling of DFmode
33867 (s390_expand_vec_compare_scalar): Remove function.
33868 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
33869 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
33871 ("*cmp<mode>_ccs"): Add wfcdb instruction.
33873 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33875 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
33877 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
33878 will anyway by matched by mov<mode>_64dfp.
33880 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33882 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
33883 vlef/vstef. Add missing operand to vleif.
33885 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33887 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
33888 pair for all vector types with 64 bit elements.
33889 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
33890 * config/s390/vector.md (V_HW_64): ... here.
33891 (V_128_NOSINGLE): New mode iterator.
33892 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
33893 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
33894 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
33895 ("*vec_load_pairv2di"): Change to ...
33896 ("*vec_load_pair<mode>"): ... this one.
33898 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33900 * config/s390/constraints.md: Add comments.
33901 (jKK): Reject element sizes > 8 bytes.
33902 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
33904 * config/s390/s390.md: Add the s_operand checks formerly in
33905 s390_split_ok_p to various splitters where they are still
33907 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
33908 for 128 bit vectors. Plus two splitters.
33910 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33912 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
33915 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33918 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
33919 error if the boundary argument is not constant.
33921 2017-03-24 Jakub Jelinek <jakub@redhat.com>
33923 PR rtl-optimization/80112
33924 * loop-doloop.c (doloop_condition_get): Don't check condition
33925 if cmp isn't SET with IF_THEN_ELSE src.
33927 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33929 PR tree-optimization/80158
33930 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
33931 replacing a candidate statement, also replace it for the
33932 candidate's alternate interpretation.
33933 (replace_rhs_if_not_dup): Likewise.
33934 (replace_one_candidate): Likewise.
33936 2017-03-24 Richard Biener <rguenther@suse.de>
33938 PR tree-optimization/80167
33939 * graphite-isl-ast-to-gimple.c
33940 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
33942 (translate_isl_ast_to_gimple::get_rename): Likewise.
33944 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33946 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
33947 handling of certain combinations of target options, including the
33948 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
33949 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
33951 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33954 * config/arm/arm.md (*load_multiple): Add reload_completed to
33955 matching condition.
33957 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33958 Richard Biener <rguenth@suse.de>
33960 PR tree-optimization/79908
33961 PR tree-optimization/80136
33962 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
33963 been cast away, gimplify_and_add suffices.
33965 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
33967 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
33969 2017-03-23 Richard Biener <rguenther@suse.de>
33971 PR tree-optimization/80032
33972 * gimplify.c (gimple_push_cleanup): Forced unconditional
33973 cleanups still have to go to the conditional_cleanups
33976 2017-03-22 Jakub Jelinek <jakub@redhat.com>
33978 PR tree-optimization/80072
33979 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
33981 (next_operand_entry_id): Change type to unsigned int.
33982 (sort_by_operand_rank): Make sure to return the right return value
33983 even if unsigned fields are bigger than INT_MAX.
33984 (struct oecount): Change cnt and id type to unsigned int.
33985 (oecount_hasher::equal): Formatting fix.
33986 (oecount_cmp): Make sure to return the right return value
33987 even if unsigned fields are bigger than INT_MAX.
33988 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
33991 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
33992 TREE_READONLY on result if writing it more than once.
33995 * doc/invoke.texi (-fsanitize=thread): Document that with
33996 -fnon-call-exceptions atomics are not able to throw
34000 * tsan.c: Include tree-eh.h.
34001 (instrument_builtin_call): Call maybe_clean_eh_stmt or
34002 maybe_clean_or_replace_eh_stmt where needed.
34003 (instrument_memory_accesses): Add cfg_changed argument.
34004 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
34005 if it returned true.
34006 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
34008 PR rtl-optimization/63191
34009 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
34010 wrapper function, moved the whole old content into ...
34011 (ix86_delegitimize_address_1): ... this. New inline function.
34012 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
34013 true as last argument instead of ix86_delegitimize_address.
34015 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34017 * config/aarch64/aarch64.c (generic_branch_cost): Copy
34018 cortexa57_branch_cost.
34020 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
34022 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
34024 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34027 * doc/md.texi (Constraints): Document wA constraint.
34028 * config/rs6000/constraints.md (wA): New.
34029 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
34030 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
34031 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
34032 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
34034 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
34037 * gimplify.c (is_oacc_declared): New function.
34038 (oacc_default_clause): Use it to set default flags for acc declared
34039 variables inside parallel regions.
34040 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
34041 declared variables.
34042 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
34043 declare attribute to any decl as necessary.
34045 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
34048 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
34049 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
34050 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
34051 (arm_arch_lpae): This.
34052 * config/arm/arm.c (arm_arch7ve): Rename into ...
34053 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
34054 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
34057 2017-03-22 Martin Liska <mliska@suse.cz>
34060 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
34061 error message instead of an ICE.
34063 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34065 * doc/extend.texi (6.11 Additional Floating Types): Revise.
34067 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
34069 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34071 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34074 2017-03-21 Martin Sebor <msebor@redhat.com>
34076 * doc/extend.texi: Use "cannot" instead of "can't."
34077 * doc/hostconfig.texi: Same.
34078 * doc/install.texi: Same.
34079 * doc/invoke.texi: Same.
34080 * doc/loop.texi: Same.
34081 * doc/md.texi: Same.
34082 * doc/objc.texi: Same.
34083 * doc/rtl.texi: Same.
34084 * doc/tm.texi: Same.
34085 * doc/tm.texi.in: Same.
34086 * doc/trouble.texi: Same.
34088 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
34091 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
34092 (collect_checksum_attributes): Set it.
34093 (die_checksum_ordered): Use it.
34095 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34097 PR tree-optimization/79908
34098 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
34099 change: For a VA_ARG whose LHS has been cast away, use
34100 force_gimple_operand to construct the side effects.
34102 2017-03-21 David Malcolm <dmalcolm@redhat.com>
34104 PR translation/80001
34105 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
34106 more amenable to translation.
34107 (oacc_loop_auto_partitions): Likewise.
34109 2017-03-21 Marek Polacek <polacek@redhat.com>
34110 Martin Sebor <msebor@redhat.com>
34112 PR tree-optimization/80109
34113 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
34114 on INTEGRAL_TYPE_P.
34116 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34117 Segher Boessenkool <segher@kernel.crashing.org>
34120 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
34121 check reg_used_between_p between insn and one of succ or succ2
34122 depending on if succ is artificial insn not inserted into insn
34125 2017-03-21 Martin Liska <mliska@suse.cz>
34127 PR gcov-profile/80081
34128 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
34129 * doc/gcc.texi: Include gcov-dump stuff.
34130 * doc/gcov-dump.texi: New file.
34132 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
34134 PR rtl-optimization/79150
34135 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
34136 conditional jump, if the jump is the last insn of the loop.
34138 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34139 Richard Biener <rguenth@suse.de>
34141 PR tree-optimization/79908
34142 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
34143 been cast away, use force_gimple_operand to construct the side
34146 2017-03-21 Martin Liska <mliska@suse.cz>
34148 PR libfortran/79956
34149 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
34152 2017-03-21 Brad Spengler <spender@grsecurity.net>
34155 * plugin.c (htab_hash_plugin): New function.
34156 (add_new_plugin): Use it and adjust.
34157 (parse_plugin_arg_opt): Adjust.
34158 (init_one_plugin): Likewise.
34160 2017-03-21 Richard Biener <rguenther@suse.de>
34162 PR tree-optimization/80032
34163 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
34164 if set force the cleanup to happen unconditionally.
34165 (gimplify_target_expr): Push inserted clobbers with force_uncond
34166 to avoid them being removed by control-dependent DCE.
34168 2017-03-21 Richard Biener <rguenther@suse.de>
34170 PR tree-optimization/80122
34171 * tree-inline.c (copy_bb): Do not expans va-arg packs or
34172 va_arg_pack_len when the inlined call stmt requires pack
34174 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
34176 2017-03-21 Jakub Jelinek <jakub@redhat.com>
34179 * tsan.c (instrument_builtin_call): If the memory model argument
34180 is not a constant, assume it is valid.
34183 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
34186 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34188 PR rtl-optimization/79910
34189 * combine.c (can_combine_p): Do not allow combining an I0 or I1
34190 if its dest is used by an insn before I2 (other than the combined
34191 insns themselves, which are properly handled already).
34193 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
34196 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34198 * combine.c (record_used_regs): New static function.
34199 (try_combine): Handle situations where there is an additional
34200 instruction between I2 and I3 which needs to have a LOG_LINK
34204 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34206 * combine.c (try_combine): Delete redundant i1 test. Call
34207 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34209 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34212 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
34213 alternatives 13/14.
34215 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34217 PR tree-optimization/80054
34218 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
34219 the optimization if a PHI or any of its arguments is not dominated
34220 by the candidate's basis. Use gphi* rather than gimple* as
34222 (replace_profitable_candidates): Clean up a gimple* variable that
34223 should be a gphi* variable.
34225 2017-03-20 Martin Sebor <msebor@redhat.com>
34228 * doc/extend.texi (attribute constructor): Document present limitation.
34230 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
34233 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
34234 __POWER9_VECTOR__ #ifdef control, change template definition to
34235 use Power9-specific built-in function.
34236 (vec_any_eq): Likewise.
34237 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
34238 to control outcomes from this test.
34239 (vector_ae_<mode>p): For VEC_F modes, likewise.
34241 2017-03-20 Ian Lance Taylor <iant@google.com>
34243 * config/i386/i386.c (ix86_function_regparm): Save an extra
34244 register for -fsplit-stack with DECL_STATIC_CHAIN.
34246 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34249 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
34250 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
34252 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
34254 * config/riscv/riscv.c (riscv_print_operand): Use "fence
34256 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
34259 2017-03-20 Marek Polacek <polacek@redhat.com>
34262 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
34264 2017-03-20 Richard Biener <rguenther@suse.de>
34266 PR tree-optimization/80113
34267 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
34268 allocate extra SSA name for PHI def.
34269 (add_close_phis_to_outer_loops): Likewise.
34270 (add_close_phis_to_merge_points): Likewise.
34271 (copy_loop_close_phi_args): Likewise.
34272 (copy_cond_phi_nodes): Likewise.
34274 2017-03-20 Martin Liska <mliska@suse.cz>
34276 PR middle-end/79753
34277 * tree-chkp.c (chkp_build_returned_bound): Do not build
34278 returned bounds for a LHS that's not a BOUNDED_P type.
34280 2017-03-20 Martin Liska <mliska@suse.cz>
34284 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
34285 COMPLEX_CST and VECTOR_CST.
34287 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34290 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
34291 target operand. A new splitter adds the clobber statement in case
34292 the target operand is dead anyway.
34294 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
34296 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
34297 to age-old versions of binutils and glibc.
34299 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
34301 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
34303 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34305 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
34307 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
34309 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
34310 requirement for binutils 2.13.
34312 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
34314 * combine.c (try_combine): Delete redundant i1 test. Call
34315 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
34317 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
34319 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
34320 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
34322 <riscv64-*-elf>: Re-arrange section
34323 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
34324 <riscv32-*-linux>: Likewise.
34325 <riscv64-*-elf>: Likewise
34326 <riscv64-*-linux>: Likewise.
34328 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
34331 * aarch64.opt(verbose-cost-dump): Fix typo.
34333 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
34336 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
34337 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
34339 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
34341 * reload.c (find_reloads): When reloading a nonoffsettable address,
34342 use RELOAD_OTHER for it and its address reloads.
34344 PR rtl-optimization/79910
34345 * combine.c (record_used_regs): New static function.
34346 (try_combine): Handle situations where there is an additional
34347 instruction between I2 and I3 which needs to have a LOG_LINK
34350 2017-03-17 Jeff Law <law@redhat.com>
34352 PR tree-optimization/71437
34353 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
34354 conditional in the hash table first.
34355 (vrp_dom_walker::before_dom_children): Extract condition from
34356 ASSERT_EXPR. Record condition, its inverion and any implied
34357 conditions as well.
34359 2017-03-17 Marek Polacek <polacek@redhat.com>
34360 Markus Trippelsdorf <markus@trippelsdorf.de>
34362 PR tree-optimization/80079
34363 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
34366 2017-03-17 Richard Biener <rguenther@suse.de>
34368 PR middle-end/80075
34369 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
34370 Properly verify the LHS before the RHS possibly claims to be
34372 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
34375 2017-03-17 Martin Jambor <mjambor@suse.cz>
34377 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
34378 (List of -O2 options): Likewise.
34379 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
34382 2017-03-17 Tom de Vries <tom@codesourcery.com>
34384 * gcov-dump.c (print_usage): Print bug_report_url.
34386 2017-03-17 Richard Biener <rguenther@suse.de>
34388 PR middle-end/80050
34389 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
34390 (parser::peek): Likewise.
34392 2017-03-17 Richard Biener <rguenther@suse.de>
34394 PR tree-optimization/80048
34395 * sese.c (free_sese_info): Properly release rename_map and
34396 copied_bb_map elements.
34398 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
34400 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
34401 Add linked-list forward and backlinks. Insert on
34402 construction, remove on destruction.
34403 (class pass_store_merging): Add m_stores_head field.
34404 (pass_store_merging::terminate_and_process_all_chains):
34405 Iterate over m_stores_head list.
34406 (pass_store_merging::terminate_all_aliasing_chains):
34408 (pass_store_merging::execute): Check for debug stmts first.
34409 Push new chains onto the m_stores_head stack.
34411 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
34414 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
34415 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
34416 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
34418 2017-03-16 Jeff Law <law@redhat.com>
34420 PR tree-optimization/71437
34421 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
34422 member function. Implementation moved into after_dom_children
34423 member function and into the threader's thread_outgoing_edges
34425 (dom_opt_dom_walker::after_dom_children): Simplify by moving
34426 some code into new thread_outgoing_edges.
34427 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
34428 definition. Simplify marker handling (do it here). Assume we always
34429 have the available expression and the const/copies tables.
34430 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
34432 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
34433 * tree-vrp.c (equiv_stack): No longer file scoped.
34434 (vrp_dom_walker): New class.
34435 (vrp_dom_walker::before_dom_children): New member function.
34436 (vrp_dom_walker::after_dom_children): Likewise.
34437 (identify_jump_threads): Setup domwalker. Use it rather than
34438 walking edges in a random order by hand. Simplify setup/finalization.
34439 (finalize_jump_threads): Remove.
34440 (vrp_finalize): Do not call identify_jump_threads here.
34441 (execute_vrp): Do it here instead and call thread_through_all_blocks
34444 PR tree-optimization/71437
34445 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
34447 (simplify_stmt_for_jump_threading): Add basic_block argument. All
34449 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
34450 (dom_opt_dom_walker::thread_across_edge): Remove
34451 handle_dominating_asserts argument. All callers changed.
34452 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
34453 changes. Remove calls to lhs_of_dominating_assert. Other
34454 uses of handle_dominating_asserts turn into unconditional code
34455 (simplify_control_stmt_condition_1): Likewise.
34456 (simplify_control_stmt_condition): Likewise.
34457 (thread_through_normal_block, thread_across_edge): Likewise.
34458 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
34459 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
34460 object if it is not an SSA_NAME.
34461 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
34462 before calling into the VRP specific simplifiers.
34463 (identify_jump_threads): Remove handle_dominating_asserts
34466 2017-03-16 Jakub Jelinek <jakub@redhat.com>
34469 * tree-diagnostic.c (default_tree_printer): No longer static.
34470 * tree-diagnostic.h (default_tree_printer): New prototype.
34472 2017-03-16 Tamar Christina <tamar.christina@arm.com>
34474 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
34475 Change ins into fmov.
34477 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34479 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
34480 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
34481 Use h_con constraint for operand 1.
34482 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
34483 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
34485 2017-03-15 Jeff Law <law@redhat.com>
34487 PR tree-optimization/71437
34488 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
34489 (record_temporary_equivalences): Use it.
34491 PR tree-optimization/71437
34492 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
34493 tree-ssa-scopedtables.
34494 (lookup_avail_expr, build_and_record_new_cond): Likewise.
34495 (record_conditions, record_cond, vuse_eq): Likewise.
34496 (record_edge_info): Adjust to API tweak of record_conditions.
34497 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
34498 (record_temporary_equivalences, optimize_stmt): Likewise.
34499 (eliminate_redundant_computations): Likewise.
34500 (record_equivalences_from_stmt): Likewise.
34501 * tree-ssa-scopedtables.c: Include options.h and params.h.
34502 (vuse_eq): New function, moved from tree-ssa-dom.c
34503 (build_and_record_new_cond): Likewise.
34504 (record_conditions): Likewise. Accept vector of conditions rather
34505 than edge_equivalence structure for first argument.
34506 for the first argument.
34507 (avail_exprs_stack::lookup_avail_expr): New member function, moved
34508 from tree-ssa-dom.c.
34509 (avail_exprs_stack::record_cond): Likewise.
34510 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
34511 from tree-ssa-dom.c.
34512 (avail_exprs_stack): Add new member functions lookup_avail_expr
34514 (record_conditions): Declare.
34516 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
34519 * lra-constraints.c (process_alt_operands): Increase reject for
34520 reloading an input/output operand.
34522 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
34525 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
34526 insns to convert from signed/unsigned char/short to IEEE 128-bit
34528 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
34530 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
34533 * config/i386/i386.c (ix86_vector_duplicate_value): Create
34534 subreg of inner mode for values already in registers.
34536 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
34538 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
34539 iteration reg is used after the loop.
34541 2017-03-14 Martin Sebor <msebor@redhat.com>
34543 PR tree-optimization/79800
34544 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
34545 precision in negative-positive range.
34546 (format_floating): Call non-const overload with adjusted precision.
34548 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34551 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
34554 2017-03-14 Martin Sebor <msebor@redhat.com>
34556 PR middle-end/80020
34557 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
34558 * builtins.def (aligned_alloc): Use it.
34561 * Makefile.in (GTFILES): Add calls.c.
34562 * calls.c: Include "gt-calls.h".
34564 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
34566 PR rtl-optimization/79728
34567 * regs.h (struct target_regs): New field
34568 x_contains_allocatable_regs_of_mode.
34569 (contains_allocatable_regs_of_mode): New macro.
34570 * reginfo.c (init_reg_sets_1): Initialize it, and change
34571 contains_reg_of_mode so it includes global regs as well.
34572 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
34573 rather than contains_regs_of_mode.
34575 2017-03-14 Martin Liska <mliska@suse.cz>
34577 * doc/invoke.texi: Document options that can't be combined with
34578 -fcheck-pointer-bounds.
34580 2017-03-14 Martin Liska <mliska@suse.cz>
34582 PR middle-end/79831
34583 * doc/invoke.texi (-Wchkp): Document the option.
34585 2017-03-14 Martin Liska <mliska@suse.cz>
34587 * Makefile.in: Install gcov-dump.
34589 2017-03-14 Martin Liska <mliska@suse.cz>
34591 * multiple_target.c (expand_target_clones): Bail out for
34592 an invalid attribute.
34594 2017-03-14 Richard Biener <rguenther@suse.de>
34596 * alias.c (struct alias_set_entry): Pack properly.
34597 * cfgloop.h (struct loop): Likewise.
34598 * cse.c (struct set): Likewise.
34599 * ipa-utils.c (struct searchc_env): Likewise.
34600 * loop-invariant.c (struct invariant): Likewise.
34601 * lra-remat.c (struct cand): Likewise.
34602 * recog.c (struct change_t): Likewise.
34603 * rtl.h (struct address_info): Likewise.
34604 * symbol-summary.h (function_summary): Likewise.
34605 * tree-loop-distribution.c (struct partition): Likewise.
34606 * tree-object-size.c (struct object_size_info): Likewise.
34607 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
34608 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
34609 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
34610 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
34611 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
34612 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
34613 (struct _stmt_vec_info): Likewise.
34615 2017-03-14 Martin Liska <mliska@suse.cz>
34618 * multiple_target.c (create_dispatcher_calls): Check that
34619 a target can create a function dispatcher.
34621 2017-03-14 Martin Liska <mliska@suse.cz>
34624 * multiple_target.c (expand_target_clones): Drop local.local
34625 flag for default implementation.
34627 2017-03-14 Richard Biener <rguenther@suse.de>
34629 PR tree-optimization/80030
34630 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
34632 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
34634 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
34635 gcc_fallthrough() instead of __attribute__((fallthrough));
34637 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
34639 * doc/gcc.texi: Remove "up" link to (DIR).
34640 * doc/gccint.texi: Ditto.
34642 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
34644 * doc/install.texi (Specific) <avr>: Remove reference to
34647 2017-03-13 Jeff Law <law@redhat.com>
34649 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
34650 attribute rather than comments.
34652 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
34653 match_scratch operand is highest.
34655 2017-03-13 Martin Liska <mliska@suse.cz>
34657 PR middle-end/78339
34658 * ipa-pure-const.c (warn_function_noreturn): If the declarations
34659 is a CHKP clone, use original declaration.
34661 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34663 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
34664 (arc_conditional_register_usage): Use a different allocation order
34665 when optimizing for size.
34666 * common/config/arc/arc-common.c (arc_option_optimization_table):
34667 Section anchors default on when optimizing for size.
34669 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34671 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
34673 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34675 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
34676 * config/arc/arc.md (cpu_facility): Add cd variant.
34677 (*movqi_insn): Add code density variant.
34678 (*movhi_insn): Likewise.
34679 (*movqi_insn): Likewise.
34680 (*addsi3_mixed): Likewise.
34681 (subsi3_insn): Likewise.
34683 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34685 * config/arc/arc.md (movsi_cond_exec): Update constraint.
34687 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
34689 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
34690 expressions with MINUS and UNARY ops.
34692 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34695 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
34697 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
34698 between vec_select and vector argument.
34699 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
34700 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
34701 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
34702 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
34703 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
34704 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
34706 2017-03-13 Richard Biener <rguenther@suse.de>
34709 * params.def (vect-max-peeling-for-alignment): Fix typo.
34711 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34713 * doc/install.texi (Specific) <mips-*-*>: Remove description of
34714 issue that only occurred with binutils below 2.18.
34716 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34718 * doc/install.texi (Specific) <cris-axis-elf>: No longer
34719 refer to binutils 2.11/2.12 minimum.
34721 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
34723 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
34724 ftp.kernel.org and simplify binutils requirement.
34726 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
34728 * doc/invoke.texi (Warning Options): Fix spelling of link-time
34730 (Optimize Options): Ditto. Also remove redundancy.
34732 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34734 PR translation/79848
34735 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
34737 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
34738 to G_ to avoid double translation.
34740 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34742 PR translation/79923
34743 * auto-profile.c (get_combined_location): Convert leading
34744 character of diagnostics to lower case and remove trailing period.
34745 (read_profile): Likewise for various diagnostics.
34746 * config/arm/arm.c (arm_option_override): Remove trailing period
34747 from various diagnostics.
34748 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
34749 (msp430_expand_delay_cycles): Likewise.
34751 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34754 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
34755 full command-line argument, rather than just "str".
34756 (aarch64_validate_march): Likewise.
34757 (aarch64_validate_mtune): Likewise.
34759 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
34761 PR rtl-optimization/78911
34762 * lra-assigns.c (must_not_spill_p): New function.
34763 (spill_for): Use it.
34765 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34767 PR tree-optimization/79981
34768 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
34769 ATOMIC_COMPARE_EXCHANGE ifn result.
34770 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
34771 IFN_ATOMIC_COMPARE_EXCHANGE.
34773 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34776 * opts.c (parse_sanitizer_options): Add missing question mark to
34777 "did you mean" message.
34779 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34781 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
34783 (VMULEUH_UNS): Likewise.
34784 (VMULOUB_UNS): Likewise.
34785 (VMULOUH_UNS): Likewise.
34786 * config/rs6000/rs6000.c (builtin_function_type): Remove
34787 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
34789 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34792 * read-rtl-function.c (function_reader::read_rtx_operand): Update
34793 x with result of extra_parsing_for_operand_code_0.
34794 (function_reader::extra_parsing_for_operand_code_0): Convert
34795 return type from void to rtx, returning x. When reading
34796 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
34797 larger size containing struct block_symbol.
34799 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
34801 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
34802 -mfloat128-hardware without -m64.
34804 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
34807 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
34808 entries to the case statement that marks unsigned arguments to
34809 overloaded functions.
34811 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34813 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
34814 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
34816 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
34819 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
34820 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
34822 2017-03-10 Martin Liska <mliska@suse.cz>
34826 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
34827 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
34830 2017-03-10 Olivier Hainque <hainque@adacore.com>
34832 * tree-switch-conversion (array_value_type): Start by resetting
34833 candidate type to it's main variant.
34835 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34837 PR rtl-optimization/79909
34838 * combine.c (try_combine): Use simplify_replace_rtx on individual
34839 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
34840 of the whole CALL_INSN_FUNCTION_USAGE.
34842 PR tree-optimization/79972
34843 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
34844 get_range_info on SSA_NAMEs. Formatting fixes.
34846 2017-03-10 Richard Biener <rguenther@suse.de>
34847 Jakub Jelinek <jakub@redhat.com>
34849 PR tree-optimization/77975
34850 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
34851 edge to be constant.
34852 (get_val_for): For constant x return it. Formatting fix.
34853 (loop_niter_by_eval): Avoid pointless looping if the next iteration
34854 would use the same bases as the current one.
34856 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34858 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
34859 instead of vec_select for V1TImode.
34860 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
34862 (VSX_LE_128): Add V1TI to this mode iterator.
34863 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
34864 (*vsx_le_perm_store_<mode>): Likewise.
34865 (pre-reload splitter for VSX stores): Likewise.
34866 (post-reload splitter for VSX stores): Likewise.
34867 (*vsx_xxpermdi2_le_<mode>): Likewise.
34868 (*vsx_lxvd2x2_le_<mode>): Likewise.
34869 (*vsx_stxvd2x2_le_<mode>): Likewise.
34871 2017-03-09 Michael Eager <eager@eagercon.com>
34873 Correct failures with --enable-checking=yes,rtl.
34875 * config/microblaze/microblaze.c (microblaze_expand_shift):
34876 Replace GET_CODE test with CONST_INT_P and INTVAL test with
34877 test for const0_rtx.
34878 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
34879 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
34881 2017-03-09 Richard Biener <rguenther@suse.de>
34883 PR tree-optimization/79977
34884 * graphite-scop-detection.c (scop_detection::merge_sese):
34885 Handle the case of extra exits to blocks dominating the entry.
34887 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
34889 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
34892 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
34894 PR rtl-optimization/79949
34895 * lra-constraints.c (process_alt_operands): Check memory when
34896 trying to predict a cycle. Print about the overall increase.
34898 2017-03-09 Richard Biener <rguenther@suse.de>
34900 PR middle-end/79971
34901 * gimple-expr.c (useless_type_conversion_p): Preserve
34902 TYPE_SATURATING for fixed-point types.
34904 2017-03-09 Richard Biener <rguenther@suse.de>
34907 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
34908 alignment of BLKmode params.
34910 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34913 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
34914 (VALL_NO_V2Q): Likewise.
34916 * config/aarch64/aarch64-simd.md
34917 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
34919 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
34920 VALL_NO_V2Q mode iterator.
34921 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
34923 2017-03-09 Martin Liska <mliska@suse.cz>
34925 PR tree-optimization/79631
34926 * tree-chkp-opt.c (chkp_is_constant_addr): Call
34927 tree_int_cst_sign_bit just for INTEGER constants.
34929 2017-03-09 Martin Liska <mliska@suse.cz>
34933 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
34936 2017-03-09 Marek Polacek <polacek@redhat.com>
34939 * tree.c (inchash::add_expr): Handle TREE_VEC.
34941 2017-03-09 Martin Liska <mliska@suse.cz>
34944 (chkp_narrow_size_and_offset): New function.
34945 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
34946 (void chkp_parse_bit_field_ref): New function.
34947 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
34948 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
34950 2017-03-09 Martin Liska <mliska@suse.cz>
34953 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
34954 (chkp_find_bounds_1): Remove gcc_unreachable.
34956 2017-03-09 Jakub Jelinek <jakub@redhat.com>
34959 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
34960 BUILT_IN_SYNC*, determine the access type from the size suffix and
34961 always build a MEM_REF with that type. Handle forgotten
34962 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
34965 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
34966 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
34967 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
34968 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
34969 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
34970 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
34971 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
34972 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
34973 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
34974 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
34975 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
34976 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
34977 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
34978 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
34979 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
34980 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
34981 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
34982 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
34983 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
34984 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
34985 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
34986 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
34987 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
34988 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
34989 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
34990 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
34991 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
34992 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
34993 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
34994 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
34995 definitions outside of __OPTIMIZE__ guarded section.
34998 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
34999 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
35000 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
35001 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
35004 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35006 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
35007 ("vfenez<mode>"): Add missing constraints.
35009 2017-03-08 Martin Sebor <msebor@redhat.com>
35012 * config/nds32/nds32.c (nds32_option_override):
35013 Fix misspelled diagnostic.
35015 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35018 * gimplify.c (gimplify_omp_for): Replace index var in outer
35019 taskloop statement with an artificial variable and add
35020 OMP_CLAUSE_PRIVATE clause for it.
35022 2017-03-08 Richard Biener <rguenther@suse.de>
35024 PR tree-optimization/79955
35025 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
35026 for accesses that are completely outside of the variable.
35028 2017-03-08 Andrew Haley <aph@redhat.com>
35030 PR tree-optimization/79943
35031 * tree-ssa-loop-split.c (compute_new_first_bound): When
35032 calculating the new upper bound, (END-BEG) should be added, not
35035 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35037 * config/avr/avr.md (setmemhi): Make sure match_dup
35038 operand number comes before match_scratch.
35040 2017-03-08 Richard Biener <rguenther@suse.de>
35042 PR tree-optimization/79920
35043 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
35044 with ncopies == 1 to ...
35045 (vect_transform_slp_perm_load): ... here. Properly compute
35046 all element loads by iterating VF times over the group. Do
35047 not handle ncopies (computed in a broken way) in
35048 vect_create_mask_and_perm.
35050 2017-03-08 Jakub Jelinek <jakub@redhat.com>
35053 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
35054 is a uniform vector, use uniform_vector_p return value instead of
35055 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
35057 2017-03-07 Marek Polacek <polacek@redhat.com>
35059 PR middle-end/79809
35060 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
35061 (alloca_call_type): Likewise.
35063 2017-03-07 Martin Liska <mliska@suse.cz>
35065 * gcov.c (process_args): Put comment to correct location.
35067 2017-03-07 Martin Liska <mliska@suse.cz>
35069 PR middle-end/68270
35070 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
35071 Use array_at_struct_end_p instead of DECL_CHAIN (field).
35072 (chkp_narrow_bounds_for_field): Likewise.
35073 (chkp_parse_array_and_component_ref): Pass one more argument to
35076 2017-03-07 Richard Biener <rguenther@suse.de>
35078 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
35081 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
35083 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
35084 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
35086 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35089 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
35090 to end of description.
35091 (PARAM_MAX_STORES_TO_MERGE): Likewise.
35093 2017-03-07 Jakub Jelinek <jakub@redhat.com>
35095 PR rtl-optimization/79901
35096 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
35098 (*avx512f_<code><mode>3<mask_name>): ... this.
35099 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
35100 iterator instead of VI8_AVX2_AVX512BW.
35102 PR rtl-optimization/79901
35103 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
35104 min/max expander, expand it using expand_vec_cond_expr.
35107 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
35110 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35113 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
35114 to void * for PCH reasons.
35115 * dwarf2out.c (output_loc_operands, output_die): Cast
35116 v.val_vec.array to unsigned char *.
35118 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
35121 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
35124 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
35126 PR rtl-optimization/79571
35127 * lra-constraints.c (process_alt_operands): Calculate static
35128 reject and subtract it from overall when only addresses will be
35131 2017-03-06 Julia Koval <julia.koval@intel.com>
35134 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
35135 incoming stack boundary to 128 for 64-bit targets.
35137 2017-03-06 Richard Biener <rguenther@suse.de>
35139 PR tree-optimization/79894
35140 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
35141 to NULL after folding it.
35143 2017-03-06 Richard Biener <rguenther@suse.de>
35145 PR tree-optimization/79824
35146 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
35147 check disabling peeling for gaps.
35149 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
35151 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
35152 attributes): Document gettimeofday.
35154 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35156 * config/s390/s390.c (s390_option_override_internal): Set
35157 PARAM_MIN_VECT_LOOP_BOUND
35159 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
35161 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
35162 * config/s390/s390.md: Likewise.
35164 2017-03-06 Jakub Jelinek <jakub@redhat.com>
35167 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
35168 (<avx2_avx512>_perm<mode>): Rename to ...
35169 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
35171 (<avx512>_perm<mode>_mask): Rename to ...
35172 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
35174 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
35175 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
35176 instead of VI8F_256_512.
35177 (avx512f_perm<mode>): New define_expand.
35178 (avx512f_perm<mode>_mask): Likewise.
35179 (avx512f_perm<mode>_1<mask_name>): New define_insn.
35180 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
35182 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35184 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
35185 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
35187 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
35189 2017-03-06 Martin Liska <mliska@suse.cz>
35192 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
35193 when having a SSA NAME w/o VAR_DECL assigned to it.
35195 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35197 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
35198 msa_dpsub_<su>_d): Fix MODE for vec_select.
35200 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
35202 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
35204 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
35206 2017-03-06 Richard Biener <rguenther@suse.de>
35208 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
35209 * plugin.c (register_plugin_info): Likewise.
35210 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
35212 2017-03-05 Jakub Jelinek <jakub@redhat.com>
35214 * config/i386/sse.md (sse_storehps, sse_storelps,
35215 avx_<castmode><avxsizesuffix>_<castmode>,
35216 avx512f_<castmode><avxsizesuffix>_<castmode>,
35217 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
35218 in condition that at least one operand is not a MEM.
35220 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35222 PR middle-end/79805
35223 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
35224 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
35226 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
35227 gimple_call_nothrow_p flag based on whether original builtin can throw.
35228 If it can, emit following stmts on the fallthrough edge.
35229 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
35230 don't create new bb if inserting just debug stmts on the edge, try to
35231 insert them on the fallthru bb or just reset debug stmts.
35233 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
35236 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
35237 restore recog_data (including the operand rtxes inside it) around
35238 the call to get_insn_template.
35240 2017-03-03 Martin Sebor <msebor@redhat.com>
35242 PR tree-optimization/79699
35243 * context.c (context::~context): Free MPFR caches to avoid
35244 a memory leak on program exit.
35246 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35248 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
35249 Use wide_int::ulow () instead of .elt (0).
35251 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35253 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
35254 (*pushxf): Limit oF constraint to 32bit targets and add oC
35255 constraint for 64bit targets.
35256 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
35257 (*pushdf): Change rmF constraint to rmC.
35259 2017-03-03 Martin Liska <mliska@suse.cz>
35261 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
35262 Remove unused variable.
35264 2017-03-03 Jakub Jelinek <jakub@redhat.com>
35267 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
35268 is a memory operand, increase num_memory.
35269 (ix86_expand_args_builtin): Likewise.
35271 2017-03-03 Jan Hubicka <jh@suse.cz>
35274 * ipa-devirt.c (maybe_record_node): Properly handle
35275 __cxa_pure_virtual visibility.
35277 2017-03-03 Martin Liska <mliska@suse.cz>
35279 PR tree-optimization/79803
35280 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
35282 (pass_loop_prefetch::execute): Disabled optimization if an
35283 assumption about L1 cache size is not met.
35285 2017-03-03 Martin Liska <mliska@suse.cz>
35287 PR rtl-optimization/79574
35288 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
35289 (hash_scan_set): Likewise.
35290 (dump_hash_table): Likewise.
35291 (hoist_code): Likewise.
35293 2017-03-03 Richard Biener <rguenther@suse.de>
35295 * fixed-value.c (fixed_from_string): Restore use of elt (1)
35296 in place of uhigh ().
35297 (fixed_convert_from_real): Likewise.
35299 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
35302 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
35304 2017-03-03 Richard Biener <rguenther@suse.de>
35306 PR middle-end/79818
35307 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
35308 TYPE_OVERFLOW_UNDEFINED check.
35310 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35312 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
35314 (vector_ae_<mode>_p): Likewise.
35315 (vector_nez_<mode>_p): Likewise.
35316 (vector_ne_v2di_p): Likewise.
35317 (vector_ae_v2di_p): Likewise.
35318 (vector_ne_<mode>_p): Likewise.
35319 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
35321 (vsx_tsqrt<mode>2_fe): Likewise.
35323 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
35326 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
35328 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35330 PR rtl-optimization/79780
35331 * cprop.c (one_cprop_pass): When second and further conditional trap
35332 in a single basic block is turned into an unconditional trap, turn it
35333 into a deleted note to avoid RTL verification failures.
35335 2017-03-02 Richard Biener <rguenther@suse.de>
35337 * fold-const.c (const_binop): Use ulow () instead of elt (0).
35339 2017-03-02 Richard Biener <rguenther@suse.de>
35341 PR tree-optimization/79345
35343 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
35344 param and abort the walk, returning -1 if it is hit.
35345 (walk_aliased_vdefs): Take a limit param and pass it on.
35346 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
35347 defaulting to 0 and return a signed int.
35348 * tree-ssa-uninit.c (struct check_defs_data): New struct.
35349 (check_defs): New helper.
35350 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
35351 about uninitialized memory.
35352 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
35353 bogus uninitialized warning.
35354 (fixed_convert_from_real): Likewise.
35356 2017-03-02 Bin Cheng <bin.cheng@arm.com>
35358 PR tree-optimization/66768
35359 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
35360 iv_use if base object can't be determined.
35362 2017-03-02 Jakub Jelinek <jakub@redhat.com>
35364 PR tree-optimization/79345
35365 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
35366 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
35367 (get_pattern_stats): Initialize it.
35368 * genemit.c (gen_expand): Verify match_scratch numbers come after
35369 match_operand/match_dup numbers.
35370 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
35371 match_scratch numbers.
35372 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
35374 * config/s390/s390.md (trunctdsd2): Likewise.
35376 2017-03-02 Richard Biener <rguenther@suse.de>
35378 * wide-int.h (wide_int_storage::operator=): Implement in terms
35381 2017-03-02 Richard Biener <rguenther@suse.de>
35383 PR tree-optimization/79777
35384 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
35385 the to insert expression to sth existing.
35387 2017-03-01 Martin Sebor <msebor@redhat.com>
35389 PR middle-end/79692
35390 * gimple-ssa-sprintf.c
35391 (directive::known_width_and_precision): New function.
35392 (format_integer): Use it.
35393 (get_mpfr_format_length): Consider the full range of precision
35394 when computing %g output with the # flag. Set the likely byte
35395 count to 3 rather than 1 when precision is indeterminate.
35396 (format_floating): Correct the lower bound of precision.
35398 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35400 * doc/invoke.texi: Document default code model for 64-bit Linux.
35402 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35405 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
35406 udiv rather than div since input pattern is unsigned.
35408 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
35410 * config/i386/i386.c (print_reg): Warn for values of
35411 unsupported size in integer register.
35413 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
35416 * config/rs6000/predicates.md (current_file_function_operand): Do
35417 not allow self calls to be local if the function is replaceable.
35419 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
35422 * config/rs6000/altivec.h (vec_ctz and others): Change the
35423 preprocessor macro that controls conditional compilation from
35424 _ARCH_PWR9 to __POWER9_VECTOR__.
35425 (vec_all_ne): Change parameterization of __altivec_scalar_pred
35426 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
35427 control (instead of _ARCH_PWR9 control) so that template
35428 definition uses power9-specific function.
35429 (vec_any_eq): Likewise.
35430 (vec_all_ne): Change macro definition to use a power9-specific
35431 expansion under #ifdef __POWER9_VECTOR__ control (instead of
35432 _ARCH_PWR9 control).
35433 (vec_any_eq) Likewise.
35434 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
35435 expansion for CMPNEF to remove support for xvcmpnesp instruction.
35436 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
35437 support for xvcmpnedp instruction.
35438 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
35439 macro expansion so that Power9 implementation of vec_all_ne does
35440 not use the AltiVec predicate framework.
35441 (VCMPNEH_P): Likewise.
35442 (VCMPNEW_P): Likewise.
35443 (VCMPNED_P): Likewise.
35444 (VCMPNEFP_P): Likewise.
35445 (VCMPNEDP_P): Likewise.
35446 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
35447 implementation of vec_any_eq to not use AltiVec predicate
35449 (VCMPAEH_P): Likewise.
35450 (VCMPAEW_P): Likewise.
35451 (VCMPAED_P): Likewise.
35452 (VCMPAEFP_P): Likewise.
35453 (VCMPAEDP_P): Likewise.
35454 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
35455 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
35456 not use the AltiVec predicate framework.
35457 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
35458 of vec_any_eq to not use AltiVec predicate framework.
35459 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
35460 support for predefined __POWER9_VECTOR__ macro to indicate that
35461 Power9 instruction selection is enabled.
35462 (altivec_overloaded_builtins): Remove extraneous
35463 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
35464 function argument types RS6000_BTI_bool_V16QI and
35465 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
35466 entry for overloaded function argument types RS6000_BTI_bool_V4SI
35467 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
35468 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
35469 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
35470 Power9 for implementations of vec_cmpne. Change the signature for
35471 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
35472 (representing vec_all_ne) to remove the previously described first
35473 argument of type RS6000_BTI_INTSI, as this was an artifact of
35474 reliance on the AltiVec predicate framework, which is no longer
35475 used in the implementation of these functions. Add
35476 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
35477 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
35478 since, unlike the AltiVec predicate framework implementation, we
35479 do not share function descriptors between vec_alle and vec_anyeq.
35480 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
35481 set of modes that receive special treatment even when
35482 TARGET_P9_VECTOR is true. The special treatment emits code that
35483 does not depend on Power9 instructions.
35484 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
35485 define_expand to not rely on AltiVec predicate framework.
35486 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35488 (vector_ne_v2di_p): Change this define_expand to not rely on
35489 AltiVec predicate framework.
35490 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
35492 (vector_ne_<mode>_p): Change this define_expand to not rely on
35493 AltiVec predicate framework.
35494 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
35496 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
35497 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
35498 define_insn pattern.
35499 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
35500 define_insn pattern because the xvcmpne<VSs>. instruction is not
35502 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
35503 instruction is not supported.
35505 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35507 * config/nvptx/nvptx.c: Include intl.h.
35509 2017-03-01 Martin Jambor <mjambor@suse.cz>
35512 * ipa-prop.h (ipa_bits): Removed field known.
35513 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
35514 to pointers. Adjusted their comments to warn about their sharing.
35515 (ipcp_transformation_summary): Change bits to a vector of pointers.
35516 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
35517 (ipa_get_ipa_bits_for_value): Declare.
35518 * tree-vrp.h (value_range): Mark as GTY((for_user)).
35519 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
35520 (ipa_bits_hash_table): Likewise.
35521 (ipa_vr_ggc_hash_traits): Likewise.
35522 (ipa_vr_hash_table): Likewise.
35523 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
35524 being pointers and vr_known being removed.
35525 (ipa_set_jf_unknown): Likewise.
35526 (ipa_get_ipa_bits_for_value): New function.
35527 (ipa_set_jfunc_bits): Likewise.
35528 (ipa_get_value_range): New overloaded functions.
35529 (ipa_set_jfunc_vr): Likewise.
35530 (ipa_compute_jump_functions_for_edge): Use the above functions to
35531 construct bits and vr parts of jump functions.
35532 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
35533 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35535 (ipcp_grow_transformations_if_necessary): Also allocate
35536 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
35538 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
35539 them. Fix too long lines.
35540 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
35541 vr_known being removed.
35542 (ipa_read_jump_function): Use new setter functions to construct bits
35543 and vr parts of jump functions or set them to NULL.
35544 (write_ipcp_transformation_info): Adjust for bits being pointers.
35545 (read_ipcp_transformation_info): Likewise.
35546 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
35548 Include gt-ipa-prop.h.
35549 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
35551 (ipcp_store_bits_results): Likewise.
35552 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
35553 Do not write to existing jump functions but use a temporary instead.
35555 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35558 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
35559 attempt to use its first operand as BIT_FIELD_REF base.
35561 2017-03-01 Richard Biener <rguenther@suse.de>
35563 PR middle-end/79721
35564 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
35565 interpolating formula in wrapping arithmetic.
35566 (chrec_apply): Convert chrec_evaluate return value to wanted type.
35568 2017-03-01 Jakub Jelinek <jakub@redhat.com>
35570 PR tree-optimization/79734
35571 * tree-vect-generic.c (expand_vector_condition): Optimize
35572 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
35573 Handle VEC_COND_EXPR where comparison has different inner width from
35574 type's inner width.
35576 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
35578 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
35579 markup, and similar issues. Remove @opindex entries for things
35580 that aren't options. Add missing -mmpy-option entries.
35582 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35584 PR tree-optimization/79737
35585 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
35586 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
35587 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
35588 instead of byte_size. Formatting fix.
35589 (shift_bytes_in_array_right): Formatting fix.
35591 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
35594 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
35595 condition on optimize for the leaf function test.
35597 2017-02-28 Martin Liska <mliska@suse.cz>
35600 * read-rtl-function.c (function_reader::handle_unknown_directive):
35601 Bail out when one uses -flto.
35603 2017-02-28 Martin Liska <mliska@suse.cz>
35605 * common.opt: Replace space with tabular for options of <number>
35607 * config/i386/i386.opt: Show <number> value for
35608 -mlarge-data-threshold.
35609 * opts.c (print_filtered_help): Do not display number in hexadecimal
35612 2017-02-28 Martin Liska <mliska@suse.cz>
35614 * common.opt: Fix --help=option -Q for options which are of
35617 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
35619 * config/i386/i386.c (print_reg): Error out for values
35620 of 8-bit size in invalid integer register.
35622 2017-02-28 Martin Sebor <msebor@redhat.com>
35624 PR tree-optimization/79691
35625 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
35627 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35630 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
35631 gcc_unreachable with output_operand_lossage.
35633 2017-02-28 Richard Biener <rguenther@suse.de>
35635 PR tree-optimization/79740
35636 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
35638 (visit_nary_op): Insert the nary into the hashtable if we
35639 pattern-matched sth.
35640 * tree-ssa-pre.c (eliminate_insert): Robustify.
35642 2017-02-28 Richard Biener <rguenther@suse.de>
35644 PR middle-end/79731
35645 * fold-const.c (decode_field_reference): Reject out-of-bound
35648 2017-02-28 Jakub Jelinek <jakub@redhat.com>
35650 * config/i386/i386.c: Include intl.h.
35651 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
35652 instead of just cond ? "..." : "...".
35653 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
35654 * coverage.c (read_counts_file): Likewise.
35655 * omp-offload.c: Include intl.h.
35656 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
35657 of just cond ? "..." : "...".
35658 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
35659 of just cond ? "..." : "...".
35661 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
35664 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
35666 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
35667 'tune for' CPU name.
35668 * config/arm/arm-cpu-data.h: Regenerated.
35670 2017-02-28 Richard Biener <rguenther@suse.de>
35672 PR tree-optimization/79732
35673 * tree-inline.c (expand_call_inline): Do not shadow var.
35675 2017-02-28 Richard Biener <rguenther@suse.de>
35677 PR tree-optimization/79723
35678 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
35679 address-space properly.
35681 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
35683 * doc/optinfo.texi (Optimization groups): Fix option used for
35685 * doc/invoke.texi (-fopt-info): Document "omp".
35686 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
35687 (OPTGROUP_ALL): Add OPTGROUP_OMP.
35688 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
35689 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
35690 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
35692 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
35694 * dumpfile.c (optgroup_options): Instead of "openmp", associate
35695 OPTGROUP_OMP with "omp".
35697 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
35700 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
35701 for arithmetic shift of unsigned V2DI.
35703 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
35705 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
35706 arc/linux.h headers.
35707 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
35708 (LINK_SPEC): Likewise.
35709 (ARC_TLS_EXTRA_START_SPEC): Likewise.
35710 (EXTRA_SPECS): Likewise.
35711 (STARTFILE_SPEC): Likewise.
35712 (ENDFILE_SPEC): Likewise.
35713 (LIB_SPEC): Likewise.
35714 (TARGET_SDATA_DEFAULT): Likewise.
35715 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
35716 (MULTILIB_DEFAULTS): Likewise.
35717 (DWARF2_UNWIND_INFO): Likewise.
35718 * config/arc/big.h: New file.
35719 * config/arc/elf.h: Likewise.
35720 * config/arc/linux.h: Likewise.
35721 * config/arc/t-uClibc: Remove.
35723 2017-02-27 Bin Cheng <bin.cheng@arm.com>
35725 PR tree-optimization/77536
35726 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
35727 (tree_transform_and_unroll_loop): Use above function to compute the
35728 estimated niter of unrolled loop and use it when scaling profile.
35729 Also use count info rather than frequency if it's non-zero.
35730 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
35731 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
35732 (vect_transform_loop): Call above function.
35734 2017-02-27 Richard Biener <rguenther@suse.de>
35736 PR tree-optimization/45397
35737 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
35738 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
35739 (visit_nary_op): Add pattern matching for CSEing sign-changed
35740 or truncated operations with wider ones.
35742 2017-02-27 Richard Biener <rguenther@suse.de>
35744 PR tree-optimization/79690
35745 * tree-vect-stmts.c (vectorizable_store): Use vector type
35746 built from the DR with address-space.
35748 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
35750 * doc/invoke.texi (Optimize Options): Refine the description
35751 of asan-use-after-return.
35753 2017-02-25 Alan Modra <amodra@gmail.com>
35755 PR rtl-optimization/79584
35756 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
35757 base, not ad->base_term, the reg within base. Remove assertion
35758 that ad->base == ad->base_term. Replace gen_int_mode using
35759 bogus mode with const0_rtx.
35761 2017-02-25 Jakub Jelinek <jakub@redhat.com>
35763 PR middle-end/79396
35764 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
35765 FMA_EXPR like tcc_binary or tcc_unary.
35767 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
35770 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
35772 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
35773 (output_loc_operands): Handle DW_OP_call_ref and
35774 DW_OP_GNU_variable_value.
35775 (struct variable_value_struct): New type.
35776 (struct variable_value_hasher): Likewise.
35777 (variable_value_hash): New variable.
35778 (string_types): Remove.
35779 (copy_loc_descr): New function.
35780 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
35781 (prepend_loc_descr_to_each): New function.
35782 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
35783 instead of add_loc_descr_to_each if the first argument is single
35784 location list and the second has multiple.
35785 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
35786 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
35787 when looking for variable value which doesn't have other location info.
35788 (loc_list_from_tree): Formatting fix.
35789 (gen_array_type_die): Simplify DW_AT_string_length handling.
35790 (adjust_string_types): Remove.
35791 (gen_subprogram_die): Don't call adjust_string_types nor test/set
35792 string_types. Call resolve_variable_values.
35793 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
35794 (resolve_addr_in_expr): Likewise. Add A argument.
35795 (copy_deref_exprloc): Remove deref argument. Adjust for the
35796 original expression being DW_OP_GNU_variable_value with optionally
35797 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
35798 optionally after it.
35799 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
35800 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
35801 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
35802 (variable_value_hasher::hash, variable_value_hasher::equal): New
35804 (resolve_variable_value_in_expr, resolve_variable_value,
35805 resolve_variable_values, note_variable_value_in_expr,
35806 note_variable_value): New functions.
35807 (dwarf2out_early_finish): Call note_variable_value on all toplevel
35810 2017-02-24 Jakub Jelinek <jakub@redhat.com>
35813 * opts.h (handle_generated_option): Add GENERATED_P argument.
35814 * opts-common.c (handle_option): Adjust function comment.
35815 (handle_generated_option): Add GENERATED_P argument, pass it to
35817 (control_warning_option): Pass false to handle_generated_option
35819 * opts.c (maybe_default_option): Pass true to handle_generated_option
35821 * optc-gen.awk: Likewise.
35823 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35825 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
35826 a REG, look at the REG it is a SUBREG of.
35827 (splitter for cmpeqsi_t): Ditto.
35829 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35831 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
35832 the special USEs with the pattern of the insn, not the insn itself.
35834 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
35837 * doc/invoke.texi: Document -mload-store-pairs.
35839 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35840 Sandra Loosemore <sandra@codesourcery.com>
35842 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
35843 argument isn't a CONST_INT.
35844 (nios2_alternate_compare_const): Assert op is a CONST_INT.
35845 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
35846 (nios2_validate_compare): Bypass alternate compare logic if *op2
35847 is not a CONST_INT.
35848 (ldstwm_operation_p): Return false if first_base is not a REG or
35849 if first_offset is not a CONST_INT.
35851 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35853 * config/cris/cris.md: Use correct operand in a define_peephole2.
35855 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35857 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
35859 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35861 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
35862 this_insn if it is an INSN or JUMP_INSN.
35863 (force_offsettable): Look at base, not at addr.
35864 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
35865 on things that aren't necessarily CONST_INTs.
35867 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
35869 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
35870 -mfpmath=sse is the default also for x86-32 targets with SSE2
35871 instruction set when @option{-ffast-math} is enabled
35873 2017-02-24 Jeff Law <law@redhat.com>
35875 PR rtl-optimizatoin/79286
35876 * ira.c (update_equiv_regs): Drop may_trap_p exception to
35879 2017-02-24 Richard Biener <rguenther@suse.de>
35881 PR tree-optimization/79389
35882 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
35885 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
35887 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
35888 function comment to reflect reality.
35889 (loop_exits_before_overflow): Fix typo in function description.
35891 2017-02-24 Richard Biener <rguenther@suse.de>
35893 PR tree-optimization/79389
35894 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
35895 properly that a threading opportunity exists. Detect conditional
35896 copy/constant propagation opportunities.
35898 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
35900 * config/visium/visium.md (type): Add trap.
35901 (b): New mode attribute.
35902 (*btst): Rename into...
35903 (*btst<mode>): ...this and adjust.
35904 (*cbranchsi4_btst_insn): Rename into...
35905 (*cbranch<mode>4_btst_insn): ...this and adjust.
35906 (trap): New define_insn.
35908 2017-02-23 Jakub Jelinek <jakub@redhat.com>
35910 PR tree-optimization/79389
35911 * ifcvt.c (struct noce_if_info): Add rev_cond field.
35912 (noce_reversed_cond_code): New function.
35913 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
35914 reversed_comparison_code. Formatting fix.
35915 (noce_try_store_flag): Test rev_cond != NULL in addition to
35916 reversed_comparison_code.
35917 (noce_try_store_flag_constants): Likewise.
35918 (noce_try_store_flag_mask): Likewise.
35919 (noce_try_addcc): Use rev_cond if non-NULL instead of
35920 reversed_comparison_code.
35921 (noce_try_cmove_arith): Likewise. Formatting fixes.
35922 (noce_try_minmax, noce_try_abs): Clear rev_cond.
35923 (noce_find_if_block): Initialize rev_cond.
35924 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
35925 instead of false as last argument never attempt to reverse it
35928 2017-02-23 Bin Cheng <bin.cheng@arm.com>
35930 PR tree-optimization/79663
35931 * tree-predcom.c (combine_chains): Process refs in reverse order
35932 only for ZERO length chains, and add explaining comment.
35934 2017-02-23 Jeff Law <law@redhat.com>
35936 PR tree-optimization/79578
35937 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
35938 in call to operand_equal_p.
35940 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
35943 * config/i386/cpuid.h: Fix another undefined behavior.
35945 2017-02-23 Richard Biener <rguenther@suse.de>
35947 PR tree-optimization/79683
35948 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
35949 vector types for data-refs.
35951 2017-02-23 Martin Liska <mliska@suse.cz>
35953 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
35955 2017-02-23 Jakub Jelinek <jakub@redhat.com>
35957 PR middle-end/79665
35958 * internal-fn.c (get_range_pos_neg): Moved to ...
35959 * tree.c (get_range_pos_neg): ... here. No longer static.
35960 * tree.h (get_range_pos_neg): New prototype.
35961 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
35962 are known to be in between 0 and signed maximum inclusive, try to
35963 expand both unsigned and signed divmod and use the cheaper one from
35966 2017-02-22 Jeff Law <law@redhat.com>
35968 PR tree-optimization/79578
35969 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
35970 to compare base operands.
35972 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
35975 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
35976 gpc_reg_operand instead of fpr_reg_operand.
35978 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
35980 * config/mips/mips.c (mips_return_in_memory): Force FP
35981 vector types to be returned in memory for o32 ABI.
35983 2017-02-22 Jakub Jelinek <jakub@redhat.com>
35985 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
35986 instead of DW_TAG_member for static data member declarations and don't
35987 set no_linkage_name for static inline data members.
35988 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
35991 2017-02-22 Martin Liska <mliska@suse.cz>
35993 * doc/invoke.texi: Replace inequality signs with square brackets
35996 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
35999 * lra-constraints.c (simplify_operand_subreg): Handle
36000 WORD_REGISTER_OPERATIONS targets.
36002 2017-02-22 Jakub Jelinek <jakub@redhat.com>
36005 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
36006 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
36007 elimination by swapping fld*.
36009 2017-02-22 Richard Biener <rguenther@suse.de>
36011 PR tree-optimization/79673
36012 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
36013 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
36014 irrelevant address-space qualifiers and avoiding a
36015 ADDR_SPACE_CONVERT_EXPR from fold_convert.
36017 2017-02-22 Richard Biener <rguenther@suse.de>
36019 PR tree-optimization/79666
36020 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
36021 to not symbolically negate if that may introduce undefined
36024 2017-02-22 Martin Liska <mliska@suse.cz>
36027 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
36028 * data-streamer-out.c (streamer_write_gcov_count_stream):
36030 * value-prof.c (stream_out_histogram_value): Make assert more
36031 precise based on type of counter.
36033 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
36036 * config/i386/i386.md (standard_x87sse_constant_load splitter):
36037 Use nonimmediate_operand instead of memory_operand for operand 1.
36038 (float-extend standard_x87sse_constant_load splitter): Ditto.
36040 2017-02-21 Jeff Law <law@redhat.com>
36042 PR tree-optimization/79621
36043 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
36044 blocks with edges to themselves.
36046 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36049 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
36050 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
36051 Use gimple_call_builtin_p.
36054 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
36055 on temporarily removed DEBUG_INSNs.
36057 PR tree-optimization/79649
36058 * tree-loop-distribution.c (classify_partition): Give up on
36059 non-generic address space loads/stores.
36061 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
36063 * doc/loop.texi (Loop manipulation): Remove nonexistent
36064 tree_ssa_loop_version from the documentation.
36065 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
36067 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36070 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
36071 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
36072 * config/rs6000/rs6000.c: Include except.h.
36073 (rs6000_expand_split_stack_prologue): Call
36074 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
36076 2017-02-21 Martin Jambor <mjambor@suse.cz>
36079 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
36080 have been analyzed.
36082 2017-02-21 Martin Jambor <mjambor@suse.cz>
36084 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
36085 for backward compatibility only.
36086 * doc/invoke.texi (Option Summary): Remove all references to
36087 -fipa-cp-alignment.
36089 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
36093 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36095 * lra-constraints.c (curr_insn_transform): Handle
36096 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36098 2017-02-21 Martin Liska <mliska@suse.cz>
36100 * config/i386/i386.opt: Replace -masm-dialect with -masm.
36102 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
36104 PR translation/79638
36105 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
36107 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
36110 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
36111 (arm_function_ok_for_sibcall): Return false for an indirect call by
36112 descriptor if all the argument registers are used.
36113 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
36114 alignment of the function.
36116 2017-02-21 Jakub Jelinek <jakub@redhat.com>
36118 PR tree-optimization/61441
36119 * simplify-rtx.c (simplify_const_unary_operation): For
36120 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
36121 the sNaN unmodified.
36123 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
36125 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
36126 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
36127 instead of SYSTEM_HEADER_DIR.
36129 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
36130 Martin Liška <mliska@suse.cz>
36132 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
36133 Fix typos and grammar, use active voice, and clarify.
36135 2017-02-20 Marek Polacek <polacek@redhat.com>
36137 PR middle-end/79537
36138 * gimplify.c (gimplify_expr): Handle unused *&&L;.
36141 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
36143 2017-02-20 Jakub Jelinek <jakub@redhat.com>
36146 * config/i386/i386.c (ix86_expand_builtin): Handle
36147 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
36148 ix86_builtins_isa[fcode].isa as a requirement of those
36149 flags and any other flag in the bitmask.
36150 (ix86_init_mmx_sse_builtins): Use 0 instead of
36151 ~OPTION_MASK_ISA_64BIT as mask.
36152 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
36153 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
36154 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
36155 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
36157 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36160 * lra-constraints.c (split_reg): Check requested split mode
36161 is supported by the register.
36163 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36165 * lra-constraints.c (simplify_operand_subreg): Remove early
36168 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36171 * lra-constraints.c (curr_insn_transform): Tighten condition
36172 for converting SUBREG reloads from OP_OUT to OP_INOUT.
36174 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
36177 * lra-constraints.c (curr_insn_transform): Handle
36178 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
36180 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
36183 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
36185 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
36187 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
36190 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
36193 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36195 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
36198 2017-02-18 Jakub Jelinek <jakub@redhat.com>
36200 * final.c (last_columnnum, override_columnnum): New variables.
36201 (final_start_function): Set last_columnnum, pass it to begin_prologue
36202 hook and pass 0 to dwarf2out_begin_prologue.
36203 (final_scan_insn): Update override_columnnum. Pass last_columnnum
36204 to source_line debug hook.
36205 (notice_source_line): Compute last_columnnum and for debug_column_info
36206 return true on column changes.
36207 * debug.h (struct gcc_debug_hooks): Add column argument to
36208 source_line and begin_prologue hooks.
36209 (debug_nothing_int_charstar_int_bool): Remove prototype.
36210 (debug_nothing_int_int_charstar,
36211 debug_nothing_int_int_charstar_int_bool): New prototypes.
36212 (dwarf2out_begin_prologue): Add column argument.
36213 * debug.c (do_nothing_debug_hooks): Adjust source_line and
36214 begin_prologue hooks.
36215 (debug_nothing_int_charstar_int_bool): Remove.
36216 (debug_nothing_int_int_charstar,
36217 debug_nothing_int_int_charstar_int_bool): New functions.
36218 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
36219 through to dwarf2out_source_line.
36220 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
36221 (dwarf2out_source_line): Add column argument, emit it if requested.
36222 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
36224 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36225 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
36226 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
36227 through to dwarf2out_begin_prologue.
36228 (vmsdbgout_source_line): Add column argument, pass it through to
36229 dwarf2out_source_line.
36230 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
36231 dbxout_source_line caller.
36232 (dbxout_source_line): Add column argument.
36234 * common.opt (gno-column-info, gcolumn-info): New options.
36235 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
36236 (check_die): Also test for multiple DW_AT_decl_column attributes.
36237 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
36238 DW_AT_decl_column if requested.
36239 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
36241 (gen_variable_die): Likewise.
36242 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
36243 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
36246 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
36247 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
36248 (ix86_handle_option): Handle OPT_m3dnowa.
36249 * doc/invoke.texi (-m3dnowa): Document.
36250 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
36251 -m3dnowa instead of -m3dnow -march=athlon.
36254 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
36255 instead of gcc_assert for K, r and R code checks. Formatting fixes.
36257 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36260 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
36261 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
36262 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
36263 generator for vsx_xxpermdi_<mode>_be.
36264 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
36265 force big-endian semantics.
36266 (vsx_xxpermdi_<mode>_be): New define_expand with same
36267 implementation as previous version of vsx_xxpermdi_<mode>.
36269 2017-02-17 Jakub Jelinek <jakub@redhat.com>
36271 PR tree-optimization/79327
36272 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
36273 variable, its initialization and use.
36275 2017-02-17 Julia Koval <julia.koval@intel.com>
36277 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
36278 (OPTION_MASK_ISA_PKU_UNSET): New.
36279 (ix86_handle_option): Handle -mrdpid.
36280 * config/i386/cpuid.h (bit_RDPID): New.
36281 * config/i386/driver-i386.c (host_detect_local_cpu):
36282 Detect RDPID feature.
36283 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
36284 * config/i386/i386-c.c (ix86_target_macros_internal):
36286 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
36287 (ix86_valid_target_attribute_inner_p): Add "rdpid".
36288 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
36289 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
36290 * config/i386/i386.md (define_insn "rdpid"): New.
36291 * config/i386/i386.opt Add -mrdpid.
36292 * config/i386/immintrin.h (_rdpid_u32): New.
36294 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
36296 PR rtl-optimization/79541
36297 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
36298 instead of transforming it into USE.
36300 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
36302 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
36303 If HONOR_SNANS (SFmode) force the input to a register.
36304 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
36305 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
36306 an frsp or similar insn.
36308 2017-02-17 Martin Liska <mliska@suse.cz>
36310 PR rtl-optimization/79577
36311 * params.def (selsched-max-sched-times): Increase minimum to 1.
36313 2017-02-17 Martin Liska <mliska@suse.cz>
36315 PR rtl-optimization/79574
36316 * gcse.c (want_to_gcse_p): Prevent integer overflow.
36318 2017-02-17 Martin Liska <mliska@suse.cz>
36320 PR tree-optimization/79529
36321 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
36322 ssa_defined_default_def_p to handle cases which are implicitly
36324 * tree-ssa.c (ssa_defined_default_def_p): New function.
36325 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
36326 which are implicitly defined.
36327 * tree-ssa.h (ssa_defined_default_def_p): Declare.
36329 2017-02-17 Richard Biener <rguenther@suse.de>
36331 PR middle-end/79576
36332 * params.def (max-ssa-name-query-depth): Limit to 10.
36334 2017-02-17 Richard Biener <rguenther@suse.de>
36336 PR tree-optimization/79552
36337 * tree-ssa-structalias.c (visit_loadstore): Properly verify
36340 2017-02-17 Richard Biener <rguenther@suse.de>
36343 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
36345 2017-02-17 Marek Polacek <polacek@redhat.com>
36347 PR middle-end/79536
36348 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
36349 (fold_negate_expr): New wrapper.
36351 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
36353 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
36354 Correct terminology and de-emphasize pre-standard behavior.
36356 2017-02-16 Alan Modra <amodra@gmail.com>
36358 PR rtl-optimization/79286
36359 * ira.c (def_dominates_uses): New function.
36360 (update_equiv_regs): Don't create an equivalence for insns that
36361 may trap where the register def does not dominate the use.
36363 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
36365 PR rtl-optimization/78127
36366 * lra.c (lra): Call lra_eliminate before finish the loop after
36369 2017-02-16 Richard Biener <rguenther@suse.de>
36371 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
36373 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
36374 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
36375 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
36376 (isl_val_int_from_wi): New function.
36377 (extract_affine_gmp): Rename to ...
36378 (extract_affine_wi): ... this, take a widest_int.
36379 (extract_affine_int): Just wrap extract_affine_wi.
36380 (add_param_constraints): Use isl_val_int_from_wi.
36381 (add_loop_constraints): Likewise, and extract_affine_wi.
36383 2017-02-15 Jeff Law <law@redhat.com>
36385 PR middle-end/79521
36386 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
36387 ira_init_register_move_cost_if_necessary.
36389 2017-02-15 Martin Sebor <msebor@redhat.com>
36391 PR middle-end/32003
36392 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
36393 removed in a prior commit.
36395 2017-02-15 Bin Cheng <bin.cheng@arm.com>
36397 PR tree-optimization/79347
36398 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
36399 counters during peeling.
36401 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
36403 * Makefile.in (site.exp): Remove "set ISLVER".
36405 2017-02-15 Jakub Jelinek <jakub@redhat.com>
36408 * real.c (real_from_integer): Call real_convert even for decimal.
36410 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
36413 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
36415 2017-02-14 Andrew Pinski <apinski@cavium.com>
36417 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
36418 cores and change the partno/implementer to be correct.
36419 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
36420 the 'B" as the implementer.
36421 * config/aarch64/aarch64-tune.md: Regenerate.
36423 2017-02-14 Carl Love <cel@us.ibm.com>
36425 * config/rs6000/rs6000.c: Add case statement entry to make the
36426 xvcvuxdsp built-in argument unsigned.
36427 * config/rs6000/vsx.md: Fix the source and return operand types so they
36428 match the instruction definitions from the ISA document. Fix typo
36429 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
36432 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
36435 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
36436 member early_clobber_alts.
36437 * lra-lives.c (reg_early_clobber_p): New.
36438 (process_bb_lives): Use it.
36439 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
36440 (debug_operand_data): Initialize early_clobber_alts.
36441 (setup_operand_alternative): Set up early_clobber_alts.
36442 (collect_non_operand_hard_regs): Ditto. Pass early clobber
36443 alternatives to new_insn_reg.
36444 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
36446 (lra_update_insn_regno_info): Pass the new arg.
36448 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36450 PR middle-end/79505
36451 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
36452 (new_oacc_loop_raw): Don't clear already cleared fields.
36455 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
36456 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
36457 _mm512_prefetch_i64gather_ps): New inline functions and macros.
36459 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
36462 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
36464 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
36467 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
36468 the extra instruction to the right place to store 128-bit constant
36471 2017-02-14 Martin Sebor <msebor@redhat.com>
36473 PR middle-end/79448
36474 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
36475 warning for strings of unknown length.
36477 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
36479 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
36481 2017-02-14 Jeff Law <law@redhat.com>
36484 * ira-costs.c (scan_one_insn): Initialize register move costs
36485 for pseudos seen in USE/CLOBBER insns.
36487 PR tree-optimization/79095
36488 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
36489 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
36490 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
36491 if the operands are known to be not equal, then the resulting range
36493 (intersect_ranges): If the new range is ~[0,0] and the old range is
36494 wide, then prefer ~[0,0].
36495 * tree-vrp.c (overflow_comparison_p_1): New function.
36496 (overflow_comparison_p): New function.
36497 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
36498 if NAME is used in an overflow test.
36499 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
36500 overflow check that can be expressed as an equality test, then adjust
36501 ops to be that equality test.
36503 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36505 * config/s390/s390-builtin-types.def: Remove flags argument.
36506 * config/s390/s390.c (s390_init_builtins): Likewise.
36508 2017-02-14 Martin Liska <mliska@suse.cz>
36510 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
36511 vector. Fix trailing white spaces.
36513 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
36515 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
36518 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36520 PR rtl-optimization/68664
36521 * config/arm/arm.c (arm_sched_can_speculate_insn):
36522 New function. Declare prototype.
36523 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36525 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36527 PR rtl-optimization/68664
36528 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
36530 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
36532 2017-02-14 Amit Pawar <amit.pawar@amd.com>
36534 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
36535 max skip bytes for function, loop and jump.
36537 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
36539 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
36540 ABS_EXPR for gimple dump.
36542 2017-02-14 Jakub Jelinek <jakub@redhat.com>
36545 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
36547 PR tree-optimization/79408
36548 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
36549 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
36550 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
36551 also if rhs1 is INTEGER_CST.
36553 2017-02-14 Richard Biener <rguenther@suse.de>
36555 PR middle-end/79432
36556 * tree-into-ssa.c (insert_phi_nodes): When the function can
36557 have abnormal edges rewrite SSA names with broken use-def
36558 dominance out of SSA and register them for PHI insertion.
36560 2017-02-13 Martin Sebor <msebor@redhat.com>
36562 PR middle-end/79496
36563 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
36564 clearing info.nowrite flag when snprintf size argument is a range.
36566 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36568 * cprop.c (cprop_jump): Add missing space in string literal.
36569 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
36570 (get_constraint_for_component_ref): Likewise.
36571 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
36572 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
36573 * lra-constraints.c (process_alt_operands): Likewise.
36574 * ipa-inline.c (inline_small_functions): Likewise.
36575 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
36576 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
36577 * trans-mem.c (diagnose_tm_1_op): Likewise.
36578 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
36579 (grid_parallel_clauses_gridifiable): Likewise.
36581 * config/nvptx/mkoffload.c (process): Add space in between
36584 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
36585 "MOD4_SSE_REGS" and "ALL_REGS".
36587 * spellcheck.c (test_data): Add , in between "foo" and "food".
36589 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36592 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
36593 boundary crossing check and subsequent code generation agree.
36595 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36597 * config/aarch64/aarch64.c (has_memory_op): Delete.
36598 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
36601 2017-02-13 Jakub Jelinek <jakub@redhat.com>
36603 PR rtl-optimization/79388
36604 PR rtl-optimization/79450
36605 * combine.c (distribute_notes): When removing TEM_INSN for which
36606 corresponding dest has last value recorded, invalidate that last
36609 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36611 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
36612 of explicit '@'. Add missing assembly comment marker on branch costs
36615 2017-02-13 Nathan Sidwell <nathan@acm.org>
36617 * gengtype-lex.l (<in_struct>): Add '/'.
36619 2017-02-13 Martin Liska <mliska@suse.cz>
36622 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
36624 2017-02-13 Richard Biener <rguenther@suse.de>
36626 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
36628 * configure: Re-generate.
36629 * config.in: Likewise.
36630 * graphite-dependences.c: Simplify as if
36631 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
36632 * graphite-isl-ast-to-gimple.c: Likewise.
36633 * graphite-optimize-isl.c: Likewise.
36634 * graphite-poly.c: Likewise.
36635 * graphite-sese-to-poly.c: Likewise.
36636 * graphite.h: Likewise.
36637 * toplev.c: Include isl/version.h and use isl_version () for
36638 printing the ISL version.
36639 * doc/install.texi: Update ISL requirement.
36641 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
36643 * doc/standards.texi (Standards): Update reference to
36646 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
36648 * doc/extend.texi (Named Address Spaces): sourceware.org now
36650 * doc/install.texi (Binaries): Ditto.
36653 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
36655 * doc/cpp.texi: Replace "stringify"/"stringification" with C
36656 standard terminology "stringize"/"stringizing" throughout.
36657 * doc/cppinternals.texi: Likewise.
36659 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
36661 * doc/extend.texi: Fix some spelling mistakes and typos.
36662 * doc/invoke.texi: Likewise.
36664 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
36667 * params.def (inline-min-speedup) Change from 10 to 8.
36669 2017-02-11 Jakub Jelinek <jakub@redhat.com>
36671 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
36674 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
36677 * ipa-inline-analysis.c (get_minimal_bb): New function.
36678 (record_modified): Use it.
36679 (remap_edge_change_prob): Handle also ancestor functions.
36681 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
36683 * doc/contrib.texi (Contributors): Remove broken link into
36684 the Mauve CVS repository.
36686 2017-02-11 Jakub Jelinek <jakub@redhat.com>
36688 PR middle-end/79454
36689 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
36690 result computation whenever lhs doesn't have vector mode, not
36691 just when it has BLKmode.
36693 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
36695 * doc/makefile.texi (profiledbootstrap): Refer to the
36696 installation instructions only in textual form.
36698 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36701 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
36703 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
36705 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
36706 (Specific): Update mingw-w64 reference.
36708 (Specific): Remove broken link to Renesas RX processor.
36710 2017-02-10 Richard Biener <rguenther@suse.de>
36712 * toplev.c (process_options): Do not mention obsolete graphite
36713 options when printing sorry message about missing graphite support.
36714 Mention -floop-nest-optimize.
36716 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
36718 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
36719 (vtst_p16): Likewise.
36720 (vtstq_p8): Likewise.
36721 (vtstq_p16): Likewise.
36723 (vtstq_p64): Likewise.
36724 * config/arm/arm_neon.h (vgetq_lane_p64): New.
36725 (vset_lane_p64): New.
36726 (vsetq_lane_p64): New.
36728 2017-02-10 Jakub Jelinek <jakub@redhat.com>
36730 PR tree-optimization/79411
36731 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
36732 stmt operands are SSA_NAMEs used in abnormal phis.
36733 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
36736 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
36739 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
36742 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
36744 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
36746 2017-02-09 Jakub Jelinek <jakub@redhat.com>
36748 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
36752 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
36753 not arbitrary TREE_CONSTANT.
36756 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
36757 "omp declare target link" attribute unless is_global_var.
36758 * omp-offload.c (find_link_var_op): Likewise.
36760 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
36761 Chung-Lin Tang <cltang@codesourcery.com>
36763 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
36765 (gimplify_adjust_omp_clauses): Don't delete TILE.
36766 (gimplify_omp_for): Deal with TILE.
36767 * internal-fn.c (expand_GOACC_TILE): New function.
36768 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
36770 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
36771 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
36773 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
36774 avoid DIV for outermost collapse var.
36775 (expand_oacc_for): Insert tile element loop as needed. Adjust.
36776 Remove out of date comments, fix whitespace.
36777 * omp-general.c (omp_extract_for_data): Deal with tiling.
36778 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
36779 adjust OLF_DIM_BASE value.
36780 (struct omp_for_data): Add tiling field.
36781 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
36782 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
36783 for auto loops. Remove default auto determining, moved to
36784 oacc_loop_fixed_partitions.
36785 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
36786 stmts, add e_mask field.
36787 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
36788 (oacc_thread_numbers): Use oacc_dim_call.
36789 (oacc_xform_tile): New.
36790 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
36791 (finish_oacc_loop): Adjust for ifns vector.
36792 (oacc_loop_discover_walk): Append loop abstraction sites to list,
36793 add case for GOACC_TILE fns.
36794 (oacc_loop_xform_loop): Delete.
36795 (oacc_loop_process): Iterate over call list directly, and add
36796 handling for GOACC_TILE fns.
36797 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
36799 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
36800 vector partitioning to outer loops. Assign 2 partitions to loops
36801 when available. Add TILE handling.
36802 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
36803 (execite_oacc_device_lower): Process GOACC_TILE fns,
36804 ignore unknown specs.
36805 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
36806 * tree.c (omp_clause_num_ops): Adjust TILE ops.
36807 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
36809 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
36811 * configure.ac (ACX_BUGURL): Update.
36812 * configure: Regenerate.
36814 2017-02-09 Richard Biener <rguenther@suse.de>
36816 PR tree-optimization/69823
36817 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
36818 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
36820 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36822 * config/arc/arc-c.def: Add __NPS400__ definition.
36823 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
36824 (TARGET_NPS400): Define.
36826 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36828 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
36830 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
36831 pointer, arch_info.
36832 (arc_cpu_types): Fill the arch_info field with a pointer into the
36833 arc_arch_types table.
36834 (arc_selected_cpu): Declare.
36835 * config/arc/arc.c (arc_selected_cpu): Make global.
36836 (arc_selected_arch): Delete.
36837 (arc_base_cpu): Delete.
36838 (arc_override_options): Remove references to deleted variables,
36839 update access to arch information.
36840 (ARC_OPT): Update access to arch information.
36841 (ARC_OPTX): Likewise.
36842 * config/arc/arc.h (arc_base_cpu): Remove declaration.
36843 (TARGET_ARC600): Update access to arch information.
36844 (TARGET_ARC601): Likewise.
36845 (TARGET_ARC700): Likewise.
36846 (TARGET_EM): Likewise.
36847 (TARGET_HS): Likewise.
36848 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
36851 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
36854 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
36855 condition/operands for integer GE/LE/GEU/LEU operations.
36857 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
36859 PR translation/79397
36860 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
36863 2017-02-08 Martin Jambor <mjambor@suse.cz>
36866 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
36867 whether allocation happened.
36868 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
36869 nothing was allocated.
36871 2017-02-08 Jakub Jelinek <jakub@redhat.com>
36873 PR tree-optimization/79408
36874 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
36875 constant, but SSA_NAME with a known integer range, use the minimum
36876 of that range instead of op1 to determine if modulo can be replaced
36877 with its first operand.
36879 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36881 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
36883 2017-02-08 Richard Biener <rguenther@suse.de>
36885 PR tree-optimization/71824
36886 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
36887 Check all loops contained in the merged region.
36889 2017-02-07 Andrew Pinski <apinski@cavium.com>
36891 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
36893 2017-02-07 Andrew Pinski <apinski@cavium.com>
36895 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
36896 (thunderxt88): Likewise.
36897 (thunderxt81): Disable LSE and change v8.1 to v8.
36898 (thunderxt83): Likewise.
36900 2017-02-07 Jakub Jelinek <jakub@redhat.com>
36901 Richard Biener <rguenther@suse.de>
36903 PR middle-end/79399
36904 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
36905 type from int to size_t.
36906 * ira-costs.c (struct_costs_size): Change type from int to size_t.
36908 2017-02-07 Jakub Jelinek <jakub@redhat.com>
36910 PR rtl-optimization/79386
36911 * cprop.c (bypass_conditional_jumps): Initialize
36912 bypass_last_basic_block already before splitting bbs after
36913 unconditional traps...
36914 (bypass_conditional_jumps): ... rather than here.
36917 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
36918 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
36919 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
36920 fix -masm=intel patterns.
36922 2017-02-07 Richard Biener <rguenther@suse.de>
36924 PR tree-optimization/79256
36925 PR middle-end/79278
36926 * builtins.c (get_object_alignment_2): Use min_align_of_type
36927 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
36928 and ADJUST_FIELD_ALIGN.
36930 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
36932 * doc/tm.texi: Regenerate.
36933 * stor-layout.c (layout_decl): Adjust.
36934 (update_alignment_for_field): Likewise.
36935 (place_field): Likewise.
36936 (min_align_of_type): Likewise.
36937 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
36938 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
36939 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
36940 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
36941 * config/frv/frv.c (frv_adjust_field_align): Likewise.
36942 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
36943 * config/i386/i386.c (x86_field_alignment): Likewise.
36944 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
36945 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
36946 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
36947 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
36948 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
36949 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
36953 2017-01-30 Richard Biener <rguenther@suse.de>
36955 PR tree-optimization/79256
36956 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
36957 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
36960 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
36962 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
36963 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
36964 builtins to SImode and emit a zero-extend, if necessary.
36966 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36968 * docs/invoke.texi (RISC-V Options): Alphabetize.
36970 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36972 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
36975 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36977 * config/riscv/riscv.c: New file.
36978 * common/config/riscv/riscv-common.c: Likewise.
36979 * config.gcc: Likewise.
36980 * config/riscv/constraints.md: Likewise.
36981 * config/riscv/elf.h: Likewise.
36982 * config/riscv/generic.md: Likewise.
36983 * config/riscv/linux.h: Likewise.
36984 * config/riscv/multilib-generator: Likewise.
36985 * config/riscv/peephole.md: Likewise.
36986 * config/riscv/pic.md: Likewise.
36987 * config/riscv/predicates.md: Likewise.
36988 * config/riscv/riscv-builtins.c: Likewise.
36989 * config/riscv/riscv-c.c: Likewise.
36990 * config/riscv/riscv-ftypes.def: Likewise.
36991 * config/riscv/riscv-modes.def: Likewise.
36992 * config/riscv/riscv-opts.h: Likewise.
36993 * config/riscv/riscv-protos.h: Likewise.
36994 * config/riscv/riscv.h: Likewise.
36995 * config/riscv/riscv.md: Likewise.
36996 * config/riscv/riscv.opt: Likewise.
36997 * config/riscv/sync.md: Likewise.
36998 * config/riscv/t-elf-multilib: Likewise.
36999 * config/riscv/t-linux: Likewise.
37000 * config/riscv/t-linux-multilib: Likewise.
37001 * config/riscv/t-riscv: Likewise.
37002 * configure.ac: Likewise.
37003 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
37004 Waterman as RISC-V maintainers.
37005 * doc/install.texi: Add RISC-V entries.
37006 * doc/invoke.texi: Add RISC-V options section.
37007 * doc/md.texi: Add RISC-V constraints section.
37008 * configure: Regenerated.
37010 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
37013 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
37014 false values to be constant vectors with all 0 or all 1 bits set.
37015 (vcondu<mode><mode>): Likewise.
37016 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
37018 (fpmask_comparison_operator): Update comment.
37019 (vecint_comparison_operator): New predicate.
37020 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
37021 vector conditionals when the true and false values are constant
37022 vectors with all 0 bits or all 1 bits set.
37024 2017-02-06 Martin Sebor <msebor@redhat.com>
37026 PR tree-optimization/79376
37027 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
37029 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
37031 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
37032 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
37033 to simplify split condition.
37035 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37037 * omp-expand.c (oxpand_omp_atomic_fetch_op,
37038 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
37041 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
37043 PR rtl-optimization/68664
37044 * target.def (can_speculate_insn): New hook.
37045 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
37046 * doc/tm.texi: Regenerate.
37047 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
37048 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
37049 (rs6000_sched_can_speculate_insn): New function.
37051 2017-02-06 Jakub Jelinek <jakub@redhat.com>
37053 PR tree-optimization/79284
37054 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
37055 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
37056 vectorizable_mask_load_store, vectorizable_operation,
37057 vect_is_simple_cond, get_same_sized_vectype): Use it instead
37058 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
37059 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
37060 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
37061 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
37062 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
37063 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
37064 is_gimple_assign (stmt). Replace another such test with
37065 is_gimple_assign (stmt).
37067 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37070 * config/avr/avr.c (rtl-iter.h): Include it.
37071 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
37072 (avr_legitimate_combined_insn): ...and implementation.
37074 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
37076 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
37077 * config/s390/s390.c (s390_const_operand_ok)
37078 (s390_canonicalize_comparison, s390_extract_part)
37079 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
37080 (s390_contiguous_bitmask_p, s390_rtx_costs)
37081 (legitimize_pic_address): Likewise.
37082 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
37083 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
37084 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
37085 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
37086 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
37088 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
37090 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
37091 REGNO($0) == REGNO($1).
37093 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37095 * config/s390/linux.h(SIZE_TYPE): Add comment.
37097 2017-02-06 Julian Brown <julian@codesourcery.com>
37098 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37099 Virendra Pathak <virendra.pathak@broadcom.com>
37101 * config/aarch64/aarch64-cores.def: Change the scheduler
37103 * config/aarch64/aarch64.md: Include thunderx2t99.md.
37104 * config/aarch64/thunderx2t99.md: New file.
37106 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37108 * doc/standards.texi (Go Language): Update link to language
37111 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37113 * tree-eh.c (lower_resx): Sanitize profile.
37114 (cleanup_empty_eh_move_lp): Likewise.
37116 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
37119 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
37121 * cfgloopmanip.h (loop_version): Update prototype.
37122 * modulo-sched.c (sms_schedule): Update call of loop_version.
37123 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
37124 * tree-parloops.c (gen_parallel_loop): Likewise.
37125 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
37126 * tree-ssa-loop-split.c (split_loop): Likewise.
37127 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
37128 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
37130 2017-02-05 Martin Liska <mliska@suse.cz>
37133 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
37135 (print_operand_address): Initialize a struct to zero.
37137 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37139 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
37140 garbage collector only in textual form.
37142 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
37144 * doc/extend.texi (x86 specific memory model extensions for
37145 transactional memory): Simplify a phrase.
37147 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
37150 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
37151 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
37152 (atomic_storedi_1): Likewise.
37154 2017-02-04 Jakub Jelinek <jakub@redhat.com>
37156 PR tree-optimization/79338
37157 * tree-parloops.c (gather_scalar_reductions): Don't call
37158 vect_analyze_loop_form for loop->inner before destroying loop's
37161 2017-02-03 Martin Sebor <msebor@redhat.com>
37163 PR tree-optimization/79327
37164 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
37165 when precision has resulted in leading zeros.
37166 (format_integer): Adjust the likely counter to assume an unknown
37167 argument that may be zero is non-zero.
37169 2017-02-03 Jason Merrill <jason@redhat.com>
37172 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
37173 avoid copying non-taken branch.
37175 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37177 PR tree-optimization/79340
37178 * tree-vect-loop.c (vectorizable_reduction): Release
37179 vec_defs elements after safe_splicing them into other vectors.
37182 PR tree-optimization/79327
37183 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
37184 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
37186 (format_integer): Use wide_int_to_tree instead of build_int_cst
37187 + to_?hwi. If argmin is NULL, just set argmin and argmax to
37188 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
37189 of shortest and longest sequence.
37191 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
37193 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
37194 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
37196 2017-02-03 Walter Lee <walt@tilera.com>
37199 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
37200 after initial stackframe link reg save.
37201 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
37203 2017-02-03 Jakub Jelinek <jakub@redhat.com>
37206 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
37207 wu for stxssp alternative.
37209 2017-02-03 Martin Sebor <msebor@redhat.com>
37211 PR tree-optimization/79352
37212 * gimple-fold.c (get_range_strlen): Add argument.
37213 (get_range_strlen): Change return type to bool.
37214 (get_maxval_strlen): Pass in a dummy argument.
37215 * gimple-fold.h (get_range_strlen): Change return type to bool.
37216 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
37217 * tree.h (array_at_struct_end_p): Add argument.
37218 * tree.c (array_at_struct_end_p): Handle it.
37220 2017-02-03 Martin Liska <mliska@suse.cz>
37223 * multiple_target.c (create_dispatcher_calls): Redirect edge
37224 from a caller of a dispatcher.
37225 (expand_target_clones): Make the clones local.
37226 (ipa_target_clone): Do both target clones and resolvers.
37227 (ipa_dispatcher_calls): Remove the pass.
37228 (pass_dispatcher_calls::gate): Likewise.
37229 (make_pass_dispatcher_calls): Likewise.
37230 * passes.def (pass_target_clone): Put as very first IPA early
37233 2017-02-03 Martin Liska <mliska@suse.cz>
37235 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
37236 in case of a function with ifunc attribute.
37238 2017-02-03 Martin Liska <mliska@suse.cz>
37240 * cgraph.c (cgraph_node::dump): Dump function version info.
37241 * symtab.c (symtab_node::dump_base): Add missing new line.
37243 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37245 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
37246 (ifcombine_ifandif): Use it.
37248 2017-02-03 Martin Liska <mliska@suse.cz>
37250 * doc/invoke.texi: Document default value for
37251 use-after-scope-direct-emission-threshold.
37253 2017-02-03 Martin Liska <mliska@suse.cz>
37255 PR tree-optimization/79339
37256 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
37257 (format_floating): Likewise.
37259 2017-02-03 Martin Liska <mliska@suse.cz>
37262 * ipa-prop.c (ipa_node_params_t::insert): Remove current
37264 (ipa_node_params_t::remove): Likewise.
37265 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
37266 initialization from removed ipa_node_params_t::insert.
37267 (ipa_node_params::~ipa_node_params): Move from removed
37268 ipa_node_params_t::release.
37269 * symbol-summary.h (symbol_summary::m_released): New member.
37270 Do not release a summary twice. Do not allow to call finalizer
37271 for types of a summary that live in GGC memory.
37273 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
37275 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
37278 2017-02-02 Martin Sebor <msebor@redhat.com>
37280 PR middle-end/79275
37281 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
37282 (format_string): Tighten up the range of output for non-constant
37283 strings and correct the expected range for wide non-constant strings.
37285 2017-02-02 Martin Sebor <msebor@redhat.com>
37287 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
37289 PR middle-end/32003
37290 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
37292 (-fdump-tree-@var): Add to index and document how to come up
37293 with pass-specific option and dump file names.
37294 (-fdump-passes): Clarify where to look for output.
37296 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
37298 PR middle-end/77445
37299 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
37300 statistics of the analyzed path; allow threading for speed when
37301 any of BBs along the path are optimized for speed.
37303 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
37305 PR middle-end/78468
37306 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
37307 settings of the virtual registers.
37310 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
37312 * explow.c (get_dynamic_stack_size): Take known alignment of stack
37313 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
37316 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37318 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
37319 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
37321 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37323 * config/s390/s390.md: Add missing comments with the expanded
37325 * config/s390/vector.md: Likewise.
37326 * config/s390/vx-builtins.md: Likewise.
37328 2017-02-02 Jakub Jelinek <jakub@redhat.com>
37331 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
37332 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
37333 conditions on a single line.
37335 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37337 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37338 __S390_VX__ to __VX__.
37340 2017-02-01 Andrew Pinski <apinski@cavium.com>
37342 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
37343 stmt_info to record_stmt_cost.
37344 (vect_get_known_peeling_cost): Pass stmt_info if known to
37346 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
37347 cpu_vector_cost field into
37348 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
37349 field into vec_int_stmt_cost and vec_fp_stmt_cost.
37350 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
37351 splitting of scalar_stmt_cost and vec_stmt_cost.
37352 (thunderx_vector_cost): Likewise.
37353 (cortexa57_vector_cost): LIkewise.
37354 (exynosm1_vector_cost): Likewise.
37355 (xgene1_vector_cost): Likewise.
37356 (thunderx2t99_vector_cost): Improve after the splitting of the two
37358 (aarch64_builtin_vectorization_cost): Update for the splitting of
37359 scalar_stmt_cost and vec_stmt_cost.
37361 2017-02-01 Torvald Riegel <triegel@redhat.com>
37362 Richard Henderson <rth@redhat.com>
37364 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
37365 conditional on existance of a fast atomic load.
37366 * optabs-query.c (can_atomic_load_p): New function.
37367 * optabs-query.h (can_atomic_load_p): Declare it.
37368 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
37369 no fast atomic load is available for the particular size of access.
37370 (expand_atomic_compare_and_swap): Likewise.
37371 (expand_atomic_load): Likewise.
37372 (expand_atomic_store): Likewise.
37373 (expand_atomic_fetch_op): Likewise.
37374 * testsuite/lib/target-supports.exp
37375 (check_effective_target_sync_int_128): Remove x86 because it provides
37376 no fast atomic load.
37377 (check_effective_target_sync_int_128_runtime): Likewise.
37379 2017-02-01 Richard Biener <rguenther@suse.de>
37381 * graphite.c: Include tree-vectorizer.h for find_loop_location.
37382 (graphite_transform_loops): Provide opt-info for optimized nests.
37383 * tree-parloop.c (parallelize_loops): Provide opt-info for
37384 parallelized loops.
37386 2017-02-01 Richard Biener <rguenther@suse.de>
37388 PR middle-end/79315
37389 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
37390 was not set before.
37392 2017-02-01 Richard Biener <rguenther@suse.de>
37394 PR tree-optimization/71824
37395 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
37396 Verify the loops are valid in the merged SESE region.
37397 (scop_detection::can_represent_loop_1): Check analyzing the
37398 evolution of the number of iterations in the region succeeds.
37400 2017-01-31 Ian Lance Taylor <iant@golang.org>
37402 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
37403 REG_ARGS_SIZE note to 32-bit push insns and call insn.
37405 2017-01-31 David Malcolm <dmalcolm@redhat.com>
37407 PR preprocessor/79210
37408 * input.c (get_substring_ranges_for_loc): Replace line_width
37409 assertion with error-handling.
37411 2017-01-31 Richard Biener <rguenther@suse.de>
37413 PR tree-optimization/77318
37414 * graphite-sese-to-poly.c (extract_affine): Fix assert.
37415 (create_pw_aff_from_tree): Take loop parameter.
37416 (add_condition_to_pbb): Pass loop of the condition to
37417 create_pw_aff_from_tree.
37419 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37421 * config/s390/s390.c (s390_asan_shadow_offset): New function.
37422 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
37424 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
37428 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
37430 (convert_int_to_float128): Likewise.
37431 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
37432 (convert_int_to_float128): Likewise.
37433 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
37434 (UNSPEC_IEEE128_CONVERT): Likewise.
37435 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
37436 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
37437 Use local variables for IBM extended format.
37438 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
37439 (fix_trunc<mode>si2_fprs): Likewise.
37440 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
37441 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
37442 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
37443 to know that we can now have integers of all sizes in vector
37445 (fix<uns>_<mode>di2_hw): Likewise.
37446 (float<uns>_<mode>si2_hw): Likewise.
37447 (fix_<mode>si2_hw): Likewise.
37448 (fixuns_<mode>si2_hw): Likewise.
37449 (float<uns>_<mode>di2_hw): Likewise.
37450 (float_<mode>di2_hw): Likewise.
37451 (float_<mode>si2_hw): Likewise.
37452 (floatuns_<mode>di2_hw): Likewise.
37453 (floatuns_<mode>si2_hw): Likewise.
37454 (xscvqp<su>wz_<mode>): Delete, no longer used.
37455 (xscvqp<su>dz_<mode>): Likewise.
37456 (xscv<su>dqp_<mode>): Likewise.
37457 (ieee128_mfvsrd_64bit): Likewise.
37458 (ieee128_mfvsrd_32bit): Likewise.
37459 (ieee128_mfvsrwz): Likewise.
37460 (ieee128_mtvsrw): Likewise.
37461 (ieee128_mtvsrd_64bit): Likewise.
37462 (ieee128_mtvsrd_32bit): Likewise.
37464 2017-01-31 Martin Liska <mliska@suse.cz>
37467 * ipa-prop.c (ipa_free_all_node_params): Call release method
37468 instead of ~sumbol_summary to not to trigger double times
37471 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
37473 PR tree-optimization/71691
37474 * bitmap.h (class auto_bitmap): New.
37475 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
37476 is_maybe_undefined instead of ssa_undefined_value_p.
37478 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37480 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
37481 __S390_ARCH_LEVEL__ to __ARCH__.
37483 2017-01-31 Jakub Jelinek <jakub@redhat.com>
37485 PR tree-optimization/79267
37486 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
37487 if should_remove_lhs_p is true.
37489 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
37492 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
37493 (add_alignment_attribute): New.
37494 (base_type_die): Add alignment attribute.
37495 (subrange_type_die): Likewise.
37496 (modified_type_die): Likewise.
37497 (gen_array_type_die): Likewise.
37498 (gen_descr_array_type_die: Likewise.
37499 (gen_enumeration_type_die): Likewise.
37500 (gen_subprogram_die): Likewise.
37501 (gen_variable_die): Likewise.
37502 (gen_field_die): Likewise.
37503 (gen_ptr_to_mbr_type_die): Likewise.
37504 (gen_struct_or_union_type_die): Likewise.
37505 (gen_subroutine_type_die): Likewise.
37506 (gen_typedef_die): Likewise.
37507 (base_type_cmp): Compare alignment attribute.
37509 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37512 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
37513 (setb_unsigned) New pattern for setb with CCUNS.
37514 * config/rs6000/rs6000.c (expand_block_compare): Use a different
37515 subfc./subfe sequence to avoid overflow problems. Generate a
37516 shorter sequence with cmpld/setb for power9.
37517 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
37518 for generating subfc. instruction.
37519 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
37520 now uses this instruction.
37522 2017-01-30 Ian Lance Taylor <iant@google.com>
37525 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
37526 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
37528 2017-01-30 Martin Sebor <msebor@redhat.com>
37530 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
37531 Move constant to the right of a relational operator.
37532 (get_mpfr_format_length, format_character, format_string): Ditto.
37533 (should_warn_p, maybe_warn): Same.
37535 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
37537 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
37540 * asan.c (get_translation_unit_decl): Remove function.
37541 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
37543 2017-01-30 Martin Liska <mliska@suse.cz>
37545 PR gcov-profile/79259
37546 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
37547 -fprofile-generate.
37549 2017-01-30 Martin Liska <mliska@suse.cz>
37552 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
37553 Initialize variables with NULL value.
37555 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
37558 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
37560 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
37562 2017-01-30 Richard Biener <rguenther@suse.de>
37564 PR tree-optimization/79276
37565 * tree-vrp.c (process_assert_insertions): Properly adjust common
37566 when removing a duplicate.
37568 2017-01-30 Richard Biener <rguenther@suse.de>
37570 PR tree-optimization/79256
37571 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
37572 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
37574 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
37576 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
37579 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
37580 ("*r<noxa>sbg_<mode>_sll_bitmask")
37581 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
37582 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
37583 Use contiguous_bitmask_nowrap_operand.
37585 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37588 * config/rs6000/altivec.h (vec_xl): Revise #define.
37589 (vec_xst): Likewise.
37591 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
37593 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
37595 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
37597 PR rtl-optimization/79194
37598 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
37599 traps before call to bypass_conditional_jumps.
37601 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37603 PR tree-optimization/71374
37604 * lra-constraints.c (check_conflict_input_operands): New.
37605 (match_reload): Use it.
37607 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
37610 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
37611 account to calculate conflict_set.
37613 2017-01-27 Bin Cheng <bin.cheng@arm.com>
37615 PR rtl-optimization/78559
37616 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
37617 other_insn in combine.
37619 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
37621 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
37622 uint16_type_node for BT_UINT16.
37624 2017-01-27 David Malcolm <dmalcolm@redhat.com>
37626 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
37627 "RTL Tests" to menu.
37628 (GIMPLE Tests): New node.
37629 (RTL Tests): New node.
37631 2017-01-27 Richard Biener <rguenther@suse.de>
37633 PR tree-optimization/79245
37634 * tree-loop-distribution.c (distribute_loop): Apply cost
37635 modeling also to detected patterns.
37637 2017-01-27 Richard Biener <rguenther@suse.de>
37639 PR tree-optimization/71433
37640 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
37641 (compare_assert_loc): New function.
37642 (process_assert_insertions): Sort and optimize assert locations
37643 to remove duplicates and push down identical assertions on
37644 edges to their destination block.
37646 2017-01-27 Richard Biener <rguenther@suse.de>
37648 PR tree-optimization/79244
37649 * tree-vrp.c (remove_range_assertions): Forcefully propagate
37650 out SSA names even if abnormal.
37652 2017-01-27 Jakub Jelinek <jakub@redhat.com>
37654 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
37655 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
37656 instead of MPFR_RNDN.
37658 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
37661 * arm.c (arm_option_override): Don't call build_target_option_node
37662 until after doing all option overrides.
37663 (arm_valid_target_attribute_tree): Likewise.
37665 2017-01-27 Martin Liska <mliska@suse.cz>
37667 * doc/invoke.texi (-fprofile-arcs): Document profiling support
37668 for {cd}tors and C++ {cd}tors.
37670 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
37672 * config/s390/s390.md ("*setmem_long_and")
37673 ("*setmem_long_and_31z"): Use zero_extend instead of and.
37675 2017-01-26 Martin Sebor <msebor@redhat.com>
37677 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
37680 2017-01-26 Martin Sebor <msebor@redhat.com>
37682 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
37683 HAVE_DFmode before using XFmode or DFmode.
37684 (parse_directive): Avoid using the z length modifier to avoid
37685 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
37687 PR middle-end/78703
37688 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
37689 to accept adjustment as an array.
37690 (get_int_range): New function.
37691 (struct directive): Make width and prec arrays.
37692 (directive::set_width, directive::set_precision): Call get_int_range.
37693 (format_integer, format_floating): Handle width and precision ranges.
37694 (format_string, parse_directive): Same.
37696 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37699 * dwarf2out.c (generate_skeleton_bottom_up): For children with
37700 comdat_type_p set, just clone them, but keep the children in the
37704 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
37705 which have direct callers with -fvar-tracking-assignments enabled
37707 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
37708 inside of type units.
37710 2017-01-26 Martin Sebor <msebor@redhat.com>
37712 PR middle-end/78703
37713 * gimple-ssa-sprintf.c (struct result_range): Add likely and
37715 (struct format_result): Replace number_chars, number_chars_min,
37716 and number_chars_max with a single member of struct result_range.
37718 (format_result::operator+=): Adjust.
37719 (struct fmtresult): Remove bounded. Handle likely and unlikely
37721 (fmtresult::adjust_for_width_or_precision): New function.
37722 (fmtresult:type_max_digits): New function.
37723 (bytes_remaining): Handle likely and unlikely counters.
37724 (min_bytes_remaining): Remove.
37725 (format_percent): Simplify.
37726 (format_integer, format_floating): Set likely and unlikely counters.
37727 (get_string_length, format_character, format_string): Same.
37728 (format_plain, should_warn_p): New function.
37729 (maybe_warn): Call should_warn_p. Update diagnostic messages
37730 and handle those for all directives, including plain strings.
37731 (format_directive): Handle likely and unlikely counters.
37732 Remove unnecessary quoting from diagnostics. Add an informational
37734 (add_bytes): Remove.
37735 (pass_sprintf_length::compute_format_length): Simplify.
37736 (try_substitute_return_value): Handle likely and unlikely counters.
37738 2017-01-26 Carl Love <cel@us.ibm.com>
37740 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
37741 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
37743 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
37746 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
37747 endianess for subregs into account.
37748 * lra-constraints.c (lra_constraints): Do risky transformations
37749 always on the first iteration.
37750 * lra-lives.c (check_pseudos_live_through_calls): Add arg
37751 last_call_used_reg_set.
37752 (process_bb_lives): Define and use last_call_used_reg_set.
37753 * lra.c (lra): Always continue after lra_constraints on the first
37756 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
37758 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
37760 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
37762 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37764 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
37765 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
37766 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
37767 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
37768 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
37769 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
37770 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
37771 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
37772 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
37774 2017-01-26 Marek Polacek <polacek@redhat.com>
37777 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
37778 for the third operand.
37780 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37782 PR middle-end/79236
37783 * omp-low.c (struct omp_context): Add simt_stmt field.
37784 (scan_omp_for): Return omp_context *.
37785 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
37786 context to the _simt_ SIMD stmt.
37787 (lower_omp_for): For combined SIMD with sibling _simt_
37788 SIMD, make sure to use the same decls in _looptemp_
37789 clauses as in the sibling.
37791 2017-01-26 David Sherwood <david.sherwood@arm.com>
37793 PR middle-end/79212
37794 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
37797 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37800 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
37801 emit fld b; fld a; if possible.
37803 * brig-builtins.def: Update copyright years.
37804 * config/arm/arm_acle_builtins.def: Update copyright years.
37806 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
37809 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
37810 constraint instead of o for the stxsd instruction.
37812 2017-01-25 Carl Love <cel@us.ibm.com>
37814 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
37815 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
37817 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
37819 * doc/invoke.texi (C++ Dialect Options): Fix typo.
37821 2017-01-25 Richard Biener <rguenther@suse.de>
37823 PR tree-optimization/69264
37824 * target.def (vector_alignment_reachable): Improve documentation.
37825 * doc/tm.texi: Regenerate.
37826 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
37828 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
37829 earlier changes with respect to TYPE_USER_ALIGN.
37830 (vector_alignment_reachable_p): Likewise. Improve dumping.
37832 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37835 * config/arm/arm.md (xordi3): Force constant operand into a register
37838 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37840 * doc/invoke.texi (-fstore-merging): Correct default optimization
37841 levels at which it is enabled.
37842 (-O): Move -fstore-merging from list to...
37845 2017-01-25 Richard Biener <rguenther@suse.de>
37848 * omp-expand.c: Include debug.h.
37849 (expand_omp_taskreg): Make sure to generate early debug before
37850 outlining anything from a function.
37851 (expand_omp_target): Likewise.
37852 (grid_expand_target_grid_body): Likewise.
37854 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
37857 * asan.c (get_translation_unit_decl): New function.
37858 (asan_add_global): Extract modules file name from globals
37859 TRANSLATION_UNIT_DECL name.
37861 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
37864 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
37865 for long calls with APCS frame and VFP.
37867 2017-01-24 David Malcolm <dmalcolm@redhat.com>
37869 * cfg.c (original_copy_tables_initialized_p): New function.
37870 * cfg.h (original_copy_tables_initialized_p): New decl.
37871 * cfgrtl.c (relink_block_chain): Guard the call to
37872 free_original_copy_tables with a call to
37873 original_copy_tables_initialized_p.
37874 * cgraph.h (symtab_node::native_rtl_p): New decl.
37875 * cgraphunit.c (symtab_node::native_rtl_p): New function.
37876 (symtab_node::needed_p): Don't assert for early assembly output
37877 for __RTL functions.
37878 (cgraph_node::finalize_function): Set "force_output" for __RTL
37880 (cgraph_node::analyze): Bail out early for __RTL functions.
37881 (analyze_functions): Update assertion to support __RTL functions.
37882 (cgraph_node::expand): Bail out early for __RTL functions.
37883 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
37885 * function.h (struct function): Update comment for field
37887 * gimple-expr.c: Include "tree-pass.h".
37888 (gimple_has_body_p): Return false for __RTL functions.
37889 * Makefile.in (OBJS): Add run-rtl-passes.o.
37890 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
37892 (gcc::pass_manager::get_clean_slate): New accessor.
37893 * passes.c: Include "insn-addr.h".
37894 (should_skip_pass_p): Add logging. Update logic for running
37895 "expand" to be compatible with both __GIMPLE and __RTL. Guard
37896 property-provider override so it is only done for gimple passes.
37898 (skip_pass): New function.
37899 (execute_one_pass): Call skip_pass when skipping passes.
37900 * read-md.c (md_reader::read_char): Support filtering
37901 the input to a subset of line numbers.
37902 (md_reader::md_reader): Initialize fields
37903 m_first_line and m_last_line.
37904 (md_reader::read_file_fragment): New function.
37905 * read-md.h (md_reader::read_file_fragment): New decl.
37906 (md_reader::m_first_line): New field.
37907 (md_reader::m_last_line): New field.
37908 * read-rtl-function.c (function_reader::create_function): Only
37909 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
37910 curr_properties. Set DECL_INITIAL to a dummy block.
37911 (read_rtl_function_body_from_file_range): New function.
37912 * read-rtl-function.h (read_rtl_function_body_from_file_range):
37914 * run-rtl-passes.c: New file.
37915 * run-rtl-passes.h: New file.
37917 2017-01-24 Jeff Law <law@redhat.com>
37919 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
37922 2017-01-24 Bin Cheng <bin.cheng@arm.com>
37924 PR tree-optimization/79159
37925 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
37926 (record_nonwrapping_iv): Improve boundary using above function if no
37927 value range information.
37929 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
37930 Martin Jambor <mjambor@suse.cz>
37932 * brig-builtins.def: New file.
37933 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
37934 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
37935 (DEF_HSAIL_SAT_BUILTIN): Likewise.
37936 (DEF_HSAIL_INTR_BUILTIN): Likewise.
37937 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
37938 * builtin-types.def (BT_INT8): New.
37939 (BT_INT16): Likewise.
37940 (BT_UINT8): Likewise.
37941 (BT_UINT16): Likewise.
37942 (BT_FN_ULONG): Likewise.
37943 (BT_FN_UINT_INT): Likewise.
37944 (BT_FN_UINT_ULONG): Likewise.
37945 (BT_FN_UINT_LONG): Likewise.
37946 (BT_FN_UINT_PTR): Likewise.
37947 (BT_FN_ULONG_PTR): Likewise.
37948 (BT_FN_INT8_FLOAT): Likewise.
37949 (BT_FN_INT16_FLOAT): Likewise.
37950 (BT_FN_UINT32_FLOAT): Likewise.
37951 (BT_FN_UINT16_FLOAT): Likewise.
37952 (BT_FN_UINT8_FLOAT): Likewise.
37953 (BT_FN_UINT64_FLOAT): Likewise.
37954 (BT_FN_UINT16_UINT32): Likewise.
37955 (BT_FN_UINT32_UINT16): Likewise.
37956 (BT_FN_UINT16_UINT16_UINT16): Likewise.
37957 (BT_FN_INT_PTR_INT): Likewise.
37958 (BT_FN_UINT_PTR_UINT): Likewise.
37959 (BT_FN_LONG_PTR_LONG): Likewise.
37960 (BT_FN_ULONG_PTR_ULONG): Likewise.
37961 (BT_FN_VOID_UINT64_UINT64): Likewise.
37962 (BT_FN_UINT8_UINT8_UINT8): Likewise.
37963 (BT_FN_INT8_INT8_INT8): Likewise.
37964 (BT_FN_INT16_INT16_INT16): Likewise.
37965 (BT_FN_INT_INT_INT): Likewise.
37966 (BT_FN_UINT_FLOAT_UINT): Likewise.
37967 (BT_FN_FLOAT_UINT_UINT): Likewise.
37968 (BT_FN_ULONG_UINT_UINT): Likewise.
37969 (BT_FN_ULONG_UINT_PTR): Likewise.
37970 (BT_FN_ULONG_ULONG_ULONG): Likewise.
37971 (BT_FN_UINT_UINT_UINT): Likewise.
37972 (BT_FN_VOID_UINT_PTR): Likewise.
37973 (BT_FN_UINT_UINT_PTR: Likewise.
37974 (BT_FN_UINT32_UINT64_PTR): Likewise.
37975 (BT_FN_INT_INT_UINT_UINT): Likewise.
37976 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
37977 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
37978 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
37979 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
37980 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
37981 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
37982 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
37983 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
37984 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
37985 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
37986 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
37987 * doc/frontends.texi: List BRIG FE.
37988 * doc/install.texi (Testing): Add BRIG tesring requirements.
37989 * doc/invoke.texi (Overall Options): Mention BRIG.
37990 * doc/standards.texi (Standards): Doucment BRIG HSA version.
37992 2017-01-24 Richard Biener <rguenther@suse.de>
37994 PR translation/79208
37995 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
37997 2017-01-24 Martin Jambor <mjambor@suse.cz>
38000 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
38001 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
38002 and known_contexts.
38004 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
38006 PR middle-end/79123
38007 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
38008 casts from signed to unsigned really don't have a range.
38010 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
38012 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
38013 GMP_RNDx for compatiblity.
38015 2017-01-24 Martin Liska <mliska@suse.cz>
38018 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
38019 that would prevent us to call alloca with -1 as argument.
38021 2017-01-24 Jakub Jelinek <jakub@redhat.com>
38023 * dwarf2out.c (output_compilation_unit_header, output_file_names):
38024 Avoid -Wformat-security warning.
38026 2017-01-23 Andrew Pinski <apinski@cavium.com>
38028 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
38031 2017-01-23 Martin Sebor <msebor@redhat.com>
38033 PR middle-end/78703
38034 * gimple-ssa-sprintf.c (warn_level): New global.
38035 (format_integer): Use it here and throughout the rest of the file.
38036 Use the same switch to compute sign as base.
38037 (maybe_warn): New function.
38038 (format_directive): Factor out warnings into maybe_warn.
38039 Add debugging output. Use warn_level.
38040 (add_bytes): Use warn_level.
38041 (pass_sprintf_length::compute_format_length): Add debugging output.
38042 (try_substitute_return_value): Same.
38043 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
38045 PR middle-end/78703
38046 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
38047 (struct fmtresult, format_integer, format_floating): Adjust.
38048 (fmtresult::fmtresult): Set max correctly in two argument ctor.
38049 (get_string_length, format_string,format_directive): Same.
38050 (pass_sprintf_length::compute_format_length): Same.
38051 (try_substitute_return_value): Simplify slightly.
38053 PR middle-end/78703
38054 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
38055 (fmtresult::operator+=): Outlined.
38056 (struct fmtresult): Add ctors.
38057 (struct conversion_spec): Rename...
38058 (struct directive): ...to this. Add and remove data members.
38059 (directive::set_width, directive::set_precision): New functions.
38060 (format_percent): Use fmtresult ctor.
38061 (get_width_and_precision): Remove.
38062 (format_integer): Make naming changes. Avoid computing width and
38064 (format_floating): Same. Adjust indentation.
38065 (format_character, format_none): New functions.
38066 (format_string): Moved character handling to format_character.
38067 (format_directive): Remove arguments, change return type.
38068 (parse_directive): New function.
38069 (pass_sprintf_length::compute_format_length): Move directive
38070 parsing to parse_directive.
38072 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38074 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
38075 (assign_assembler_name_if_needed): ... this.
38076 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
38077 (assign_assembler_name_if_needed): ... this.
38078 (free_lang_data_in_cgraph): Adjust callers.
38079 * cgraphunit.c (cgraph_node::analyze): Likewise.
38080 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
38083 2017-01-23 Richard Biener <rguenther@suse.de>
38085 PR tree-optimization/79088
38086 PR tree-optimization/79188
38087 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
38088 resetting loop bounds after last path deletion. Reset loop
38089 bounds of the target loop, make code match the comments.
38090 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
38091 Make sure loops need no fixups.
38093 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
38095 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
38096 exponent support with double type for first argument.
38097 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
38098 type returned by __builtin_vec_extract_sig,
38099 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
38100 functions from "vector int" to "vector unsigned int" or from
38101 "vector long long int" to "vector unsigned long long int".
38102 Changed type returned by __builtin_vec_extract_exp,
38103 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
38104 functions from "vector int" to "vector unsigned int" or from
38105 "vector long long int" to "vector unsigned long long int".
38106 Changed return type of __builtin_vec_test_data_class,
38107 __builtin_vec_test_data_class_sp, and
38108 __builtin_vec_test_data_class_dp from "vector int" to
38109 "vector bool int" or from "vector long long int" to "vector bool
38110 long long int" and changed second argument type from "unsigned
38111 int" to "int". Added new overloaded function forms "vector float
38112 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
38113 "vector float __builtin_vec_insert_exp_sp (vector float, vector
38114 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
38115 double, vector unsigned long long int)" and "vector double
38116 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
38117 long int)". Changed return type of
38118 __builtin_scalar_test_data_class and
38119 __builtin_scalar_test_data_class_sp and
38120 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
38121 int" and changed second argument from "unsigned int" to "int".
38122 Changed type returned by __builtin_scalar_test_neg,
38123 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
38124 from "int" to "bool int". Added new overloaded function form
38125 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
38126 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
38127 exponent double-precision with floating point first argument.
38128 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
38129 documentation of scalar_test_data_class, scalar_test_neg,
38130 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
38131 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
38132 vec_test_data_class built-in functions to reflect refinements in
38133 their type signatures.
38135 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
38137 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
38139 (aarch64_elf_asm_destructor): Likewise.
38141 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
38143 PR rtl-optimization/78634
38144 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
38145 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
38146 * ifcvt.c (noce_try_cmove): Add missing cost check.
38148 PR rtl-optimization/71724
38149 * combine.c (if_then_else_cond): Look for situations where it is
38150 beneficial to undo the work of one of the recursive calls.
38152 2017-01-23 Bin Cheng <bin.cheng@arm.com>
38154 PR tree-optimization/70754
38155 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
38156 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
38157 combined stmt before it if not NULL.
38158 (combine_chains): Process refs reversely and compute dominance point
38161 2017-01-23 Martin Liska <mliska@suse.cz>
38163 PR tree-optimization/79196
38164 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
38165 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
38167 (strlen_optimize_stmt): Call the renamed function.
38169 2017-01-23 Michael Matz <matz@suse.de>
38171 PR tree-optimization/78384
38172 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
38174 2017-01-23 Richard Biener <rguenther@suse.de>
38176 PR tree-optimization/79186
38177 * tree-vrp.c (register_new_assert_for): Make sure we've seen
38178 both incoming edges before moving an assert.
38180 2017-01-23 Martin Jambor <mjambor@suse.cz>
38182 * ipa-prop.c (load_from_param_1): Removed.
38183 (load_from_unmodified_param): Bits from load_from_param_1 put back
38185 (load_from_param): Removed.
38186 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
38187 with stmt. Reverted back to use of load_from_unmodified_param.
38189 2017-01-23 Martin Jambor <mjambor@suse.cz>
38192 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
38193 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
38194 field a pointer to garbage collected vector, mark lattices and
38195 ipcp_orig_node with GTY((skip)).
38196 (ipa_get_param_count): Adjust to descriptors being a pointer.
38197 (ipa_get_param): Likewise.
38198 (ipa_get_type): Likewise.
38199 (ipa_get_param_move_cost): Likewise.
38200 (ipa_set_param_used): Likewise.
38201 (ipa_get_controlled_uses): Likewise.
38202 (ipa_set_controlled_uses): Likewise.
38203 (ipa_is_param_used): Likewise.
38204 (ipa_node_params_t): Move into garbage collector. New methods insert
38206 (ipa_node_params_sum): Annotate wth GTY(()).
38207 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
38209 (ipa_load_from_parm_agg): Adjust declaration.
38210 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
38211 * ipa-profile.c (ipa_profile): Likewise.
38212 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
38213 (ipa_populate_param_decls): Make descriptors parameter garbage
38215 (ipa_dump_param): Adjust to descriptors being a pointer.
38216 (ipa_alloc_node_params): Likewise.
38217 (ipa_initialize_node_params): Likewise.
38218 (load_from_param_1): Make descriptors parameter garbage collected.
38219 (load_from_unmodified_param): Likewise.
38220 (load_from_param): Likewise.
38221 (ipa_load_from_parm_agg): Likewise.
38222 (ipa_node_params::~ipa_node_params): Removed.
38223 (ipa_free_all_node_params): Remove call to delete operator.
38224 (ipa_node_params_t::insert): New.
38225 (ipa_node_params_t::remove): Likewise.
38226 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
38227 copy known_csts and known_contexts vectors.
38228 (ipa_read_node_info): Adjust to descriptors being a pointer.
38229 (ipcp_modif_dom_walker): Make m_descriptors field garbage
38231 (ipcp_transform_function): Make descriptors variable garbage
38234 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
38236 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
38237 * config/i386/avx512dqintrin.h: Ditto.
38238 * config/i386/avx512fintrin.h: Ditto.
38239 * config/i386/i386.c: Handle new builtins.
38240 * config/i386/i386-builtin.def: Add new builtins.
38241 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
38242 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
38244 2017-01-23 Jakub Jelinek <jakub@redhat.com>
38245 Martin Liska <mliska@suse.cz>
38247 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
38248 * asan.c (asan_expand_poison_ifn): Support stores and use
38249 appropriate ASAN report function.
38250 * internal-fn.c (expand_ASAN_POISON_USE): New function.
38251 * internal-fn.def (ASAN_POISON_USE): Declare.
38252 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
38253 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
38254 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
38255 ASAN_POISON calls w/o LHS.
38256 * tree-ssa.c (execute_update_addresses_taken): Create clobber
38257 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
38258 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
38259 * gimplify.c (asan_poison_variables): Add attribute
38260 use_after_scope_memory to variables that really needs to live
38262 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
38263 having the attribute.
38265 2017-01-23 Martin Liska <mliska@suse.cz>
38267 * asan.c (create_asan_shadow_var): New function.
38268 (asan_expand_poison_ifn): Likewise.
38269 * asan.h (asan_expand_poison_ifn): New declaration.
38270 * internal-fn.c (expand_ASAN_POISON): Likewise.
38271 * internal-fn.def (ASAN_POISON): New builtin.
38272 * sanopt.c (pass_sanopt::execute): Expand
38273 asan_expand_poison_ifn.
38274 * tree-inline.c (copy_decl_for_dup_finish): Make function
38276 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
38277 * tree-ssa.c (is_asan_mark_p): New function.
38278 (execute_update_addresses_taken): Rewrite local variables
38279 (identified just by use-after-scope as addressable) into SSA.
38281 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38283 * doc/install.texi (Specific): opensource.apple.com uses https
38284 now. Remove trailing slash.
38286 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
38288 * README.Portability: Remove note on an Irix compatibility issue.
38290 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
38292 * gcov.c (INCLUDE_ALGORITHM): Define.
38293 (INCLUDE_VECTOR): Define.
38294 No longer include <vector> and <algorithm> directly.
38296 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
38298 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
38300 * doc/invoke.texi (Code Gen Options): Ditto.
38302 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
38305 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
38307 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
38309 rtl-optimization/79125
38310 * cprop.c (local_cprop_pass): Handle cases where we make an
38311 unconditional trap.
38313 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
38317 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
38318 read from, for big endian.
38320 2017-01-20 Jiong Wang <jiong.wang@arm.com>
38322 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
38323 register pauth builtins for LP64 only.
38325 2017-01-20 Marek Polacek <polacek@redhat.com>
38328 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
38331 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38333 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
38335 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
38336 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
38337 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
38339 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38342 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
38343 in big-endian mode when they are not a single duplicated value.
38345 2017-01-20 Richard Biener <rguenther@suse.de>
38347 * BASE-VER: Bump to 7.0.1.
38349 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
38351 * omp-low.c (omplow_simd_context): New struct. Use it...
38352 (lower_rec_simd_input_clauses): ...here and...
38353 (lower_rec_input_clauses): ...here to hold common data. Adjust all
38354 references to idx, lane, max_vf, is_simt.
38356 2017-01-20 Graham Markall <graham.markall@embecosm.com>
38358 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
38361 2017-01-20 Martin Jambor <mjambor@suse.cz>
38363 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
38364 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
38366 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
38367 (GTFILES): Rename hsa.c to hsa-common.c.
38368 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
38369 * hsa-dump.c: Likewise.
38370 * hsa-gen.c: Likewise.
38371 * hsa-regalloc.c: Likewise.
38372 * ipa-hsa.c: Likewise.
38373 * omp-expand.c: Likewise.
38374 * omp-low.c: Likewise.
38375 * toplev.c: Likewise.
38377 2017-01-20 Marek Polacek <polacek@redhat.com>
38380 * doc/invoke.texi: Document -Wduplicated-branches.
38381 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
38382 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
38383 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
38384 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
38385 return 0 only when not OEP_LEXICOGRAPHIC.
38386 (fold_build_cleanup_point_expr): Use the expression
38387 location when building CLEANUP_POINT_EXPR.
38388 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
38389 * tree.c (add_expr): Handle error_mark_node.
38391 2017-01-20 Martin Liska <mliska@suse.cz>
38394 * tree-profile.c (init_ic_make_global_vars): Do not call
38396 (gimple_init_gcov_profiler): Likewise.
38398 2017-01-20 Martin Liska <mliska@suse.cz>
38401 * cgraph.h (maybe_create_reference): Remove argument and
38403 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
38405 * ipa-cp.c (create_specialized_node): Likewise.
38406 * symtab.c (symtab_node::maybe_create_reference): Handle
38407 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
38409 2017-01-20 Martin Liska <mliska@suse.cz>
38411 * read-rtl-function.c (function_reader::create_function): Use
38412 build_decl instread of build_decl_stat.
38414 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
38416 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
38417 * config/i386/avx512dqintrin.h: Ditto.
38418 * config/i386/avx512fintrin.h: Ditto.
38419 * config/i386/i386-builtin-types.def: Add new types.
38420 * config/i386/i386.c: Handle new types.
38421 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
38422 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
38423 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
38424 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
38425 (__builtin_ia32_kshiftridi): New.
38426 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
38428 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
38432 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
38433 define to rs6000_init_stack_protect_guard.
38434 (rs6000_init_stack_protect_guard): New function.
38436 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38437 Yunqiang Su <yunqiang.su@imgtec.com>
38439 * config.gcc (supported_defaults): Add madd4.
38440 (with_madd4): Add validation.
38441 (all_defaults): Add madd4.
38442 * config/mips/mips.opt (mmadd4): New option.
38443 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38445 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38447 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
38448 (ISA_HAS_FUSED_MADD4): Likewise.
38449 * doc/invoke.texi (-mmadd4): Document the new option.
38450 * doc/install.texi (--with-madd4): Document the new option.
38452 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38454 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
38455 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
38456 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
38457 (aarch64_init_pauth_hint_builtins): New.
38458 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
38459 (aarch64_expand_builtin): Expand new builtins.
38461 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38463 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
38464 * combine-stack-adj.c (no_unhandled_cfa): Handle
38465 REG_CFA_TOGGLE_RA_MANGLE.
38466 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
38467 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
38468 info for return address signing.
38469 (aarch64_expand_epilogue): Likewise.
38471 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38473 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
38474 * config/aarch64/aarch64-protos.h
38475 (aarch64_return_address_signing_enabled): New declaration.
38476 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
38478 (aarch64_expand_prologue): Sign return address before it's pushed onto
38480 (aarch64_expand_epilogue): Authenticate return address fetched from
38482 (aarch64_override_options): Sanity check for ILP32 and ISA level.
38483 (aarch64_attributes): New function attributes for "sign-return-address".
38484 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
38485 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
38486 ("*do_return"): Generate combined instructions according to key index.
38487 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
38488 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
38490 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
38491 * config/aarch64/aarch64.opt (msign-return-address=): New.
38492 * doc/extend.texi (AArch64 Function Attributes): Documents
38493 "sign-return-address=".
38494 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
38496 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
38498 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
38499 overall option summary.
38501 2017-01-19 Jiong Wang <jiong.wang@arm.com>
38503 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
38504 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
38505 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
38506 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
38508 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
38510 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
38511 -mpower9-minmax by default for -mcpu=power9.
38512 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
38513 128-bit floating point.
38515 2017-01-20 Alan Modra <amodra@gmail.com>
38517 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
38518 optimizing for size.
38520 2017-01-20 Alan Modra <amodra@gmail.com>
38523 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
38524 for strcmp and strncmp from corresponding builtin decl.
38526 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38528 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
38529 instead of i386/rtems-64.h.
38530 * config/i386/rtems-64.h: Remove.
38532 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
38536 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
38538 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
38540 2017-01-19 Tamar Christina <tamar.christina@arm.com>
38542 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
38543 Change int to HOST_WIDE_INT.
38544 * config/aarch64/aarch64-protos.h
38545 (aarch64_simd_gen_const_vector_dup): Likewise.
38546 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
38548 2017-01-19 David Malcolm <dmalcolm@redhat.com>
38550 * langhooks-def.h (lhd_type_for_size): New decl.
38551 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
38552 * langhooks.c (lhd_type_for_size): New function, taken from
38555 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
38557 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
38558 define_bypass for CR latency.
38559 (power9-cracked-alu): Update bypass latency and remove power9-branch.
38560 (power9-alu2): Add define_bypass for CR latency.
38562 (power9-mul): Update insn latency.
38563 (power9-mul-compare): Update insn latency, bypass latency and remove
38566 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38568 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
38570 * config/aarch64/aarch64.md
38571 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
38572 aarch64_nopcrelative_literal_loads.
38573 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
38575 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
38577 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
38578 TARGET_LOONGSON_3A.
38579 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
38581 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
38584 * config.gcc (supported_defaults): Add lxc1-sxc1.
38585 (with_lxc1_sxc1): Add validation.
38586 (all_defaults): Add lxc1-sxc1.
38587 * config/mips/mips.opt (mlxc1-sxc1): New option.
38588 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
38590 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
38591 __mips_no_lxc1_sxc1.
38592 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
38593 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
38594 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
38596 2017-01-19 Richard Biener <rguenther@suse.de>
38598 PR tree-optimization/72488
38599 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
38600 sure to restore SSA info.
38601 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
38603 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
38605 PR rtl-optimization/79121
38606 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
38607 of the inner type when shifting an extended value.
38609 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
38612 * symtab.c (symtab_node::equal_address_to): Fix comparing of
38613 interposable aliases.
38615 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
38618 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
38619 Use the evmergelohi instruction.
38620 (mov_si<mode>_e500_subreg4_2_le): Likewise.
38621 (mov_sitf_e500_subreg8_2_be): Likewise.
38622 (mov_sitf_e500_subreg12_2_le): Likewise.
38623 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
38624 (mov_si<mode>_e500_subreg4_2_be): Likewise.
38625 (mov_sitf_e500_subreg8_2_le): Likewise.
38626 (mov_sitf_e500_subreg12_2_be): Likewise.
38628 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38630 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
38631 attribute from vecsimple to vecperm.
38632 (altivec_vbpermq2): Likewise.
38634 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38637 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
38639 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38640 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
38641 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
38642 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
38643 case where N arg is SIZE_MAX.
38644 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
38645 (cmpstrsi): Add pattern.
38647 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
38649 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
38650 __builtin_vec_revb builtins.
38651 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
38652 built-in functions to support generation of the ISA 3.0 XXBR<x>
38653 vector byte reverse instructions.
38654 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
38655 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
38656 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
38657 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
38658 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
38659 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
38660 (P9V_BUILTIN_VEC_REVB): Likewise.
38661 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
38662 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
38663 (p9_xxbrq_v16qi): Likewise.
38664 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
38665 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
38666 (p9_xxbrh_v8hi): Likewise.
38667 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
38668 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
38669 vec_revb built-in functions.
38671 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
38673 PR rtl-optimization/78952
38674 * config/i386/i386.md (any_extract): New code iterator.
38675 (*insvqi_2): Use any_extract for source operand.
38676 (*insvqi_3): Use any_shiftrt for source operand.
38678 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
38680 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
38682 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
38684 2017-01-18 Matthias Klose <doko@ubuntu.com>
38686 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
38688 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38690 * config/rs6000/altivec.h (vec_bperm): Change #define.
38691 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
38692 (altivec_vbpermq2): New define_insn.
38693 (altivec_vbpermd): Likewise.
38694 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
38695 function interface.
38696 (VBPERMD): Likewise.
38697 (VBPERM): New polymorphic function interface.
38698 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
38699 Add entries for P9V_BUILTIN_VEC_VBPERM.
38700 * doc/extend.texi: Add interfaces for vec_bperm.
38702 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38704 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
38705 first letter of error messages.
38706 (s390_resolve_overloaded_builtin): Likewise.
38707 * config/s390/s390.c (s390_expand_builtin): Likewise.
38708 (s390_invalid_arg_for_unprototyped_fn): Likewise.
38709 (s390_valid_target_attribute_inner_p): Likewise.
38710 * config/s390/s390.md ("tabort"): Likewise.
38712 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
38714 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
38715 (ISA_AVOID_DIV_HILO): New macro.
38716 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
38717 (ISA_HAS_DDIV): Likewise.
38719 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38721 * doc/invoke.texi (fabi-version): Correct number of occurrences.
38723 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38725 * doc/invoke.texi (fabi-version): Spelling fix.
38727 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38730 * doc/invoke.texi (fabi-version): Mention mangling fix for
38733 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
38736 * doc/invoke.texi (fabi-version): Document discriminator mangling.
38738 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
38741 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
38742 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
38744 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
38746 (stack_protect_test): Ditto.
38747 * config/rs6000/rs6000.opt (mstack-protector-guard=,
38748 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
38750 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
38751 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
38752 -mstack-protector-guard-offset=.
38753 (RS/6000 and PowerPC Options): Ditto.
38755 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
38757 * config/i386/i386.h (MASK_CLASS_P): New define.
38758 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
38759 there are no registers from different register sets also when
38760 mask registers are used. Update function comment.
38761 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
38762 to (*k/*r) and (*k/*km) alternatives.
38764 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
38766 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
38767 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
38768 (EH_RETURN_HANDLER_RTX): New define.
38769 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
38770 Force frame pointer in EH return functions.
38771 (aarch64_expand_epilogue): Add barrier for eh_return.
38772 (aarch64_final_eh_return_addr): Remove.
38773 (aarch64_eh_return_handler_rtx): New function.
38774 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
38776 (aarch64_eh_return_handler_rtx): New prototype.
38778 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38780 * config/rs6000/altivec.h (vec_rlmi): New #define.
38781 (vec_vrlnm): Likewise.
38782 (vec_rlnm): Likewise.
38783 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
38784 (UNSPEC_VRLNM): Likewise.
38785 (VIlong): New mode iterator.
38786 (altivec_vrl<VI_char>mi): New define_insn.
38787 (altivec_vrl<VI_char>nm): Likewise.
38788 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
38790 (VRLDNM): Likewise.
38791 (RLNM): New polymorphic function entry.
38792 (VRLWMI): New monomorphic function entry.
38793 (VRLDMI): Likewise.
38794 (RLMI): New polymorphic function entry.
38795 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
38796 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
38797 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
38800 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38803 * dwarf2out.c (field_byte_offset): Restore the
38804 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
38805 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
38806 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
38809 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38812 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
38814 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38817 * dwarf2out.c (add_data_member_location_attribute): For constant
38818 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
38819 instead of DW_AT_data_member_location, DW_AT_bit_offset and
38820 DW_AT_byte_size attributes.
38822 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38824 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
38825 after forcing to constant memory when the code model is medium.
38827 2017-01-17 Julia Koval <julia.koval@intel.com>
38830 * config/i386/avx512fintrin.h
38831 (_mm512_i32gather_ps): Change __addr type to void const*.
38832 (_mm512_mask_i32gather_ps): Ditto.
38833 (_mm512_i32gather_pd): Ditto.
38834 (_mm512_mask_i32gather_pd): Ditto.
38835 (_mm512_i64gather_ps): Ditto.
38836 (_mm512_mask_i64gather_ps): Ditto.
38837 (_mm512_i64gather_pd): Ditto.
38838 (_mm512_mask_i64gather_pd): Ditto.
38839 (_mm512_i32gather_epi32): Ditto.
38840 (_mm512_mask_i32gather_epi32): Ditto.
38841 (_mm512_i32gather_epi64): Ditto.
38842 (_mm512_mask_i32gather_epi64): Ditto.
38843 (_mm512_i64gather_epi32): Ditto.
38844 (_mm512_mask_i64gather_epi32): Ditto.
38845 (_mm512_i64gather_epi64): Ditto.
38846 (_mm512_mask_i64gather_epi64): Ditto.
38847 (_mm512_i32scatter_ps): Change __addr type to void*.
38848 (_mm512_mask_i32scatter_ps): Ditto.
38849 (_mm512_i32scatter_pd): Ditto.
38850 (_mm512_mask_i32scatter_pd): Ditto.
38851 (_mm512_i64scatter_ps): Ditto.
38852 (_mm512_mask_i64scatter_ps): Ditto.
38853 (_mm512_i64scatter_pd): Ditto.
38854 (_mm512_mask_i64scatter_pd): Ditto.
38855 (_mm512_i32scatter_epi32): Ditto.
38856 (_mm512_mask_i32scatter_epi32): Ditto.
38857 (_mm512_i32scatter_epi64): Ditto.
38858 (_mm512_mask_i32scatter_epi64): Ditto.
38859 (_mm512_i64scatter_epi32): Ditto.
38860 (_mm512_mask_i64scatter_epi32): Ditto.
38861 (_mm512_i64scatter_epi64): Ditto.
38862 (_mm512_mask_i64scatter_epi64): Ditto.
38863 * config/i386/avx512pfintrin.h
38864 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
38865 (_mm512_mask_prefetch_i32gather_ps): Ditto.
38866 (_mm512_mask_prefetch_i64gather_pd): Ditto.
38867 (_mm512_mask_prefetch_i64gather_ps): Ditto.
38868 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
38869 (_mm512_prefetch_i32scatter_ps): Ditto.
38870 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
38871 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
38872 (_mm512_prefetch_i64scatter_pd): Ditto.
38873 (_mm512_prefetch_i64scatter_ps): Ditto.
38874 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
38875 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
38876 * config/i386/avx512vlintrin.h
38877 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
38878 (_mm_mmask_i32gather_ps): Ditto.
38879 (_mm256_mmask_i32gather_pd): Ditto.
38880 (_mm_mmask_i32gather_pd): Ditto.
38881 (_mm256_mmask_i64gather_ps): Ditto.
38882 (_mm_mmask_i64gather_ps): Ditto.
38883 (_mm256_mmask_i64gather_pd): Ditto.
38884 (_mm_mmask_i64gather_pd): Ditto.
38885 (_mm256_mmask_i32gather_epi32): Ditto.
38886 (_mm_mmask_i32gather_epi32): Ditto.
38887 (_mm256_mmask_i32gather_epi64): Ditto.
38888 (_mm_mmask_i32gather_epi64): Ditto.
38889 (_mm256_mmask_i64gather_epi32): Ditto.
38890 (_mm_mmask_i64gather_epi32): Ditto.
38891 (_mm256_mmask_i64gather_epi64): Ditto.
38892 (_mm_mmask_i64gather_epi64): Ditto.
38893 (_mm256_i32scatter_ps): Change __addr type to void*.
38894 (_mm256_mask_i32scatter_ps): Ditto.
38895 (_mm_i32scatter_ps): Ditto.
38896 (_mm_mask_i32scatter_ps): Ditto.
38897 (_mm256_i32scatter_pd): Ditto.
38898 (_mm256_mask_i32scatter_pd): Ditto.
38899 (_mm_i32scatter_pd): Ditto.
38900 (_mm_mask_i32scatter_pd): Ditto.
38901 (_mm256_i64scatter_ps): Ditto.
38902 (_mm256_mask_i64scatter_ps): Ditto.
38903 (_mm_i64scatter_ps): Ditto.
38904 (_mm_mask_i64scatter_ps): Ditto.
38905 (_mm256_i64scatter_pd): Ditto.
38906 (_mm256_mask_i64scatter_pd): Ditto.
38907 (_mm_i64scatter_pd): Ditto.
38908 (_mm_mask_i64scatter_pd): Ditto.
38909 (_mm256_i32scatter_epi32): Ditto.
38910 (_mm256_mask_i32scatter_epi32): Ditto.
38911 (_mm_i32scatter_epi32): Ditto.
38912 (_mm_mask_i32scatter_epi32): Ditto.
38913 (_mm256_i32scatter_epi64): Ditto.
38914 (_mm256_mask_i32scatter_epi64): Ditto.
38915 (_mm_i32scatter_epi64): Ditto.
38916 (_mm_mask_i32scatter_epi64): Ditto.
38917 (_mm256_i64scatter_epi32): Ditto.
38918 (_mm256_mask_i64scatter_epi32): Ditto.
38919 (_mm_i64scatter_epi32): Ditto.
38920 (_mm_mask_i64scatter_epi32): Ditto.
38921 (_mm256_i64scatter_epi64): Ditto.
38922 (_mm256_mask_i64scatter_epi64): Ditto.
38923 (_mm_i64scatter_epi64): Ditto.
38924 (_mm_mask_i64scatter_epi64): Ditto.
38925 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
38926 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
38927 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
38928 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
38929 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
38930 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
38931 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
38932 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
38933 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
38934 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
38935 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
38936 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
38937 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
38938 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
38939 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
38940 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
38941 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
38942 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
38943 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
38944 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
38945 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
38946 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
38947 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
38948 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
38949 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
38950 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
38951 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
38952 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
38953 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
38954 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
38955 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
38956 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
38957 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
38958 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
38959 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
38960 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
38961 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
38962 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
38963 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
38964 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
38965 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
38966 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
38967 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
38968 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
38969 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
38970 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
38971 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
38972 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
38973 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
38974 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
38975 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
38976 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
38977 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
38978 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
38979 definitions accordingly.
38981 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
38982 Kuan-Lin Chen <kuanlinchentw@gmail.com>
38985 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
38988 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
38991 * ira-conflicts.c (ira_build_conflicts): Update total conflict
38992 hard regs for inner regno.
38994 2017-01-17 Martin Liska <mliska@suse.cz>
38997 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
38998 assumption and add comment.
39000 2017-01-17 Nathan Sidwell <nathan@acm.org>
39002 * ipa-visibility.c (localize_node): New function, broken out of ...
39003 (function_and_variable_visibility): ... here. Call it.
39005 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
39007 PR middle-end/77445
39008 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
39009 correctly set frequency of oudgoing edge.
39010 (duplicate_thread_path): Fix profile updating.
39012 2017-01-17 Jakub Jelinek <jakub@redhat.com>
39015 * configure.ac: Add GCC_BASE_VER.
39016 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
39017 version from BASE-VER file.
39018 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
39019 (gcc.o): Depend on $(BASEVER).
39020 * common.opt (dumpfullversion): New option.
39021 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
39022 * doc/invoke.texi: Document -dumpfullversion.
39023 * doc/install.texi: Document --with-gcc-major-version-only.
39024 * configure: Regenerated.
39026 2017-01-17 Richard Biener <rguenther@suse.de>
39028 PR tree-optimization/71433
39029 * tree-vrp.c (register_new_assert_for): Merge same asserts
39030 on all incoming edges.
39031 (process_assert_insertions_for): Handle insertions at the
39034 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
39036 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
39037 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
39039 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
39042 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
39045 2017-01-17 Alan Modra <amodra@gmail.com>
39048 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
39049 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
39050 symbolic stack limit when pic.
39052 2017-01-16 Martin Sebor <msebor@redhat.com>
39054 PR tree-optimization/78608
39055 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
39057 2017-01-16 Jeff Law <law@redhat.com>
39060 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
39061 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
39062 for several include directories that may be relative to sysroot.
39063 * config/i386/x-mingw32 (gplus_includedir): Define.
39064 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
39065 (native_system_includedir): Likewise.
39066 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
39067 override if TARGET_SYSTEM_ROOT is defined.
39068 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
39070 PR tree-optimization/79090
39071 PR tree-optimization/33562
39072 PR tree-optimization/61912
39073 PR tree-optimization/77485
39074 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
39075 and computed trims into the dump file.
39077 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
39079 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
39081 2017-01-16 Jakub Jelinek <jakub@redhat.com>
39084 * gimplify.c (gimplify_init_constructor): If want_value and
39085 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
39089 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
39090 sequence. Formatting fixes.
39091 (doloop_optimize): Formatting fixes.
39094 * gcc.c (debug_level_greater_than_spec_func): New function.
39095 (static_spec_functions): Add debug-level-gt spec function.
39096 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
39098 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39099 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
39100 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
39101 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
39102 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
39103 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
39105 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
39107 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
39108 QImode fixups to general and mask registers only.
39110 2017-01-16 Carl Love <cel@us.ibm.com>
39112 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
39113 for built-in functions
39114 vector signed char vec_nabs (vector signed char)
39115 vector signed short vec_nabs (vector signed short)
39116 vector signed int vec_nabs (vector signed int)
39117 vector signed long long vec_nabs (vector signed long long)
39118 vector float vec_nabs (vector float)
39119 vector double vec_nabs (vector double)
39120 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
39122 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
39123 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
39124 * doc/extend.texi: Update the documentation file for the new built-in
39127 2017-01-16 Martin Sebor <msebor@redhat.com>
39129 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
39132 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39134 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
39135 UNSPEC_VSX__XXSPLTD to require special splat handling.
39137 2017-01-16 David Malcolm <dmalcolm@redhat.com>
39140 * system.h: Poison strndup.
39142 2017-01-16 Alan Modra <amodra@gmail.com>
39145 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
39148 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
39150 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
39152 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
39154 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
39155 call recog here. Assert that INSN_CODE (insn) is non-negative.
39157 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39160 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
39162 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
39163 in the currently scheduled RTL fragment.
39165 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
39167 PR rtl-optimization/78751
39168 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
39171 2017-01-14 Jeff Law <law@redhat.com>
39173 PR tree-optimization/79090
39174 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
39175 variable length stores.
39176 (compute_trims): Delete dead assignment to *trim_tail.
39177 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
39180 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
39182 PR rtl-optimization/78626
39183 PR rtl-optimization/78727
39184 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
39185 of a block, and split such blocks after everything else is finished.
39187 2017-01-14 Alan Modra <amodra@gmail.com>
39190 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
39191 target legitimate_combined_insn.
39192 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
39193 (rs6000_legitimate_combined_insn): New function.
39194 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
39196 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
39197 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
39198 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
39200 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
39202 * doc/frontends.texi (G++ and GCC): Remove references to Java.
39204 2017-01-13 Jeff Law <law@redhat.com>
39206 PR tree-optimization/33562
39207 PR tree-optimization/61912
39208 PR tree-optimization/77485
39209 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
39211 (delete_dead_assignment): Likewise.
39212 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
39213 statement to delete_dead_call and delete_dead_assignment.
39215 2017-01-13 David Malcolm <dmalcolm@redhat.com>
39218 * substring-locations.c (format_warning_va): Strengthen case 1 so
39219 that both endpoints of the substring must be within the format
39220 range for just the substring to be printed.
39222 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
39224 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
39225 * config/i386/i386.c (ix86_target_string): Add missing options
39226 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
39227 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
39228 flags_other and ix86_target_other to flags2_other. Display unknown
39230 (ix86_valid_target_attribute_inner_p): Add missing options and
39231 reorder options by implied ISAs, as in ix86_target_string.
39233 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39235 * hash-table.h (hash_table::too_empty_p): New function.
39236 (hash_table::expand): Use it.
39237 (hash_table::traverse): Likewise.
39238 (hash_table::empty_slot): Use sizeof (value_type) instead of
39239 sizeof (PTR) to convert bytes to elements. Shrink the table
39240 if the current size is excessive for the current number of
39243 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
39245 * ira-costs.c (record_reg_classes): Break from the inner loop
39246 early once alt_fail is known to be true. Update outer loop
39247 handling accordingly.
39249 2017-01-13 Jeff Law <law@redhat.com>
39251 * tree-ssa-dse.c (decrement_count): New function.
39252 (increment_start_addr, maybe_trim_memstar_call): Likewise.
39253 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
39254 when we know the partially dead statement is a mem* function.
39256 PR tree-optimization/61912
39257 PR tree-optimization/77485
39258 * tree-ssa-dse.c: Include expr.h.
39259 (maybe_trim_constructor_store): New function.
39260 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
39262 PR tree-optimization/33562
39263 PR tree-optimization/61912
39264 PR tree-optimization/77485
39265 * doc/invoke.texi: Document new dse-max-object-size param.
39266 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
39267 * tree-ssa-dse.c: Include params.h.
39268 (dse_store_status): New enum.
39269 (initialize_ao_ref_for_dse): New, partially extracted from
39271 (valid_ao_ref_for_dse, normalize_ref): New.
39272 (setup_live_bytes_from_ref, compute_trims): Likewise.
39273 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
39274 (maybe_trim_partially_dead_store): Likewise.
39275 (maybe_trim_complex_store): Likewise.
39276 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
39277 Track what bytes live from the original store. Return tri-state
39278 for dead, partially dead or live.
39279 (dse_dom_walker): Add constructor, destructor and new private members.
39280 (delete_dead_call, delete_dead_assignment): New extracted from
39282 (dse_optimize_stmt): Make a member of dse_dom_walker.
39283 Use initialize_ao_ref_for_dse.
39285 PR tree-optimization/33562
39286 PR tree-optimization/61912
39287 PR tree-optimization/77485
39288 * sbitmap.h (bitmap_count_bits): Prototype.
39289 (bitmap_clear_range, bitmap_set_range): Likewise.
39290 * sbitmap.c (bitmap_clear_range): New function.
39291 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
39293 2017-01-13 Martin Liska <mliska@suse.cz>
39296 * function.c (set_cfun): Add new argument force.
39297 * function.h (set_cfun): Likewise.
39298 * ipa-inline-transform.c (inline_call): Use the function when
39299 strict alising from is dropped for function we inline to.
39301 2017-01-13 Richard Biener <rguenther@suse.de>
39303 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
39304 for dumping GIMPLE INTEGER_CSTs.
39306 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
39308 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
39309 to 201112L since C++17.
39311 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
39314 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
39315 if -fsanitize=kernel-address is present.
39317 2017-01-13 Richard Biener <rguenther@suse.de>
39319 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
39320 as _Literal ( type ) number in case usual suffixes do not
39321 preserve all information.
39323 2017-01-13 Richard Biener <rguenther@suse.de>
39325 PR tree-optimization/77283
39326 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
39327 and ssa-iterators.h.
39328 (is_feasible_trace): Implement a cost model based on joiner
39331 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
39334 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
39335 char or short to __float128/_Float128 directly.
39337 2017-01-12 Martin Sebor <msebor@redhat.com>
39339 to -Wformat-overflow.
39340 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
39341 (min_bytes_remaining): Same.
39342 (get_string_length): Same.
39343 (format_string): Same.
39344 (format_directive): Same.
39346 (pass_sprintf_length::handle_gimple_call): Same.
39348 2017-01-12 Jakub Jelinek <jakub@redhat.com>
39350 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
39351 info.nowrite calls with no lhs that can't throw. Return bool
39352 whether gsi_remove has been called or not.
39353 (pass_sprintf_length::handle_gimple_call): Return bool whether
39354 try_substitute_return_value called gsi_remove. Formatting fix.
39355 (pass_sprintf_length::execute): Don't use gsi_remove if
39356 handle_gimple_call returned true.
39359 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
39360 be removed due to side-effects, don't remove following barrier nor
39361 turn the successor edge into fallthru edge.
39363 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39366 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
39367 element-reversing loads and stores as not swappable.
39369 2017-01-12 Nathan Sidwell <nathan@acm.org>
39370 Nicolai Stange <nicstange@gmail.com>
39372 * combine.c (try_combine): Don't ignore result of overlap checking
39373 loop. Combine overlap & asm check into single loop.
39375 2017-01-12 Richard Biener <rguenther@suse.de>
39377 * tree-pretty-print.c (dump_generic_node): Provide -gimple
39378 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
39380 2017-01-12 Richard Biener <rguenther@suse.de>
39382 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
39383 and TS_TARGET_OPTION directly derive from TS_BASE.
39384 * tree-core.h (tree_optimization_option): Derive from tree_base.
39385 (tree_target_option): Likewise.
39387 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
39389 * config/i386/i386.c (memory_address_length): Increase len
39390 only when rip_relative_addr_p returns false.
39392 2017-01-11 Julia Koval <julia.koval@intel.com>
39394 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
39395 (OPTION_MASK_ISA_SGX_SET): New.
39396 (ix86_handle_option): Handle OPT_msgx.
39397 * config.gcc: Added sgxintrin.h.
39398 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
39399 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
39400 * config/i386/i386.c (ix86_target_string): Add -msgx.
39402 (ix86_option_override_internal): Handle new options.
39403 (ix86_valid_target_attribute_inner_p): Add sgx.
39404 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
39405 * config/i386/i386.opt: Add msgx.
39406 * config/i386/sgxintrin.h: New file.
39407 * config/i386/x86intrin.h: Add sgxintrin.h.
39409 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39412 * fold-const.c (maybe_nonzero_address): Return 1 for function
39414 (tree_single_nonzero_warnv_p): Don't handle function local objects
39418 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
39421 2017-01-11 David Malcolm <dmalcolm@redhat.com>
39424 * opts.c: Include "spellcheck.h"
39425 (struct string_fragment): New struct.
39426 (struct edit_distance_traits<const string_fragment &>): New
39428 (get_closest_sanitizer_option): New function.
39429 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
39431 2017-01-11 Jakub Jelinek <jakub@redhat.com>
39433 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
39435 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
39436 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
39437 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
39438 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
39439 for initial die_offset if dwarf_split_debug_info.
39440 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
39441 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
39443 (output_skeleton_debug_sections): Formatting fix. Use
39444 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
39445 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
39447 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
39449 * config/arm/cortex-a53.md: Add bypasses for
39450 cortex_a53_r2f_cvt.
39451 (cortex_a53_r2f): Only use for transfers.
39452 (cortex_a53_f2r): Likewise.
39453 (cortex_a53_r2f_cvt): Add reservation for conversions.
39454 (cortex_a53_f2r_cvt): Likewise.
39456 2017-01-11 Tamar Christina <tamar.christina@arm.com>
39458 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
39459 to all inlined functions, change static to extern.
39461 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
39464 * config/arm/arm.c (legitimize_pic_address): Handle reference to
39466 (arm_assemble_integer): Likewise.
39468 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
39470 * config.gcc: Use new awk script to check CPU, FPU and architecture
39471 parameters for --with-... options.
39472 * config/arm/parsecpu.awk: New file
39473 * config/arm/arm-cpus.in: New file.
39474 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
39476 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
39478 * config/arm/t-arm: Update dependency rules.
39479 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
39480 of processing .def files.
39481 * config/arm/genopt.sh: Deleted.
39482 * config/arm/gentune.sh: Deleted.
39483 * config/arm/arm-cores.def: Deleted.
39484 * config/arm/arm-arches.def: Deleted.
39485 * config/arm/arm-fpus.def: Deleted.
39486 * config/arm/arm-tune.md: Regenerated.
39487 * config/arm/arm-tables.opt: Regenerated.
39488 * config/arm/arm-cpu.h: New generated file.
39489 * config/arm/arm-cpu-data.h: New generated file.
39490 * config/arm/arm-cpu-cdata.h: New generated file.
39492 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
39495 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
39497 (input_varpool_node): Unpack dynamically_initialized bit.
39499 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
39501 PR rtl-optimization/79032
39502 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
39503 the alignment of the adjusted memory reference against that of MODE,
39504 instead of the alignment of the original memory reference.
39506 2017-01-11 Martin Jambor <mjambor@suse.cz>
39508 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
39510 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
39511 decorated functions.
39513 2017-01-11 Richard Biener <rguenther@suse.de>
39515 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
39516 set range/nonnull info for PHI results. Do not set it on
39517 stmts marked for removal.
39519 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
39521 * expr.c (store_field): In the bitfield case, fetch the return value
39522 from the registers before applying a single big-endian adjustment.
39523 Always do a final load for a BLKmode value not larger than a word.
39525 2017-01-10 David Malcolm <dmalcolm@redhat.com>
39528 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39529 that we correctly handle column numbers greater than
39530 LINE_MAP_MAX_COLUMN_NUMBER.
39532 2017-01-10 Martin Sebor <msebor@redhat.com>
39534 PR middle-end/78245
39535 * gimple-ssa-sprintf.c (get_destination_size): Call
39536 {init,fini}object_sizes.
39537 * tree-object-size.c (addr_object_size): Adjust.
39538 (pass_through_call): Adjust.
39539 (pass_object_sizes::execute): Adjust.
39540 * tree-object-size.h (fini_object_sizes): Declare.
39542 2017-01-10 Martin Sebor <msebor@redhat.com>
39544 PR tree-optimization/78775
39545 * builtins.c (get_size_range): Move...
39546 * calls.c: ...to here.
39547 (alloc_max_size): Accept zero argument.
39548 (operand_signed_p): Remove.
39549 (maybe_warn_alloc_args_overflow): Call get_size_range.
39550 * calls.h (get_size_range): Declare.
39552 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
39554 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
39555 from TI's devices.csv file as of September 2016.
39556 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
39558 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
39560 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
39561 * doc/invoke.texi: Likewise.
39562 * doc/md.texi: Likewise.
39563 * doc/objc.texi: Likewise.
39565 2017-01-10 Joshua Conner <joshconner@google.com>
39567 * config/arm/fuchsia-elf.h: New file.
39568 * config/fuchsia.h: New file.
39569 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
39570 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
39572 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
39574 2016-01-10 Richard Biener <rguenther@suse.de>
39576 PR tree-optimization/79034
39577 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
39578 Propagate out degenerate PHIs in the joiner.
39580 2017-01-10 Martin Liska <mliska@suse.cz>
39582 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
39583 (sort_congruence_classes_by_decl_uid): Likewise.
39584 (sort_congruence_class_groups_by_decl_uid): Likewise.
39585 (sem_item_optimizer::merge_classes): Sort class, groups in these
39586 classes and members in the groups by DECL_UID of declarations.
39587 This would make merge operations stable.
39589 2017-01-10 Martin Liska <mliska@suse.cz>
39591 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
39592 usage of m_classes_vec.
39593 (sem_item_optimizer::~sem_item_optimizer): Likewise.
39594 (sem_item_optimizer::get_group_by_hash): Likewise.
39595 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
39596 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
39597 (sem_item_optimizer::verify_classes): Likewise.
39598 (sem_item_optimizer::process_cong_reduction): Likewise.
39599 (sem_item_optimizer::dump_cong_classes): Likewise.
39600 (sem_item_optimizer::merge_classes): Likewise.
39601 * ipa-icf.h (congruence_class_hash): Rename from
39602 congruence_class_group_hash. Remove declaration of m_classes_vec.
39604 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
39606 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
39607 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
39608 * config.gcc: Add avx512vpopcntdqintrin.h.
39609 * config/i386/avx512vpopcntdqintrin.h: New.
39610 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
39611 * config/i386/i386-builtin-types.def: Add new types.
39612 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
39613 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
39614 __builtin_ia32_vpopcountq_v8di_mask): New.
39615 * config/i386/i386-c.c (ix86_target_macros_internal): Define
39616 __AVX512VPOPCNTDQ__.
39617 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
39618 (PTA_AVX512VPOPCNTDQ): Define.
39619 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
39620 TARGET_AVX512VPOPCNTDQ_P): Define.
39621 * config/i386/i386.opt: Add mavx512vpopcntdq.
39622 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
39623 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
39625 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39627 PR middle-end/77484
39628 * predict.def (PRED_CALL): Set to 67.
39630 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
39632 * expr.c (store_field): In the bitfield case, if the value comes from
39633 a function call and is of an aggregate type returned in registers, do
39634 not modify the field mode; extract the value in all cases if the mode
39635 is BLKmode and the size is not larger than a word.
39637 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
39640 * config/i386/cpuid.h: Fix undefined behavior.
39642 2017-01-04 Jeff Law <law@redhat.com>
39644 PR tree-optimization/79007
39645 PR tree-optimization/67955
39646 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
39647 conservative for pt.null when flag_non_call_exceptions is on.
39649 2017-01-09 Jakub Jelinek <jakub@redhat.com>
39651 PR translation/79019
39652 PR translation/79020
39653 * params.def (PARAM_INLINE_MIN_SPEEDUP,
39654 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
39655 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
39657 * config/avr/avr.opt (maccumulate-args): Likewise.
39658 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
39659 * common.opt (freport-bug): Likewise.
39660 * cif-code.def (CIF_FINAL_ERROR): Likewise.
39661 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
39662 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
39663 translatable string.
39664 * config/i386/i386.c (function_value_32): Likewise.
39665 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
39666 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
39668 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
39669 * common/config/msp430/msp430-common.c (msp430_handle_option):
39671 * symtab.c (symtab_node::verify_base): Likewise.
39672 * opts.c (set_debug_level): Likewise.
39673 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
39674 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
39675 missing whitespace to translatable strings.
39676 * config/avr/avr.md (bswapsi2): Fix typo in comment.
39677 * config/sh/superh.h: Likewise.
39678 * config/i386/xopintrin.h: Likewise.
39679 * config/i386/znver1.md: Likewise.
39680 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
39681 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
39682 * double-int.h (struct double_int): Likewise.
39683 * double-int.c (div_and_round_double): Likewise.
39684 * wide-int.cc: Likewise.
39685 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
39686 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
39687 * cfgcleanup.c (crossjumps_occured): Renamed to ...
39688 (crossjumps_occurred): ... this.
39689 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
39692 PR tree-optimization/78899
39693 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
39694 returning bool return struct loop *, NULL for failure and the new
39696 (versionable_outer_loop_p): Don't version outer loop if it has
39697 dont_vectorized bit set.
39698 (tree_if_conversion): When versioning outer loop, ensure
39699 tree_if_conversion is performed also on the inner loop of the
39700 non-vectorizable outer loop copy.
39701 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
39702 LOOP_VECTORIZED in inner loop of the scalar outer loop and
39703 prevent vectorization of it.
39704 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
39705 the outer loop vectorization of the non-scalar version is attempted
39706 before vectorization of the inner loop in scalar version. If
39707 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
39708 vectorization of its inner loop.
39709 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
39710 has 2 inner loops, rename also on edges from bb whose single pred
39711 is outer_loop->header. Fix typo in function comment.
39713 2017-01-09 Martin Sebor <msebor@redhat.com>
39716 * asan.c (asan_emit_stack_protection): Increase local buffer size
39717 to avoid snprintf truncation warning.
39719 2017-01-09 Andrew Pinski <apinski@cavium.com>
39721 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
39722 to reference thunderx2t99 for the tuning structure
39723 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
39725 (thunderx2t99_extra_costs): This.
39726 * config/aarch64/aarch64-tune.md: Regenerate.
39727 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
39728 (vulcan_addrcost_table): This.
39729 (vulcan_regmove_cost): Rename to ...
39730 (thunderx2t99_regmove_cost): This.
39731 (vulcan_vector_cost): Rename to ...
39732 (thunderx2t99_vector_cost): this.
39733 (vulcan_branch_cost): Rename to ...
39734 (thunderx2t99_branch_cost): This.
39735 (vulcan_tunings): Rename to ...
39736 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
39737 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
39739 2017-01-09 Martin Jambor <mjambor@suse.cz>
39743 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
39744 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
39745 (propagate_vr_accross_jump_function): Use the above function for all
39746 value range computations for pass-through jump functions and type
39747 converasion from explicit value range values.
39748 (ipcp_propagate_stage): Do not attempt to deduce types of formal
39749 parameters from TYPE_ARG_TYPES.
39750 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
39751 (ipa_write_node_info): Stream type of the actual argument.
39752 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
39754 2017-01-09 Martin Liska <mliska@suse.cz>
39757 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
39758 (lookup_compiler): Do not show error message with have_E.
39760 2017-01-09 Jakub Jelinek <jakub@redhat.com>
39762 PR tree-optimization/78938
39763 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
39764 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
39765 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
39766 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
39769 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39771 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
39774 2017-01-09 Richard Biener <rguenther@suse.de>
39776 PR tree-optimization/78997
39777 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
39778 name condition properly.
39780 2017-01-09 Richard Biener <rguenther@suse.de>
39783 * dwarf2out.c (is_cxx): New overload with context.
39784 (is_naming_typedef_decl): Use it.
39786 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39788 * invoke.texi (Option Summary): Correct spacing in option lists
39789 and add line breaks to fix over-long lines.
39791 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39793 PR middle-end/17660
39795 * extend.texi (Common Variable Attributes): Add xref to GCC
39796 Internals manual to explain mode attribute keywords.
39798 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39801 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
39802 and Preprocessor Options.
39803 (Options for Linking): Document -pthread here....
39804 (RS/6000 and PowerPC Options): ...not here.
39805 (Solaris 2 Options): ...or here.
39806 * doc/cppopts.texi: Document -pthread.
39808 2017-01-08 Martin Sebor <msebor@redhat.com>
39810 PR middle-end/77708
39811 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
39812 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
39813 New member functions.
39814 (format_directive): Used them.
39816 (pass_sprintf_length::handle_gimple_call): Same.
39817 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
39818 to avoid truncation for any argument.
39819 (extract_affine_mul): Same.
39820 * tree.c (get_file_function_name): Same.
39822 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39824 PR middle-end/77484
39825 * predict.def (PRED_INDIR_CALL): Set to 86.
39827 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39829 PR preprocessor/54124
39830 * doc/cppopts.texi: Reformat -d subtable to list the full name
39831 of the options. Add cross-reference to the docs for the general
39832 compiler -d options.
39833 * doc/invoke.texi (Developer Options): Add cross-reference to the
39834 preprocessor-specific -d option documentation.
39836 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39838 PR preprocessor/13498
39839 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
39840 redudant material, and reflect new command-line options.
39841 (System Headers): Likewise.
39843 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39845 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
39846 -isystem, and -idirafter. Copy-edit.
39847 * doc/cppopts.texi: Copy-edit. Remove contradiction about
39848 default for -ftrack-macro-expansion. Delete obsolete and
39849 badly-formatted implementation details about -fdebug-cpp output.
39850 * doc/cppwarnopts.texi: Copy-edit.
39852 2017-01-07 David Malcolm <dmalcolm@redhat.com>
39855 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39856 that the transition from a max line width >= 1<<10 to narrower
39857 lines works correctly.
39859 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
39861 * doc/options.texi (PerFunction): New.
39862 * opt-functions.awk (switch_flags): Map both Optimization and
39863 PerFunction to CL_OPTIMIZATION.
39864 * opth-gen.awk: Test for PerFunction flag along with
39866 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
39867 it only when the latter is present. Skip those that don't in
39868 the hash function generator.
39869 * common.opt (fvar-tracking): Mark as PerFunction instead of
39871 (fvar-tracking-assignments): Likewise.
39872 (fvar-tracking-assignments-toggle): Likewise.
39873 (fvar-tracking-uninit): Likewise.
39875 2017-01-07 Jakub Jelinek <jakub@redhat.com>
39877 PR translation/79018
39878 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
39881 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
39884 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
39885 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
39886 TARGET_LONG_JUMP_TABLE_OFFSETS.
39887 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
39888 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
39889 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
39890 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
39891 * config/m68k/m68k.md (tablejump expander): Likewise.
39892 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
39893 TARGET_LONG_JUMP_TABLE_OFFSETS.
39894 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
39895 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
39897 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
39898 David Holsgrove <david.holsgrove@xilinx.com>
39900 * common/config/microblaze/microblaze-common.c
39901 (TARGET_EXCEPT_UNWIND_INFO): Remove.
39902 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
39904 * config/microblaze/microblaze.c (microblaze_must_save_register)
39905 (microblaze_expand_epilogue, microblaze_return_addr): Handle
39907 (microblaze_eh_return): New function.
39908 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
39909 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
39910 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
39911 * config/microblaze/microblaze.md (eh_return): New pattern.
39913 2017-01-06 Jakub Jelinek <jakub@redhat.com>
39915 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
39916 GCC_DIAGNOSTIC_STRINGIFY): Define.
39918 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
39920 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39922 * config/arm/arm.md (<mcrr>): New.
39924 * config/arm/arm.c (arm_arch5te): New.
39925 (arm_option_override): Set arm_arch5te.
39926 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
39928 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
39929 (arm_mcrr_qualifiers): ... this. New.
39930 (MRRC_QUALIFIERS): Define to...
39931 (arm_mrrc_qualifiers): ... this. New.
39932 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
39934 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
39935 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
39936 (MRRCI, mrrc, MRRC): New.
39937 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
39938 VUNSPEC_MRRC2): New.
39940 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39942 * config/arm/arm.md (<mcr>): New.
39944 * config/arm/arm.c (arm_coproc_builtin_available): Add
39945 support for mcr, mrc, mcr2 and mrc2.
39946 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
39947 (arm_mcr_qualifiers): ... this. New.
39948 (MRC_QUALIFIERS): Define to ...
39949 (arm_mrc_qualifiers): ... this. New.
39950 (MCR_QUALIFIERS): Define to ...
39951 (arm_mcr_qualifiers): ... this. New.
39952 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
39954 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
39955 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
39956 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
39957 VUNSPEC_MRC2): New.
39959 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39961 * config/arm/arm.md (*ldc): New.
39965 * config/arm/arm.c (arm_coproc_builtin_available): Add
39966 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
39967 (arm_coproc_ldc_stc_legitimate_address): New.
39968 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
39969 'qualifier_const_pointer'.
39970 (LDC_QUALIFIERS): Define to...
39971 (arm_ldc_qualifiers): ... this. New.
39972 (STC_QUALIFIERS): Define to...
39973 (arm_stc_qualifiers): ... this. New.
39974 * config/arm/arm-protos.h
39975 (arm_coproc_ldc_stc_legitimate_address): New.
39976 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
39977 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
39978 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
39979 stc2, stcl, stc2l): New.
39980 * config/arm/constraints.md (Uz): New.
39981 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
39982 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
39983 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
39984 VUNSPEC_STC2L): New.
39986 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39988 * config/arm/arm.md (<cdp>): New.
39989 * config/arm/arm.c (neon_const_bounds): Rename this ...
39990 (arm_const_bounds): ... this.
39991 (arm_coproc_builtin_available): New.
39992 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
39993 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
39994 (CDP_QUALIFIERS): Define to...
39995 (arm_cdp_qualifiers): ... this. New.
39997 (arm_expand_builtin_args): Add case for 6 arguments.
39998 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
39999 (arm_const_bounds): ... this.
40000 (arm_coproc_builtin_available): New.
40001 * config/arm/arm_acle.h (__arm_cdp): New.
40003 * config/arm/arm_acle_builtins.def (cdp): New.
40005 * config/arm/iterators.md (CDPI,CDP,cdp): New.
40006 * config/arm/neon.md: Rename all 'neon_const_bounds' to
40007 'arm_const_bounds'.
40008 * config/arm/types.md (coproc): New.
40009 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
40010 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
40011 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
40012 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
40014 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40016 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
40017 (UBINOP_QUALIFIERS): New.
40019 (acle_builtin_data): New. Change comment.
40020 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
40021 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
40022 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
40023 arm_acle_builtins.def.
40024 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
40025 (arm_init_acle_builtins): New.
40026 (CRC32_BUILTIN): Remove.
40027 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
40028 crc32cb, crc32ch and crc32cw.
40029 (arm_init_crc32_builtins): Remove.
40030 (arm_init_builtins): Use arm_init_acle_builtins rather
40031 than arm_init_crc32_builtins.
40032 (arm_expand_acle_builtin): New.
40033 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
40034 * config/arm/arm_acle_builtins.def: New.
40036 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
40038 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
40039 (arm_builtin_datum): ... this.
40040 (arm_init_neon_builtin): Rename to ...
40041 (arm_init_builtin): ... this. Add a new parameters PREFIX
40042 and USE_SIG_IN_NAME.
40043 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
40044 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
40045 'arm_builtin_datum'.
40046 (arm_init_vfp_builtins): Likewise.
40047 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
40048 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
40049 (arm_expand_neon_args): Rename to ...
40050 (arm_expand_builtin_args): ... this. Rename builtin_arg
40051 enum values and differentiate between ARG_BUILTIN_MEMORY
40052 and ARG_BUILTIN_NEON_MEMORY.
40053 (arm_expand_neon_builtin_1): Rename to ...
40054 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
40055 values, arm_expand_builtin_args and add bool parameter NEON.
40056 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
40057 (arm_expand_vfp_builtin): Likewise.
40058 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
40060 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40062 PR middle-end/77484
40063 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
40064 * predict.c (tree_estimate_probability_bb): Reverse direction of
40065 polymorphic call predictor.
40067 2017-01-06 David Malcolm <dmalcolm@redhat.com>
40069 * passes.c (execute_one_pass): Split out pass-skipping logic into...
40070 (determine_pass_name_match): ...this new function and...
40071 (should_skip_pass_p): ...this new function.
40073 2017-01-06 Nathan Sidwell <nathan@acm.org>
40075 * ipa-visibility.c (function_and_variable_visibility): Reformat
40076 comments and long lines. Remove extrneous if.
40077 * symtab.c (symtab_node::make_decl_local): Fix code format.
40078 (symtab_node::set_section_for_node): Fix comment typo.
40080 2017-01-06 Martin Liska <mliska@suse.cz>
40083 * lra-constraints.c: Rename invariant to lra_invariant.
40084 * predict.c (set_even_probabilities): Initialize e to NULL.
40086 2017-01-05 Martin Sebor <msebor@redhat.com>
40088 PR tree-optimization/78910
40089 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
40090 (format_integer): Correct off-by-one error in the handling
40091 of precision with negative numbers in signed conversions..
40093 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
40095 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
40097 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40099 PR tree-optimization/71016
40100 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
40101 factor_out_conditional_conversion. Formatting fix.
40102 (factor_out_conditional_conversion): Add cond_stmt argument.
40103 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
40104 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
40107 2017-01-05 David Malcolm <dmalcolm@redhat.com>
40109 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
40110 read-rtl-function.o, and selftest-rtl.o.
40111 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
40112 (selftest::aarch64_test_loading_full_dump): New function.
40113 (selftest::aarch64_run_selftests): New function.
40114 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
40115 selftest::aarch64_run_selftests.
40116 * config/i386/i386.c
40117 (selftest::ix86_test_loading_dump_fragment_1): New function.
40118 (selftest::ix86_test_loading_call_insn): New function.
40119 (selftest::ix86_test_loading_full_dump): New function.
40120 (selftest::ix86_test_loading_unspec): New function.
40121 (selftest::ix86_run_selftests): Call the new functions.
40122 * emit-rtl.c (maybe_set_max_label_num): New function.
40123 * emit-rtl.h (maybe_set_max_label_num): New decl.
40124 * function.c (instantiate_decls): Guard call to
40125 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
40126 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
40128 * gensupport.c (gen_reader::gen_reader): Pass "false"
40129 for new "compact" param of rtx_reader.
40130 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
40131 rather than an empty string for NULL strings.
40132 * read-md.c: Potentially include config.h rather than bconfig.h.
40133 Wrap include of errors.h with #ifdef GENERATOR_FILE.
40134 (have_error): New global, copied from errors.c.
40135 (md_reader::read_name): Rename to...
40136 (md_reader::read_name_1): ...this, adding "out_loc" param,
40137 and converting "missing name or number" to returning false, rather
40139 (md_reader::read_name): Reimplement in terms of read_name_1.
40140 (md_reader::read_name_or_nil): New function.
40141 (md_reader::read_string): Handle "(nil)" by returning NULL.
40142 (md_reader::md_reader): Add new param "compact".
40143 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
40144 (md_reader::read_file): New method.
40145 * read-md.h (md_reader::md_reader): Add new param "compact".
40146 (md_reader::read_file): New method.
40147 (md_reader::is_compact): New accessor.
40148 (md_reader::read_name): Convert return type from void to file_location.
40149 (md_reader::read_name_or_nil): New decl.
40150 (md_reader::read_name_1): New decl.
40151 (md_reader::m_compact): New field.
40152 (noop_reader::noop_reader): Pass "false" for new "compact" param
40154 (rtx_reader::rtx_reader): Add new "compact" param.
40155 (rtx_reader::read_rtx_operand): Make virtual and convert return
40156 type from void to rtx.
40157 (rtx_reader::read_until): New decl.
40158 (rtx_reader::handle_any_trailing_information): New virtual function.
40159 (rtx_reader::postprocess): New virtual function.
40160 (rtx_reader::finalize_string): New virtual function.
40161 (rtx_reader::m_in_call_function_usage): New field.
40162 (rtx_reader::m_reuse_rtx_by_id): New field.
40163 * read-rtl-function.c: New file.
40164 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
40165 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
40166 (selftest::verify_three_block_rtl_cfg): New decl.
40167 * read-rtl-function.h: New file.
40168 * read-rtl.c: Potentially include config.h rather than bconfig.h.
40169 For host, include function.h, memmodel.h, and emit-rtl.h.
40170 (one_time_initialization): New function.
40171 (struct compact_insn_name): New struct.
40172 (compact_insn_names): New array.
40173 (find_code): Handle insn codes in compact dumps.
40174 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
40175 (bind_subst_iter_and_attr): Likewise.
40176 (add_condition_to_string): Likewise.
40177 (add_condition_to_rtx): Likewise.
40178 (apply_attribute_uses): Likewise.
40179 (add_current_iterators): Likewise.
40180 (apply_iterators): Likewise.
40181 (initialize_iterators): Guard usage of apply_subst_iterator with
40182 #ifdef GENERATOR_FILE.
40183 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
40184 (md_reader::read_mapping): Likewise.
40185 (add_define_attr_for_define_subst): Likewise.
40186 (add_define_subst_attr): Likewise.
40187 (read_subst_mapping): Likewise.
40188 (check_code_iterator): Likewise.
40189 (rtx_reader::read_rtx): Likewise. Move one-time initialization
40191 (one_time_initialization): New function.
40192 (rtx_reader::read_until): New method.
40193 (read_flags): New function.
40194 (parse_reg_note_name): New function.
40195 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
40196 Handle reuse_rtx ids.
40197 Wrap iterator lookup within #ifdef GENERATOR_FILE.
40198 Add parsing support for RTL dumps, mirroring the special-cases in
40199 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
40200 values, and calling handle_any_trailing_information.
40201 (rtx_reader::read_rtx_operand): Convert return type from void
40202 to rtx, returning return_rtx. Handle case 'e'. Call
40203 finalize_string on XSTR and XTMPL fields.
40204 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
40205 "(nil)" values were omitted. Call the postprocess vfunc on the
40207 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
40208 class ctor. Initialize m_in_call_function_usage. Call
40209 one_time_initialization.
40210 * rtl-tests.c (selftest::test_uncond_jump): Call
40211 set_new_first_and_last_insn.
40212 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
40213 * selftest-rtl.c: New file.
40214 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
40215 (selftest::get_insn_by_uid): New decl.
40216 * selftest-run-tests.c (selftest::run_tests): Call
40217 read_rtl_function_c_tests.
40218 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
40219 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
40222 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
40224 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
40225 operands in a special way. Assert that pos+len <= mode precision.
40227 2017-01-05 Jakub Jelinek <jakub@redhat.com>
40229 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
40230 3 argument Alias with unlimited for the negative form.
40231 (fno-vect-cost-model): Removed.
40233 2017-01-05 Martin Liska <mliska@suse.cz>
40235 * hsa-gen.c (gen_hsa_divmod): New function.
40236 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
40238 2017-01-05 Martin Liska <mliska@suse.cz>
40241 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
40244 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40246 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
40247 small constant length operands.
40249 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40251 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
40252 between loop iterations.
40254 2017-01-05 Martin Liska <mliska@suse.cz>
40257 * gimplify.c (gimplify_decl_expr): Compare to
40258 asan_poisoned_variables instread of checking flags.
40259 (gimplify_target_expr): Likewise.
40260 (gimplify_expr): Likewise.
40261 (gimplify_function_tree): Conditionally initialize
40262 asan_poisoned_variables.
40264 2017-01-04 Jeff Law <law@redhat.com>
40266 PR tree-optimizatin/78812
40267 * rtl.h (contains_mem_rtx_p): Prototype.
40268 * ifcvt.c (containts_mem_rtx_p): Move from here to...
40269 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
40270 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
40271 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
40272 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
40274 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40276 * input.c (assert_char_at_range): Default-initialize actual_range.
40278 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40280 * df-scan.c (df_ref_create_structure): Make regno unsigned,
40281 to match the caller.
40283 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40285 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
40286 insns after final jump in test to emit dummy move.
40288 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40290 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
40291 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
40293 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
40295 * multiple_target.c (create_dispatcher_calls): Init e_next.
40296 * tree-ssa-loop-split.c (split_loop): Init border.
40297 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
40300 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
40305 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
40306 (altivec_register_operand): Do not return true if the operand
40307 contains a SUBREG mixing SImode and SFmode.
40308 (vsx_register_operand): Likewise.
40309 (vsx_reg_sfsubreg_ok): New predicate.
40310 (vfloat_operand): Do not return true if the operand contains a
40311 SUBREG mixing SImode and SFmode.
40312 (vint_operand): Likewise.
40313 (vlogical_operand): Likewise.
40314 (gpc_reg_operand): Likewise.
40315 (int_reg_operand): Likewise.
40316 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
40317 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
40318 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
40320 (rs6000_emit_move_si_sf_subreg): New helper function.
40321 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
40322 fixup SUBREGs involving SImode and SFmode.
40323 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
40324 numbers for the new peephole2 optimization.
40325 (peephole2 for SFmode unions): New peephole2 to optimize cases in
40326 the GLIBC math library that do AND/IOR/XOR operations on single
40327 precision floating point.
40328 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
40329 target macros to say whether we need to avoid SUBREGs mixing
40331 (TARGET_ALLOW_SF_SUBREG): Likewise.
40332 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
40333 (UNSPEC_SI_FROM_SF): Likewise.
40334 (iorxor): Change spacing.
40335 (and_ior_xor): New iterator for AND, IOR, and XOR.
40336 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
40337 (movdi_from_sf_zero_ext): Likewise.
40338 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
40339 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
40340 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
40341 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
40342 (fms<mode>4): Likewise.
40343 (fnma<mode>4): Likewise.
40344 (fnms<mode>4): Likewise.
40345 (nfma<mode>4): Likewise.
40346 (nfms<mode>4): Likewise.
40348 2017-01-04 Marek Polacek <polacek@redhat.com>
40351 * doc/invoke.texi: Document -Wpointer-compare.
40353 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40355 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
40358 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
40359 descriptions for -gdwarf-5 and emit them as uleb128 instead of
40362 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
40365 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
40366 documentation of the powerpc_popcntb_ok attribute.
40367 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
40368 code to issue warning messages if a requested CPU configuration is
40369 not supported by the binary (assembler and loader) toolchain.
40370 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
40371 made to define a built-in function that has been disabled.
40372 (paired_init_builtins): Add assertion to prevent ICE if attempt is
40373 made to define a built-in function that has been disabled.
40374 (altivec_init_builtins): Add comment explaining why definition
40375 of the DST built-in functions is not preceded by an assertion
40376 check. Add assertions to prevent ICE if attempts are made to
40377 define an altivec predicate or an abs* built-in function that has
40379 (htm_init_builtins): Add comment explaining why definition of the
40380 htm built-in functions is not preceded by an assertion check.
40382 2017-01-04 Jeff Law <law@redhat.com>
40384 PR tree-optimizatin/67955
40385 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
40386 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
40387 the points-to solution does not include pt_null. Use DECL_PT_UID
40390 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
40392 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
40393 Use gen_int_mode instead of gen_lopwart for const_int operands.
40395 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40397 PR tree-optimization/71563
40398 * match.pd: Simplify X << Y into X if Y is known to be 0 or
40399 out of range value - has low bits known to be zero.
40401 2017-01-04 Alan Modra <amodra@gmail.com>
40403 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
40404 * configure: Regenerate.
40405 * config.in: Regenerate.
40407 2017-01-04 Jakub Jelinek <jakub@redhat.com>
40410 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
40411 a substring of the message, but strcmp with the whole message. Ifdef
40412 ENABLE_NLS, translate the message first using dgettext.
40414 2017-01-03 Jeff Law <law@redhat.com>
40416 PR tree-optimizatin/78856
40417 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
40418 (mark_threaded_blocks): Remove code to truncate thread paths that
40419 cross multiple loop headers. Instead invalidate the cached loop
40420 iteration information and handle case of a thread path walking
40421 into an irreducible region.
40423 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
40426 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
40427 assertions. Add support for doing the signbit if the IEEE 128-bit
40428 floating point value is in a GPR.
40429 * config/rs6000/rs6000.md (Fsignbit): Delete.
40430 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
40431 Update the length attribute if the value is in a GPR.
40432 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
40433 the sign or zero extension instruction, since the value is always 0/1.
40434 (signbit<mode>2_dm2): Delete using <Fsignbit>.
40437 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
40438 extracting SImode to a GPR register so that we can generate a
40439 store, limit the vector to be in a traditional Altivec register
40440 for the vextuwrx instruction.
40442 2017-01-03 Ian Lance Taylor <iant@google.com>
40444 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
40446 2017-01-03 Martin Sebor <msebor@redhat.com>
40448 PR tree-optimization/78696
40449 * gimple-ssa-sprintf.c (format_floating): Correct handling of
40450 precision. Use MPFR for %f for greater fidelity. Correct handling
40452 (pass_sprintf_length::compute_format_length): Set width and precision
40453 specified by asrerisk to void_node for vararg functions.
40454 (try_substitute_return_value): Adjust dump output.
40456 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
40458 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
40460 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
40462 * doc/invoke.texi (SPARC options): Document -mlra as the default.
40463 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
40464 -mlra/-mno-lra was passed to the compiler.
40466 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
40468 PR rtl-optimization/65618
40469 * emit-rtl.c (try_split): Move initialization of "before" and
40470 "after" to just before the call to emit_insn_after_setloc.
40472 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
40474 * doc/md.texi (Standard Names): Remove reference to Java frontend.
40476 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
40478 * dwarf2out.c (gen_enumeration_type_die): When
40479 -gno-strict-dwarf, add a DW_AT_encoding attribute.
40481 2017-01-03 Jakub Jelinek <jakub@redhat.com>
40483 PR tree-optimization/78965
40484 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
40485 Change first argument from const call_info & to call_info &. For %n
40486 set info.nowrite to false.
40488 PR middle-end/78901
40489 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
40490 possibly throwing calls.
40492 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
40493 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
40494 and fns handling, rather than in a separate case SSA_NAME.
40496 2017-01-02 Jeff Law <law@redhat.com>
40498 * config/darwin-driver.c (darwin_driver_init): Const-correctness
40499 fixes for first_period and second_period variables.
40501 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
40504 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
40505 (*insvqi_1): New insn pattern.
40506 (*insvqi_1_mem_rex64): Ditto.
40507 (*insvqi_2): Ditto.
40508 (*insvqi_3): Rename from *insvqi.
40510 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
40512 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
40514 * doc/cfg.texi (Edges): Remove reference to Java.
40515 (Maintaining the CFG): Ditto.
40517 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40519 PR middle-end/77674
40520 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
40521 transparent aliases.
40523 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
40525 PR middle-end/77484
40526 * predict.def (PRED_CALL): Update hitrate.
40527 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
40528 * predict.c (tree_estimate_probability_bb): Split CALL predictor
40529 into direct/indirect/polymorphic variants.
40531 2017-01-01 Jakub Jelinek <jakub@redhat.com>
40533 Update copyright years.
40535 * gcc.c (process_command): Update copyright notice dates.
40536 * gcov-dump.c (print_version): Ditto.
40537 * gcov.c (print_version): Ditto.
40538 * gcov-tool.c (print_version): Ditto.
40539 * gengtype.c (create_file): Ditto.
40540 * doc/cpp.texi: Bump @copying's copyright year.
40541 * doc/cppinternals.texi: Ditto.
40542 * doc/gcc.texi: Ditto.
40543 * doc/gccint.texi: Ditto.
40544 * doc/gcov.texi: Ditto.
40545 * doc/install.texi: Ditto.
40546 * doc/invoke.texi: Ditto.
40548 Copyright (C) 2017 Free Software Foundation, Inc.
40550 Copying and distribution of this file, with or without modification,
40551 are permitted in any medium without royalty provided the copyright
40552 notice and this notice are preserved.