1 2017-10-10 Jim Wilson <wilson@tuliptree.org>
3 PR rtl-optimization/81434
4 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
5 comment for main loop. In sched_group_found if, also add checks for
6 pass and min_cost_group.
8 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
10 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
11 (rs6000_insn_cost): New function.
12 * config/rs6000/rs6000.md (cost): New attribute.
14 2017-10-10 Jakub Jelinek <jakub@redhat.com>
15 H.J. Lu <hongjiu.lu@intel.com>
19 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
20 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
21 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
22 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
23 to OPTION_MASK_ISA_AVX512VL - builtins that have both
24 OPTION_MASK_ISA_MMX and some other bit set require both
25 mmx and the ISAs without the mmx bit.
26 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
27 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
28 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
29 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
30 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
31 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
32 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
33 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
34 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
35 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
36 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
37 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
38 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
39 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
40 Add OPTION_MASK_ISA_MMX.
42 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
44 * config.gcc (armv7*-*-freebsd*): New target.
45 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
47 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
49 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI, X86_TUNE_ADJUST_UNROLL,
50 X86_TUNE_ONE_IF_CONV_INSN): Move to right spot in the file.
52 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
54 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
56 (WIDE_INT_REF_FOR): Update accordingly.
57 * tree.h (wi::int_traits <const_tree>): Delete.
58 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
59 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
60 (wi::tree_to_wide_ref): New typedef.
61 (wi::to_wide): New function.
62 * calls.c (get_size_range): Use wi::to_wide when operating on
64 * cgraph.c (cgraph_node::create_thunk): Likewise.
65 * config/i386/i386.c (ix86_data_alignment): Likewise.
66 (ix86_local_alignment): Likewise.
67 * dbxout.c (stabstr_O): Likewise.
68 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
69 * expr.c (const_vector_from_tree): Likewise.
70 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
71 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
72 (fold_negate_expr_1, int_const_binop_1, const_binop)
73 (fold_convert_const_int_from_real, optimize_bit_field_compare)
74 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
75 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
76 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
77 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
78 (fold_not_const, round_up_loc): Likewise.
79 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
80 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
81 (alloca_call_type): Likewise.
82 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
83 * godump.c (go_output_typedef): Likewise.
84 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
85 * internal-fn.c (get_min_precision): Likewise.
86 * ipa-cp.c (ipcp_store_vr_results): Likewise.
87 * ipa-polymorphic-call.c
88 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
89 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
90 (ipa_modify_call_arguments): Likewise.
92 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
93 * print-tree.c (print_node_brief, print_node): Likewise.
94 * stmt.c (expand_case): Likewise.
95 * stor-layout.c (layout_type): Likewise.
96 * tree-affine.c (tree_to_aff_combination): Likewise.
97 * tree-cfg.c (group_case_labels_stmt): Likewise.
98 * tree-data-ref.c (dr_analyze_indices): Likewise.
99 (prune_runtime_alias_test_list): Likewise.
100 * tree-dump.c (dequeue_and_dump): Likewise.
101 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
102 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
103 * tree-pretty-print.c (dump_generic_node): Likewise.
104 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
105 (simple_iv_with_niters): Likewise.
106 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
107 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
108 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
109 * tree-ssa-loop-niter.c (split_to_var_and_offset)
110 (refine_value_range_using_guard, number_of_iterations_ne_max)
111 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
112 (get_cst_init_from_scev, record_nonwrapping_iv)
113 (scev_var_range_cant_overflow): Likewise.
114 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
115 * tree-ssa-pre.c (compute_avail): Likewise.
116 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
117 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
118 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
119 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
120 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
121 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
122 (dump_case_nodes, try_switch_expansion): Likewise.
123 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
124 (vect_do_peeling): Likewise.
125 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
126 * tree-vect-stmts.c (vectorizable_load): Likewise.
127 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
128 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
129 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
130 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
131 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
132 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
133 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
134 (evrp_dom_walker::before_dom_children): Likewise.
135 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
136 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
137 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
138 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
139 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
140 (get_range_pos_neg): Likewise.
141 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
142 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
143 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
144 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
145 * config/avr/avr.c (avr_fold_builtin): Likewise.
146 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
147 * config/msp430/msp430.c (msp430_attr): Likewise.
148 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
149 * config/powerpcspe/powerpcspe-c.c
150 (altivec_resolve_overloaded_builtin): Likewise.
151 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
152 (rs6000_expand_ternop_builtin): Likewise.
153 * config/rs6000/rs6000-c.c
154 (altivec_resolve_overloaded_builtin): Likewise.
155 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
156 (rs6000_expand_ternop_builtin): Likewise.
157 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
159 2017-10-10 Bin Cheng <bin.cheng@arm.com>
161 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
162 when copying loop nest with only one inner loop.
164 2017-10-10 Richard Biener <rguenther@suse.de>
166 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
167 blocks if SCEV is active.
168 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
170 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
171 (scev_initialize): Assert we are not yet initialized.
173 2017-10-10 Bin Cheng <bin.cheng@arm.com>
175 * tree-loop-distribution.c (generate_loops_for_partition): Remove
176 inner loop's exit stmt by making it always exit the loop, otherwise
177 we would generate an infinite empty loop.
179 2017-10-10 Bin Cheng <bin.cheng@arm.com>
181 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
182 renaming variables in new preheader if it's deleted.
184 2017-10-10 Bin Cheng <bin.cheng@arm.com>
186 * tree-loop-distribution.c (struct partition): Remove unused field
187 loops of the structure.
188 (partition_alloc, partition_free): Ditto.
189 (build_rdg_partition_for_vertex): Ditto.
191 2017-10-09 Jeff Law <law@redhat.com>
193 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
194 return type to match prototype and documentation.
196 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
198 * config/rs6000/rs6000.c (processor_costs): Move to ...
199 * config/rs6000/rs6000.h: ... here.
200 (rs6000_cost): Declare.
202 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
204 * except.c (setjmp_fn): New global variable.
205 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
206 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
207 if DONT_USE_BUILTIN_SETJMP is defined.
209 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
211 * target.def (insn_cost): New hook.
212 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
213 * doc/tm.texi: Regenerate.
214 * rtlanal.c (insn_cost): Use the new hook.
216 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
218 * combine.c (combine_validate_cost): Compute the new insn_cost,
219 not just pattern_cost.
220 (try_combine): Adjust comment.
222 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
224 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
226 * combine.c (uid_insn_cost): Adjust comment.
227 (combine_validate_cost): Adjust comment. Use pattern_cost instead
229 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
230 * dse.c (find_shift_sequence): Ditto.
231 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
232 (bb_valid_for_noce_process_p): Use pattern_cost.
233 * rtl.h (insn_rtx_cost): Delete.
234 (pattern_cost): New prototype.
235 (insn_cost): New prototype.
236 * rtlanal.c (insn_rtx_cost): Rename to...
237 (pattern_cost): ... this.
240 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
242 * config/i386/i386.md (*jcc_2): Remove insn pattern.
243 (*jcc<mode>_0_r_i387): Ditto.
244 (*jccxf_r_i387): Ditto.
245 (*jcc<mode>_r_i387): Ditto.
246 (*jccu<mode>_r_i387): Ditto.
247 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
248 (*jcc): Rename from *jcc_1.
250 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
252 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
253 deferred rescans after the lvx/stvx recombination pre-pass.
255 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
257 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
258 memory operation instruction support.
259 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
260 (rs6000-ibm-aix[789]*): Likewise.
261 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
262 Document new functions.
264 2017-10-09 Richard Biener <rguenther@suse.de>
266 PR tree-optimization/82397
267 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
268 equality only for semantically equal trees.
270 2017-10-09 Richard Biener <rguenther@suse.de>
272 PR tree-optimization/82449
273 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
274 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
275 allow constant addresses.
276 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
279 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
281 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
284 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
287 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
290 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
293 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
295 2017-10-09 Jakub Jelinek <jakub@redhat.com>
298 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
299 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
301 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
303 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
304 (WI_BINARY_PREDICATE_RESULT): Likewise.
305 (wi::binary_traits::operator_result): New type.
306 (wi::binary_traits::predicate_result): Likewise.
307 (generic_wide_int::operator~, unary generic_wide_int::operator-)
308 (generic_wide_int::operator==, generic_wide_int::operator!=)
309 (generic_wide_int::operator&, generic_wide_int::and_not)
310 (generic_wide_int::operator|, generic_wide_int::or_not)
311 (generic_wide_int::operator^, generic_wide_int::operator+
312 (binary generic_wide_int::operator-, generic_wide_int::operator*):
314 (operator~, unary operator-, operator==, operator!=, operator&)
315 (operator|, operator^, operator+, binary operator-, operator*): New
317 * expr.c (get_inner_reference): Use wi::bit_and_not.
318 * fold-const.c (fold_binary_loc): Likewise.
319 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
320 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
321 (bit_value_binop): Likewise.
322 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
323 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
324 (extract_range_from_binary_expr_1): Likewise.
325 (masked_increment): Likewise.
326 (simplify_bit_ops_using_ranges): Likewise.
328 2017-10-09 Martin Jambor <mjambor@suse.cz>
331 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
332 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
333 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
335 (hsa_fixup_mov_insn_type): New function.
336 (hsa_op_with_type::get_in_type): Use it.
337 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
338 immediates in an assert.
339 (hsa_op_with_type::extend_int_to_32bit): New method.
340 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
341 types. Convert to dest type if necessary.
342 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
343 (reg_for_gimple_ssa): Pass false as min32int to
344 hsa_type_for_scalar_tree_type.
345 (gen_hsa_addr): Fixup type when creating addresable temporary.
346 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
347 (gen_hsa_unary_operation): Extend operands and convert to dest type if
348 necessary. Call hsa_fixup_mov_insn_type.
349 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
350 extend operands and convert to dest type if necessary.
351 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
352 to dest type if necessary.
353 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
354 if conversion nt necessary and size matches.
355 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
356 to dest type if necessary.
357 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
358 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
360 (gen_hsa_clrsb): Likewise.
361 (gen_hsa_ffs): Likewise.
362 (gen_hsa_divmod): Extend operands and convert to dest type if
364 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
366 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
368 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
369 Remove empty default arguments. Use a brace block as output
371 (conditional return): Ditto.
373 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
374 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
376 (group_ending_nop): Ditto.
378 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
379 (splitters for those): Ditto.
381 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
383 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
384 a conditional jump (and the compare for it) so that pc_rtx is the
386 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
387 for the deleted and renamed ctr<mode>_internal[234] patterns.
388 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
389 Delete second conditional return pattern.
390 (ctr<mode>_internal2): Delete this second bdnz pattern.
391 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
392 (ctr<mode>_internal4): Delete this second bdz pattern.
394 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
396 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
397 (always_initialized_rtx_for_ssa_name_p): New predicate.
398 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
399 (finish_out_of_ssa): Free new field of SA.
400 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
401 * tree-ssa-coalesce.c: Include tree-ssa.h.
402 (get_parm_default_def_partitions): Remove extern keyword.
403 (get_undefined_value_partitions): New function.
404 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
405 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
407 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
409 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
411 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
413 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
414 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
416 * i386.c (ix86_expand_set_or_movmem): Disable 512bit loops for targets
419 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
421 * i386.c (has_dispatch): Disable for Ryzen.
423 2017-10-08 Olivier Hainque <hainque@adacore.com>
425 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
426 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
428 (thumb_set_return_address): Likewise.
430 2017-10-08 Olivier Hainque <hainque@adacore.com>
432 * common/config/arm/arm-common.c (arm_except_unwind_info):
433 Handle DWARF2_UNWIND_INFO.
435 2017-10-07 Michael Collison <michael.collison@arm.com>
437 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
440 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
442 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
443 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
444 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
445 defined, force the creation of a new block for a dispatch label.
447 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
449 * invoke.texi (Wsuggest-attribute=cold): Document.
450 * common.opt (Wsuggest-attribute=cold): New
451 * ipa-pure-const.c (warn_function_cold): New function.
452 * predict.c (compute_function_frequency): Use it.
453 * predict.h (warn_function_cold): Declare.
455 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
457 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
460 2017-10-06 Martin Liska <mliska@suse.cz>
462 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
463 keyword for member functions.
464 (struct sanopt_tree_couple): New struct.
465 (struct sanopt_tree_couple_hash): New function.
466 (struct sanopt_ctx): Add new hash_map.
467 (has_dominating_ubsan_ptr_check): New function.
468 (record_ubsan_ptr_check_stmt): Likewise.
469 (maybe_optimize_ubsan_ptr_ifn): Likewise.
470 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
471 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
473 2017-10-06 Sudakshina Das <sudi.das@arm.com>
476 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
477 aarch64_simd_valid_immediate on CONST_VECTORs.
478 (aarch64_reg_or_bic_imm): Likewise.
480 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
482 PR rtl-optimization/82396
483 * haifa-sched.c (ready_sort_real): Disable qsort checking.
485 2017-10-06 Sebastian Pop <sebpop@gmail.com>
487 * graphite-dependences.c (scop_get_reads): Move code to...
488 (scop_get_must_writes): Move code to...
489 (scop_get_may_writes): Move code to...
490 (scop_get_reads_and_writes): ... here.
491 (scop_get_dependences): Call scop_get_reads_and_writes.
493 2017-10-06 Jakub Jelinek <jakub@redhat.com>
495 PR tree-optimization/82434
496 * fold-const.h (can_native_encode_type_p,
497 can_native_encode_string_p): Remove.
498 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
499 don't encode anything, just return what would be otherwise returned.
500 (native_encode_fixed, native_encode_complex, native_encode_vector):
502 (native_encode_string): Likewise. Inline by hand
503 can_native_encode_string_p.
504 (can_native_encode_type_p): Remove.
505 (can_native_encode_string_p): Remove.
506 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
507 STRING_CSTs using can_native_encode_string_p, test all
508 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
509 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
510 argument from native_encode_expr.
511 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
512 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
515 2017-10-06 Richard Biener <rguenther@suse.de>
517 PR tree-optimization/82397
518 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
519 operand_equal_p but rely on data_ref_compare_tree for detecting
521 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
522 to match up with dr_group_sort_cmp.
524 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
527 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
529 * config/s390/s390-builtin-types.def: Regenerate.
531 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
534 * config/s390/s390-builtin-types.def: Regenerate.
535 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
536 Change flag from B_VXE to B_VX.
537 (s390_vec_min_dbl): Remove B_VXE flag.
539 2017-10-06 Richard Biener <rguenther@suse.de>
541 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
542 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
543 translate_isl_ast_to_gimple::is_valid_rename,
544 translate_isl_ast_to_gimple::get_rename,
545 translate_isl_ast_to_gimple::get_def_bb_for_const,
546 translate_isl_ast_to_gimple::get_new_name,
547 translate_isl_ast_to_gimple::collect_all_ssa_names,
548 translate_isl_ast_to_gimple::copy_loop_phi_args,
549 translate_isl_ast_to_gimple::collect_all_ssa_names,
550 translate_isl_ast_to_gimple::copy_loop_phi_args,
551 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
552 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
553 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
554 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
555 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
556 translate_isl_ast_to_gimple::copy_cond_phi_args,
557 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
558 translate_isl_ast_to_gimple::edge_for_new_close_phis,
559 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
560 translate_isl_ast_to_gimple::rename_uses,
561 translate_isl_ast_to_gimple::rename_all_uses): Remove.
562 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
563 (set_rename_for_each_def): Likewise.
564 (graphite_copy_stmts_from_block): Handle debug stmt resetting
565 here. Handle rewriting SCEV analyzable uses here.
566 (copy_bb_and_scalar_dependences): Generate code for PHI
568 (graphite_regenerate_ast_isl): Adjust.
569 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
570 (add_write, add_read): New functions.
571 (build_cross_bb_scalars_def): Use it and simplify.
572 (build_cross_bb_scalars_use): Likewise.
573 (graphite_find_cross_bb_scalar_vars): Inline into...
574 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
575 simulating out-of-SSA. Compute liveout and add dependencies.
576 (build_scops): Force an empty entry block.
577 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
579 (sese_build_liveouts): Declare.
580 (sese_trivially_empty_bb_p): Likewise.
581 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
582 compute liveout and debug_liveout.
583 (sese_bad_liveouts_use): Remove.
584 (sese_reset_debug_liveouts_bb): Likewise.
585 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
586 (sese_build_liveouts): Build liveout and debug_liveout and store
588 (new_sese_info): Adjust.
589 (free_sese_info): Likewise.
590 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
591 do not build liveout here.
592 (move_sese_in_condition): Adjust region entry.
593 (scev_analyzable_p): Match up with chrec_apply requirements.
594 (sese_trivially_empty_bb_p): New.
595 * tree-into-ssa.c (get_reaching_def): Properly support generating
596 default-defs for incremental rewrite of anonymous names.
598 2017-10-06 Richard Biener <rguenther@suse.de>
600 * graphite-sese-to-poly.c (extract_affine): For casts increasing
601 precision do not perform modulo reduction.
603 2017-10-06 Richard Biener <rguenther@suse.de>
605 PR tree-optimization/82436
606 * tree-vect-slp.c (vect_supported_load_permutation_p): More
607 conservatively choose the vectorization factor when checking
608 whether we can perform the required load permutation.
609 (vect_transform_slp_perm_load): Assert when we may not fail.
611 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
613 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
614 message for incompatible -msdata=* and -mcall-* options.
616 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
618 * i386.c (ia32_multipass_dfa_lookahead): Default to issue rate
619 for post-reload scheduling.
621 2017-10-05 Tamar Christina <tamar.christina@arm.com>
623 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
625 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
627 * i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
628 to improve monte carlo in scimark.
630 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
632 * i386.c (ix86_size_cost, i386_cost, i486_cost, pentium_cost,
633 lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
634 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
635 pentium4_cost, nocona_cost): Set reassociation width to 1.
636 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
637 width to 2 for fp operations and 1 otherwise.
638 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
640 (atom_cost): Set reassociation width to 2.
641 (slm_cost, generic_cost): Set fp reassociation width to 2 and 1 otherwise.
642 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
643 (core_cost): Set fp reassociation width to 4 and vector to 2.
644 (ix86_reassociation_width): Rewrite using cost table; special case
645 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
646 and TARGET_AVX128_OPTIMAL.
647 * i386.h (processor_costs): Add
648 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
649 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
650 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
651 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
652 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
653 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
655 2017-10-05 Nathan Sidwell <nathan@acm.org>
657 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
659 2017-10-05 Tamar Christina <tamar.christina@arm.com>
661 * config/arm/arm.c (arm_test_fpu_data): New.
662 (arm_run_selftests): Call arm_test_fpu_data.
664 2017-10-04 Nathan Sidwell <nathan@acm.org>
666 * toplev.c (toplev::main): Remove excess parens on pretty_printer
668 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
670 2017-10-04 Sudakshina Das <sudi.das@arm.com>
672 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
673 check type for aarch64_simd_valid_immediate.
674 (aarch64_output_simd_mov_immediate): Update prototype.
675 (aarch64_simd_valid_immediate): Update prototype.
676 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
677 support for ORR-immediate.
678 (and<mode>3): modified pattern to add support for BIC-immediate.
679 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
680 now checks for valid immediate for BIC and ORR based on new enum
682 (aarch64_output_simd_mov_immediate): Function now used to output
683 BIC/ORR imm as well based on new enum argument.
684 * config/aarch64/constraints.md (Do): New vector immediate constraint.
686 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
687 (aarch64_reg_or_bic_imm): Likewise.
689 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
691 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
692 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
694 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
698 PR rtl-optimization/82396
699 * haifa-sched.c (autopref_multipass_init): Simplify
701 (autopref_rank_data): Simplify sort order.
702 * sched-int.h (autopref_multipass_data_): Remove
703 multi_mem_insn_p, min_offset and max_offset.
705 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
707 * doc/sourcebuild.texi: Document vect_peeling_profitable.
709 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
711 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
714 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
716 * doc/sourcebuild.texi: Document vect_long_mult.
718 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
720 PR tree-optimization/82413
721 * fold-const.c (build_range_check): Use widest_int when comparing
722 the maximum ETYPE value with HIGH.
724 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
726 PR rtl-optimization/82396
727 * haifa-sched.c (autopref_multipass_init): Simplify
729 (autopref_rank_data): Simplify sort order.
730 * sched-int.h (autopref_multipass_data_): Remove
731 multi_mem_insn_p, min_offset and max_offset.
733 2017-10-04 Jakub Jelinek <jakub@redhat.com>
735 PR tree-optimization/82381
736 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
737 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
740 PR tree-optimization/82374
741 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
742 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
743 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
744 current_function_decl to the new decl.
746 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
748 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
749 helper macro for IEEE float128 hardware built-in functions.
750 (SQRTF128_ODD): Add built-in functions with the round-to-odd
752 (TRUNCF128_ODD): Likewise.
753 (ADDF128_ODD): Likewise.
754 (SUBF128_ODD): Likewise.
755 (MULF128_ODD): Likewise.
756 (DIVF128_ODD): Likewise.
757 (FMAF128_ODD): Likewise.
758 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
759 UNSPEC_TRUNC_ROUND_TO_ODD.
760 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
761 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
762 floating point round to odd instructions.
763 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
764 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
765 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
766 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
767 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
768 (trunc<mode>sf2_hw): Change the truncate with round to odd
769 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
770 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
771 to odd hardware instructions.
772 (sub<mode>3_odd): Likewise.
773 (mul<mode>3_odd): Likewise.
774 (div<mode>3_odd): Likewise.
775 (sqrt<mode>2_odd): Likewise.
776 (fma<mode>4_odd): Likewise.
777 (fms<mode>4_odd): Likewise.
778 (nfma<mode>4_odd): Likewise.
779 (nfms<mode>4_odd): Likewise.
780 (trunc<mode>df2_odd): Change the truncate with round to odd
781 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
783 * doc/extend.texi (PowerPC built-in functions): Update documentation
784 for existing IEEE float128-bit built-in functions. Add built-in
785 functions that generate the IEEE 128-bit floating point round to
788 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
790 PR rtl-optimization/77729
791 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
792 to (X&(C1&~C2))|C2 transformations.
794 2017-10-03 Martin Jambor <mjambor@suse.cz>
796 PR tree-optimization/82363
797 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
798 mismatch, mark lacc written regardless of racc.
800 2017-10-03 Jakub Jelinek <jakub@redhat.com>
802 PR tree-optimization/82381
803 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
804 stmt_to_insert nor wheather SSA_NAMEs are default defs.
805 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
806 fallthrough into reassoc_stmt_dominates_stmt_p.
809 * combine.c (combine_instructions): Don't combine in unreachable
812 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
815 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
816 function to not use the have_cpu variable. Do not set cpu_index,
817 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
819 (rs6000_valid_attribute_p): Remove duplicate initializations of
820 old_optimize and func_optimize.
821 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
822 (rs6000_activate_target_options): Make global.
823 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
826 2017-10-02 Jakub Jelinek <jakub@redhat.com>
828 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
829 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
830 Set *poffset to 0 and *psize and *pmax_size to -1 if
831 *poffset + *psize overflows in HOST_WIDE_INT.
833 PR tree-optimization/82387
834 PR tree-optimization/82388
835 PR tree-optimization/82389
836 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
837 instead of live_bytes non-NULL.
839 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
842 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
845 2017-10-02 Richard Biener <rguenther@suse.de>
847 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
848 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
849 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
851 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
853 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
854 requested precision matches the type's.
855 * calls.c (alloc_max_size): Calculate the new candidate size as
856 a widest_int and use wi::to_widest when comparing it with the
857 current candidate size.
858 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
859 zero rather than integer_zero_node.
860 * match.pd: Check for a no-op conversion before using wi::add
861 rather than after. Use tree_to_uhwi when summing small shift
862 counts into an unsigned int.
864 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
865 Alan Hayward <alan.hayward@arm.com>
866 David Sherwood <david.sherwood@arm.com>
869 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
870 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
871 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
874 2017-10-02 Richard Biener <rguenther@suse.de>
876 PR tree-optimization/82355
877 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
878 a mapping for the enclosing loop but avoid generating one for
880 (copy_bb_and_scalar_dependences): Remove premature codegen
881 error on PHIs in blocks duplicated into multiple places.
882 * graphite-scop-detection.c
883 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
884 in the region use it as loop and nest to analyze the DR in.
885 (try_generate_gimple_bb): Likewise.
886 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
887 (add_loop_constraints): For blocks in a loop not in the region
888 create a dimension with a single iteration.
889 * sese.h (gbb_loop_at_index): Remove assert.
891 2017-10-01 Kevin Buettner <kevinb@redhat.com>
893 * omp-expand.c (adjust_context_scope): New function.
894 (expand_parallel_call): Call adjust_context_scope.
896 2017-10-01 Jeff Law <law@redhat.com>
898 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
899 dom_opt_dom_walker class with direct access to private members.
900 Add comments. Call test_for_singularity.
901 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
902 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
903 m_dummy_cond anymore.
904 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
906 (pass_dominator:execute): Build the dummy_cond here and pass it
907 to the dom_opt_dom_walker ctor.
908 (test_for_singularity): New function.
910 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
911 Maya Rashish <coypu@sdf.org>
913 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
914 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
915 (alpha*-*-netbsd*) Use nbsd_tm_file.
916 (arm*-*-netbsdelf*) Likewise.
917 (i[34567]86-*-netbsdelf*) Likewise.
918 (x86_64-*-netbsd*) Likewise.
919 (mips*-*-netbsd*) Likewise.
920 (powerpc-*-netbsd*) Likewise.
921 (sh*-*-netbsd*) Likewise.
922 (sparc-*-netbsdelf*) Likewise.
923 (sparc64-*-netbsd*) Likewise.
924 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
926 (vax-*-netbsdelf*) Likewise.
927 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
928 (UINT_FAST8_TYPE) Likewise.
929 (INT_FAST16_TYPE) Check CHAR_FAST16.
930 (UINT_FAST16_TYPE) Likewise.
932 2017-09-30 Jakub Jelinek <jakub@redhat.com>
935 * config/i386/i386.md
936 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
937 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
938 *divmodsi4_zext_2): New define_insn_and_split.
939 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
940 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
941 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
942 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
943 New define_insn_and_split.
944 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
945 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
946 operands[1] having DImode when mode is SImode.
948 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
949 always SImode for DIV and MOD in REG_EQUAL notes.
951 2017-09-29 Yury Gribov <tetra2005@gmail.com>
954 * match.pd: Fix handling of NaNs in pattern.
956 2017-09-29 Jeff Law <law@redhat.com>
958 * sbitmap.c (bitmap_bit_in_range_p): New function.
959 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
960 * tree-ssa-dse.c (live_bytes_read): New function.
961 (dse_classify_store): Ignore reads of dead bytes.
963 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
964 typos and whitespace errors.
965 * config/i386/predicates.md (address_no_seg_operand): Likewise.
966 * config/s390/s390.c (s390_emit_prologue): Likewise.
968 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
971 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
972 with a symbol for LRA.
974 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
976 PR rtl-optimization/82338
977 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
979 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
981 * genmodes.c (calc_wider_mode): Suppress qsort macro.
982 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
983 (qsort_chk): Declare.
984 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
985 (qsort_chk): New function.
987 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
989 PR tree-optimization/82337
990 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
991 phi definition if the PHI result appears in an abnormal PHI.
992 (find_basis_for_base_expr): Don't record a basis if the LHS of the
993 basis appears in an abnormal PHI.
995 2017-09-29 Richard Biener <rguenther@suse.de>
997 * graphite-isl-ast-to-gimple.c
998 (translate_isl_ast_to_gimple::set_codegen_error): New function.
999 (binary_op_to_tree): Use it.
1000 (get_rename_from_scev): Likewise.
1001 (copy_loop_phi_nodes): Likewise.
1002 (copy_bb_and_scalar_dependences): Likewise.
1003 (translate_pending_phi_nodes): Likewise.
1005 2017-09-29 Jakub Jelinek <jakub@redhat.com>
1008 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
1009 for movabsq $(i32 << shift), r64.
1011 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
1013 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
1014 index when encoding %esp as %rsp to avoid 0x67 prefix.
1016 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
1018 * config/i386/i386.md (*movsf_internal, *movdf_internal):
1019 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
1021 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
1023 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
1024 Extensions with more than 16 double VFP registers.
1025 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
1026 to_clear_mask and all code related to it. Replace the remaining
1027 entry by a sbitmap and adapt code accordingly.
1029 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
1031 * brig-builtins.def: Change pure attributes to const.
1033 2017-09-28 Joseph Myers <joseph@codesourcery.com>
1035 * config.gcc (default_gnu_indirect_function): Default to yes for
1036 sparc*-*-linux* with glibc.
1038 2017-09-28 Joseph Myers <joseph@codesourcery.com>
1040 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
1041 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
1042 when creating .init_array and .fini_array sections with priority
1045 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
1048 * config/aarch64/aarch64.c
1049 (aarch64_builtin_support_vector_misalignment): Always return false
1050 when misalignment is unknown.
1052 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
1054 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
1055 this function to return false if the definition used by the swap
1056 instruction is artificial, or if the memory address from which the
1057 constant value is loaded is not represented by a base address held
1058 in a register or if the base address register is a frame or stack
1059 pointer. Additionally, return false if the base address of the
1060 loaded constant is a SYMBOL_REF but is not considered to be a
1062 (replace_swapped_load_constant): New function.
1063 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
1064 loaded constant vector with a load of a swapped constant vector.
1066 2017-09-27 Carl Love <cel@us.ibm.com>
1068 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
1069 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
1070 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
1073 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
1075 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
1076 first, always call autopref_rank_data otherwise.
1078 2017-09-27 Richard Biener <rguenther@suse.de>
1080 * graphite-scop-detection.c (find_scop_parameters): Move
1081 loop bound handling ...
1082 (gather_bbs::before_dom_children): ... here, avoiding the need
1083 to build scop_info->loop_nest.
1084 (record_loop_in_sese): Remove.
1085 * sese.h (sese_info_t::loop_nest): Remove.
1086 * sese.c (new_sese_info): Do not allocate loop_nest.
1087 (free_sese_info): Do not free loop_nest.
1089 2017-09-27 Jakub Jelinek <jakub@redhat.com>
1092 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
1093 lhs from calls if the lhs has addressable type.
1095 2017-09-27 Richard Biener <rguenther@suse.de>
1097 * graphite.h (scop::max_alias_set): New member.
1098 * graphite-scop-detection.c: Remove references to non-existing
1099 --param in comments.
1100 (build_alias_sets): Record the maximum alias set used for drs.
1101 (build_scops): Support zero as unlimited for
1102 --param graphite-max-arrays-per-scop.
1103 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
1105 (build_poly_sr_1): ... here. Compute alias set based on the
1106 maximum alias set used for drs rather than
1107 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
1109 2017-09-27 Richard Biener <rguenther@suse.de>
1111 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
1112 --param loop-block-tile-size=0 to disable tiling.
1114 2017-09-27 Richard Biener <rguenther@suse.de>
1116 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
1117 (graphite-max-nb-scop-params): Document special value zero.
1118 * domwalk.h (dom_walker::STOP): New symbolical constant.
1119 (dom_walker::dom_walker): Add optional parameter for bb to
1121 (dom_walker::~dom_walker): Declare.
1122 (dom_walker::before_dom_children): Document STOP return value.
1123 (dom_walker::m_user_bb_to_rpo): New member.
1124 (dom_walker::m_bb_to_rpo): Likewise.
1125 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
1126 mapping here if not provided by the user.
1127 (dom_walker::~dom_walker): Free bb to RPO mapping if not
1128 provided by the user.
1129 (dom_walker::STOP): Define.
1130 (dom_walker::walk): Do not compute bb to RPO mapping here.
1131 Support STOP return value from before_dom_children to stop
1133 * graphite-optimize-isl.c (optimize_isl): If the schedule
1134 is the same still generate code if -fgraphite-identity
1135 or -floop-parallelize-all are given.
1136 * graphite-scop-detection.c: Include cfganal.h.
1137 (gather_bbs::gather_bbs): Get and pass through bb to RPO
1139 (gather_bbs::before_dom_children): Return STOP for BBs
1141 (build_scops): Compute bb to RPO mapping and pass it to
1142 the domwalk. Treat --param graphite-max-nb-scop-params=0
1143 as not limiting the number of params.
1144 * graphite.c (graphite_initialize): Remove limit on the
1145 number of basic-blocks in a function.
1146 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
1147 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
1148 default value of 10.
1150 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1152 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
1153 Adjust code to eliminate needing to do the shift right 32-bits
1154 operation after XSCVDPSPN.
1156 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1158 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
1159 ((X / Y) != 0 -> X >= Y): Likewise.
1161 2017-09-26 Carl Love <cel@us.ibm.com>
1163 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
1164 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
1165 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
1166 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
1167 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
1168 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
1169 definitions and overloading.
1170 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
1171 statement for P9V_BUILTIN_XST_LEN_R.
1172 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
1173 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
1174 define_expand and define_insn for the instructions and builtins.
1175 * doc/extend.texi: Update the built-in documentation file for the new
1177 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
1178 define_insn for the instructions
1180 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
1183 * gcc/config/netbsd-protos.h: New file.
1184 * gcc/config/netbsd.c: New file.
1185 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
1186 * gcc/config/t-netbsd: New file.
1187 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
1188 (tmake_file) Add t-netbsd.
1189 (extra_objs) Add netbsd.o.
1191 2017-09-26 Janus Weil <janus@gcc.gnu.org>
1195 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
1197 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1199 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
1200 sign extension from a vector register to a GPR by doing a 32-bit
1201 direct move and then an EXTSW.
1202 (extendsi<mode>2 splitter): Likewise.
1203 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
1204 right or vector extract after doing XSCVDPSPN. Use
1205 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
1207 (movdi_from_sf_zero_ext): Likewise.
1208 (reload_gpr_from_vsxsf): Likewise.
1209 (p8_mfvsrd_4_disf): Delete, no longer used.
1210 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
1211 then needing to move the SFmode to a GPR to use the XSCVDPSP
1212 instruction instead of FRSP and XSCVDPSPN.
1213 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
1214 it is adjacent to the other XSCVSPDP insns.
1215 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
1216 SFmode to be in traditional Altivec registers.
1217 (vsx_xscvdpspn): Eliminate useless alternative constraint.
1218 (vsx_xscvspdpn): Likewise.
1219 (vsx_xscvspdpn_scalar): Likewise.
1221 2017-09-26 Martin Jambor <mjambor@suse.cz>
1223 * tree-sra.c (compare_access_positions): Put integral types first,
1224 stabilize sorting of integral types, remove conditions putting
1225 non-full-precision integers last.
1226 (sort_and_splice_var_accesses): Disable scalarization if a
1227 non-integert would be represented by a non-full-precision integer.
1229 2017-09-26 Joseph Myers <joseph@codesourcery.com>
1231 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
1232 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
1233 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
1234 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
1235 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
1236 conditionals inside the function instead of around it. Call
1237 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
1238 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
1240 2017-09-26 Richard Biener <rguenther@suse.de>
1242 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
1244 (scop_detection::build_scop_breadth): ... this. Removed.
1245 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
1246 (scop_detection::harmful_stmt_in_bb): Likewise.
1247 (scop_detection::graphite_can_represent_stmt): Likewise.
1248 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
1249 (scop_detection::can_represent_loop): Remove recursion, fold in ...
1250 (scop_detection::can_represent_loop_1): ... this. Removed.
1251 (scop_detection::harmful_loop_in_region): Simplify after inlining
1252 the above and remove more quadraticness.
1253 (build_scops): Adjust.
1254 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
1257 2017-09-26 Jakub Jelinek <jakub@redhat.com>
1260 * config/i386/i386.c (ix86_print_operand_address_as): Only test
1261 REGNO (base) == SP_REG if base is a REG.
1264 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
1265 if it is different SSA_NAME.
1266 (optimize_range_tests_cmp_bitwise): New function.
1267 (optimize_range_tests): Call it.
1269 2017-09-26 Richard Biener <rguenther@suse.de>
1271 PR tree-optimization/82321
1272 * graphite.c (canonicalize_loop_closed_ssa): Properly check
1273 for the def being inside the loop.
1275 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1277 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
1279 * config/s390/s390-builtins.def: Fix constraint on op4.
1281 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1283 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
1284 independent expanders.
1285 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
1286 ("vec_ordered", "vec_unordered"): New expanders.
1288 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1290 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
1293 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1295 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
1296 vec_unpacks_lo_v16qi.
1297 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
1299 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1301 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
1302 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
1303 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
1305 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1307 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
1309 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
1311 ("*vec_slb<mode>"): New insn pattern.
1312 ("vec_shr_<mode>"): New expander.
1313 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
1314 and force the shift count operand to V16QImode.
1315 ("vec_srb<mode>"): Set shift count mode to V16QI.
1317 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1319 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
1320 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
1321 ("vec_widen_smult_hi_<mode>"): New expander definitions.
1323 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
1326 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
1328 2017-09-26 Richard Biener <rguenther@suse.de>
1330 PR tree-optimization/82320
1331 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
1334 2017-09-25 Jeff Law <law@redhat.com>
1336 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
1337 prototype for new argument.
1338 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
1339 mostly extracted from rs6000_emit_allocate_stack.
1340 (rs6000_emit_probe_stack_range_stack_clash): New function.
1341 (rs6000_emit_allocate_stack): Call
1342 rs6000_emit_probe_stack_range_stack_clash as needed.
1343 (rs6000_emit_probe_stack_range): Add additional argument
1344 to call to gen_probe_stack_range{si,di}.
1345 (output_probe_stack_range): New.
1346 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
1347 (output_probe_stack_range_stack_clash): New.
1348 (rs6000_emit_prologue): Emit notes into dump file as requested.
1349 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
1350 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
1351 Add additional operand and pass it to output_probe_stack_range.
1353 2017-09-25 Bin Cheng <bin.cheng@arm.com>
1355 PR tree-optimization/82163
1356 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
1357 (checking_verify_loop_closed_ssa): New parameter.
1358 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
1359 (check_loop_closed_ssa_stmt): Delete.
1360 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
1361 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
1362 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
1365 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
1367 * brig-builtins.def: Treat HSAIL barrier builtins as
1368 setjmp/longjump style functions.
1370 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1372 * target.def (constant_alignment): New hook.
1373 * defaults.h (CONSTANT_ALIGNMENT): Delete.
1374 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
1375 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
1376 * doc/tm.texi: Regenerate.
1377 * targhooks.h (default_constant_alignment): Declare.
1378 (constant_alignment_word_strings): Likewise.
1379 * targhooks.c (default_constant_alignment): New function.
1380 (constant_alignment_word_strings): Likewise.
1381 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
1382 instead of CONSTANT_ALIGNMENT.
1383 * varasm.c (align_variable, get_variable_align, build_constant_desc)
1384 (force_const_mem): Likewise.
1385 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
1386 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
1387 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
1388 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1389 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
1391 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
1392 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1393 constant_alignment_word_strings.
1394 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
1395 (CONSTANT_ALIGNMENT): Likewise.
1396 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1397 (arm_constant_alignment): New function.
1398 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
1399 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1400 constant_alignment_word_strings.
1401 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
1402 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1403 constant_alignment_word_strings.
1404 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
1405 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1406 (cris_constant_alignment): New function.
1407 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
1408 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1409 (epiphany_constant_alignment): New function.
1410 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
1411 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1412 constant_alignment_word_strings.
1413 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
1414 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1415 constant_alignment_word_strings.
1416 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
1417 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1418 constant_alignment_word_strings.
1419 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
1420 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
1421 * config/i386/i386.c (ix86_constant_alignment): Make static.
1422 Use the same interface as the target hook.
1423 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1424 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
1425 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1426 constant_alignment_word_strings.
1427 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
1428 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
1429 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1430 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
1431 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1432 constant_alignment_word_strings.
1433 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
1434 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1435 constant_alignment_word_strings.
1436 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
1437 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1438 constant_alignment_word_strings.
1439 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
1440 * config/microblaze/microblaze.c (microblaze_constant_alignment):
1442 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1443 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
1444 * config/mips/mips.c (mips_constant_alignment): New function.
1445 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1446 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
1447 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
1448 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1449 (mmix_constant_alignment): Make static. Use the same interface
1451 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
1452 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1453 constant_alignment_word_strings.
1454 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
1455 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1456 constant_alignment_word_strings.
1457 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
1458 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1459 constant_alignment_word_strings.
1460 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
1461 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1462 (rs6000_constant_alignment): New function.
1463 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
1464 * config/riscv/riscv.c (riscv_constant_alignment): New function.
1465 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1466 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
1467 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1468 (rs6000_constant_alignment): New function.
1469 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
1470 * config/s390/s390.c (s390_constant_alignment): New function.
1471 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1472 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
1473 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1474 constant_alignment_word_strings.
1475 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
1476 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1477 (sparc_constant_alignment): New function.
1478 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
1479 * config/spu/spu.c (spu_constant_alignment): New function.
1480 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1481 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
1482 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1483 constant_alignment_word_strings.
1484 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
1485 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1486 constant_alignment_word_strings.
1487 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
1488 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1489 constant_alignment_word_strings.
1490 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
1491 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1492 (visium_constant_alignment): New function.
1493 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
1494 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1495 (xtensa_constant_alignment): New function.
1496 * system.h (CONSTANT_ALIGNMENT): Poison.
1498 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
1500 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1501 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
1502 (rs6000_builtin_valid_without_lhs): New helper function.
1503 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1504 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
1506 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1508 * target.h (vec_perm_indices): Use unsigned short rather than
1510 (auto_vec_perm_indices): Likewise.
1511 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
1512 Use unsigned int rather than unsigned char.
1513 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
1515 2017-09-25 Richard Biener <rguenther@suse.de>
1517 * cfgloop.h (sort_sibling_loops): Declare.
1518 * cfgloop.c (sort_sibling_loops_cmp): New helper.
1519 (sort_sibling_loops): New function sorting the sibling loop list
1521 * graphite.c (graphite_transform_loops): Sort sibling loops.
1523 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
1525 * target.def (vec_perm_const_ok): Change sel parameter to
1527 * optabs-query.c (can_vec_perm_p): Update accordingly.
1528 * doc/tm.texi: Regenerate.
1529 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
1530 auto_vec_perm_indices and remove separate nelt field.
1531 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
1532 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
1533 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
1534 (aarch64_expand_vec_perm_const): Update accordingly.
1535 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
1536 to vec_perm_indices.
1537 * config/arm/arm.c (expand_vec_perm_d): Change perm to
1538 auto_vec_perm_indices and remove separate nelt field.
1539 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
1540 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
1541 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
1543 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
1544 to vec_perm_indices.
1545 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
1546 sel to vec_perm_indices.
1547 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
1548 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
1549 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
1551 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
1554 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
1557 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
1558 on the FUNCTION_DECL function context if it has a DIE that is a
1561 2017-09-25 Richard Biener <rguenther@suse.de>
1563 PR tree-optimization/82285
1564 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
1567 2017-09-25 Tom de Vries <tom@codesourcery.com>
1571 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
1574 2017-09-25 Richard Biener <rguenther@suse.de>
1576 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
1577 ISL errors other than isl_error_quota happen. Dump if the
1578 schedule is the same.
1579 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
1580 errors instead of aborting inside ISL.
1582 2017-09-25 Iain Sandoe <iain@codesourcery.com>
1585 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
1586 of libgcc_eh for m64.
1587 * config/i386/darwin64.h: Likewise.
1589 2017-09-25 Richard Biener <rguenther@suse.de>
1592 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
1593 attribute for incomplete types nor twice for complete ones.
1595 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
1598 * config/i386/i386.c (ix86_print_operand_address_as): Encode
1599 %esp as %rsp to avoid 0x67 prefix if there is no index or base
1602 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
1605 * config/i386/i386.opt (mprefer-avx256): Use
1606 ix86_target_flags variable.
1607 * config/i386/i386.c (ix86_target_string): Move
1608 -mprefer-avx256 to flag2_opts.
1610 2017-09-22 Jakub Jelinek <jakub@redhat.com>
1613 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
1614 and x != -1 | y != -1 into (x & y) != -1.
1616 2017-09-22 Steve Ellcey <sellcey@cavium.com>
1618 * config.gcc: Add new case statement to set
1619 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
1620 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
1621 s390x-*-linux* case statements. Added aarch64 to the list of
1622 supported architectures.
1624 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1626 PR tree-optimization/82289
1627 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
1628 STMT_VINFO_RELEVANT_P.
1630 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1631 Alan Hayward <alan.hayward@arm.com>
1632 David Sherwood <david.sherwood@arm.com>
1634 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
1635 for VR_RANGE only; don't allow VR_ANTI_RANGE.
1636 (extract_range_from_binary_expr_1): Don't call
1637 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
1639 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1640 Alan Hayward <alan.hayward@arm.com>
1641 David Sherwood <david.sherwood@arm.com>
1643 * target.def (preferred_vector_alignment): New hook.
1644 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
1646 * doc/tm.texi: Regenerate.
1647 * targhooks.h (default_preferred_vector_alignment): Declare.
1648 * targhooks.c (default_preferred_vector_alignment): New function.
1649 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
1651 (DR_TARGET_ALIGNMENT): New macro.
1652 (aligned_access_p): Update commentary.
1653 (vect_known_alignment_in_bytes): New function.
1654 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
1656 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
1657 Calculate the misalignment based on the target alignment rather than
1659 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
1660 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
1661 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
1662 the target alignment, rather than masking the element misalignment
1663 with the number of elements in a vector. Also use the target
1664 alignment when calculating the maximum number of peels.
1665 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
1666 instead of TYPE_ALIGN_UNIT.
1667 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
1668 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
1669 (vect_create_addr_base_for_vector_ref): Update call accordingly.
1670 (vect_create_data_ref_ptr): Likewise.
1671 (vect_setup_realignment): Realign by ANDing with
1672 -DR_TARGET_MISALIGNMENT.
1673 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
1674 the number of peels based on DR_TARGET_ALIGNMENT.
1675 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
1676 with the guaranteed alignment boundary when deciding whether
1678 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
1679 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
1680 (ensure_base_align): Remove stmt_info parameter. Get the
1681 target base alignment from DR_TARGET_ALIGNMENT.
1682 (vectorizable_store): Update call accordingly. Interpret
1683 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
1685 (vectorizable_load): Likewise.
1687 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1688 Alan Hayward <alan.hayward@arm.com>
1689 David Sherwood <david.sherwood@arm.com>
1691 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
1692 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
1693 (vect_enhance_data_refs_alignment): Likewise.
1695 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1697 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1698 error. Only quit immediately if parsing is complete.
1699 (BEGIN): Initialize fatal_err and parse_done.
1700 (begin fpu, end fpu): Check number of arguments.
1701 (begin arch, end arch): Likewise.
1702 (begin cpu, end cpu): Likewise.
1703 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
1704 (optalias): Likewise.
1706 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1708 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
1709 * config/arm/arm-isa.h: Delete. Move definitions to ...
1710 * arm-cpus.in: ... here. Use new feature and fgroup values.
1711 * config/arm/arm.c (arm_option_override): Use lower case for feature
1713 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
1714 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
1715 * config/arm/parsecpu.awk (END): Add new command 'isa'.
1717 (print_isa_bits_for): New function.
1718 (gen_isa): New function.
1719 (gen_comm_data): Use print_isa_bits_for.
1720 (define feature): New keyword.
1721 (define fgroup): New keyword.
1722 * config/arm/t-arm (TM_H): Remove.
1723 (GTM_H): Add arm-isa.h.
1724 (arm-isa.h): Add rule to generate file.
1725 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
1726 case for feature bit names.
1728 2017-09-22 Richard Biener <rguenther@suse.de>
1730 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
1732 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
1733 print of no dependency loops ...
1734 * graphite.c (graphite_transform_loops): ... here.
1735 (canonicalize_loop_closed_ssa_form): Work from inner to outer
1737 (same_close_phi_node, remove_duplicate_close_phi,
1738 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
1739 (canonicalize_loop_closed_ssa): ... here and simplify.
1740 * graphite-optimize-isl.c: Include tree-vectorizer.h.
1741 (optimize_isl): Use dump_printf_loc to tell when we stopped
1742 optimizing because of an ISL timeout.
1744 2017-09-22 Richard Biener <rguenther@suse.de>
1746 PR tree-optimization/82291
1747 * tree-if-conv.c (predicate_mem_writes): Make sure to
1748 remove writes in blocks predicated with false.
1750 2017-09-22 Richard Biener <rguenther@suse.de>
1752 * sese.c: Include cfganal.h.
1753 (if_region_set_false_region): Remove.
1754 (create_if_region_on_edge): Likewise.
1755 (move_sese_in_condition): Re-implement without destroying
1758 2017-09-22 Richard Biener <rguenther@suse.de>
1760 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
1761 Verify both BBs contain loop PHI nodes before dispatching to
1763 (graphite_regenerate_ast_isl): Do not recompute dominators,
1764 do not verify three times. Restructure for clarity.
1765 * graphite-scop-detection.c (same_close_phi_node,
1766 remove_duplicate_close_phi, make_close_phi_nodes_unique,
1767 defined_in_loop_p, canonicalize_loop_closed_ssa,
1768 canonicalize_loop_closed_ssa_form): Simplify, remove excess
1769 checking and SSA rewrite, move to ...
1770 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
1771 (graphite_initialize): Do not pass in ctx, do not reset the
1772 SCEV cache, compute only dominators.
1773 (graphite_transform_loops): Allocate ISL ctx after
1774 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
1775 Maintain post-dominators only around build_scops.
1776 * sese.c (if_region_set_false_region): Make static. Free
1777 and recompute dominators.
1778 (move_sese_in_condition): Assert we don't get called with
1779 post-dominators computed.
1780 * sese.h (if_region_set_false_region): Remove.
1782 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
1784 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
1785 mode attribute for TARGET_AVX512VL.
1787 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
1789 * config/i386/i386.opt (mprefer-avx256): New option.
1790 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
1792 (ix86_preferred_simd_mode): Return 256-bit AVX modes
1793 for TARGET_PREFER_AVX256.
1794 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
1796 2017-09-21 Jeff Law <law@redhat.com>
1798 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
1799 Fix dump output if the only stack space is for pushed registers.
1801 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1803 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
1806 2017-09-21 Martin Sebor <msebor@redhat.com>
1809 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
1810 code (in C++) or code that triggers warnings.
1812 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
1814 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
1816 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
1818 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
1819 * sched-rgn.c: Ditto.
1820 * sel-sched-ir.c: Ditto.
1822 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
1824 * toplev.h (set_random_seed): Adjust return type.
1825 * toplev.c (init_local_tick): Move eager initialization of random_seed
1826 to get_random_seed. Adjust comment.
1827 (init_random_seed): Inline to get_random_seed, delete.
1828 (get_random_seed): Initialize random_seed lazily.
1829 (set_random_seed): Do not return previous value.
1830 (print_switch_value): Do not call get_random_seed.
1832 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
1834 * cgraph.c (delete_function_version): New, broken out from...
1835 (cgraph_node::delete_function_version): ...here. Rename to
1836 cgraph_node::delete_function_version_by_decl. Update all uses.
1837 (cgraph_node::remove): Call delete_function_version.
1839 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1842 * tree-inline.c (expand_call_inline): Emit clobber stmts for
1843 VAR_DECLs to which addressable non-volatile parameters are mapped
1844 and for id->retvar after the return value assignment. Clear
1845 id->retval and id->retbnd after inlining.
1847 2017-09-21 Richard Biener <rguenther@suse.de>
1849 PR tree-optimization/82276
1850 PR tree-optimization/82244
1851 * tree-vrp.c (build_assert_expr_for): Set
1852 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
1854 (remove_range_assertions): Revert earlier change.
1856 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
1859 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
1861 2017-09-21 Richard Biener <rguenther@suse.de>
1863 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
1864 Restore valid IL after code generation errors.
1865 * graphite.c (graphite_transform_loops): Diagnose code
1866 generation issues as MSG_MISSED_OPTIMIZATION and continue
1867 with processing SCOPs.
1869 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1870 Alan Hayward <alan.hayward@arm.com>
1871 David Sherwood <david.sherwood@arm.com>
1873 * calls.c (compute_argument_addresses): Use simplify_gen_binary
1874 rather than choosing between plus_constant and gen_rtx_<CODE>.
1875 * expr.c (emit_push_insn): Likewise.
1876 (expand_expr_real_2): Likewise.
1878 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1879 Alan Hayward <alan.hayward@arm.com>
1880 David Sherwood <david.sherwood@arm.com>
1882 * loop-unroll.c (split_iv): Call copy_rtx on the step.
1884 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1885 Alan Hayward <alan.hayward@arm.com>
1886 David Sherwood <david.sherwood@arm.com>
1888 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
1889 calling tree_to_uhwi.
1891 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1892 Alan Hayward <alan.hayward@arm.com>
1893 David Sherwood <david.sherwood@arm.com>
1895 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
1896 INTEGER_CST rather than a negative test for ADDR_EXPR.
1898 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
1899 Alan Hayward <alan.hayward@arm.com>
1900 David Sherwood <david.sherwood@arm.com>
1902 * tree-vrp.c (extract_range_from_binary_expr_1): Check
1903 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
1905 2017-09-21 Richard Biener <rguenther@suse.de>
1907 PR tree-optimization/71351
1908 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
1909 graphite_create_new_loop_guard): Remove, fold remaining parts
1911 (translate_isl_ast_node_for): ... here and simplify.
1913 2017-09-21 Jakub Jelinek <jakub@redhat.com>
1916 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
1917 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
1918 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
1919 alternative always use QI mode, for -Os imov (=R,R) alternative
1920 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
1923 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1924 Jeff Law <law@redhat.com>
1926 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
1927 (allocate_stack_space): New function, partially extracted from
1929 (s390_emit_prologue): Track offset to most recent stack probe.
1930 Code to allocate space moved into allocate_stack_space.
1931 Dump actions when no stack is allocated.
1932 (s390_prologue_plus_offset): New function.
1933 (s390_emit_stack_probe): Likewise.
1935 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
1937 * common.opt (Wa, Wl, Wp, g, gz=): Add
1939 (gno-column-info): Remove.
1940 (gcolumn-info): Drop RejectNegative.
1942 (gno-record-gcc-switches): Remove.
1943 (grecord-gcc-switches): Drop RejectNegative.
1944 (gno-split-dwarf): Remove.
1945 (gsplit-dwarf): Drop RejectNegative.
1946 (gno-strict-dwarf): Remove.
1947 (gstrict-dwarf): Drop RejectNegative.
1948 * config/darwin.opt (gfull, gused): Add RejectNegative.
1949 * dwarf2out.c (gen_producer_string): Drop
1950 gno-record-gcc-switches handler.
1951 * optc-gen.awk: Add g to prefixes with negative forms.
1952 * opts-common.c (remapping_prefix_p): New.
1953 (find_opt): Check it.
1954 (generate_canonical_option): Test g prefix.
1955 (option_map): Add -gno- mapping.
1956 (add_misspelling_candidates): Check remapping_prefix_p.
1958 2017-09-20 Jeff Law <law@redhat.com>
1960 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
1961 thinko in stack clash protection support.
1963 * explow.c (compute_stack_clash_protection_loop_data): Use
1964 CONST_INT_P instead of explicit test. Verify object is a
1965 CONST_INT_P before looking at INTVAL.
1966 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
1967 instead of explicit test.
1969 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
1972 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
1973 address instead of to r1 and r11.
1975 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
1977 * config.gcc: Support "knm".
1978 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
1979 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
1981 * config/i386/i386.c (m_KNM): Define.
1982 (processor_target_table): Add "knm".
1984 (ix86_option_override_internal): Add "knm".
1985 (ix86_issue_rate): Add PROCESSOR_KNM.
1986 (ix86_adjust_cost): Ditto.
1987 (ia32_multipass_dfa_lookahead): Ditto.
1988 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
1989 (fold_builtin_cpu): Add M_INTEL_KNM.
1990 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
1991 (processor_type): Add PROCESSOR_KNM.
1992 * config/i386/x86-tune.def: Add m_KNM.
1993 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
1995 2017-09-20 Richard Biener <rguenther@suse.de>
1997 PR tree-optimization/80213
1998 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
1999 are allowed in empty BBs as well.
2000 (canonicalize_loop_closed_ssa): Also look for other complex
2002 (scop_detection::get_sese): Include the loop-closed PHI block
2004 (scop_detection::merge_sese): Remove code adding extra blocks.
2005 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
2006 (build_scops): Assert the final returned scop is invalid.
2008 2017-09-20 Richard Biener <rguenther@suse.de>
2010 PR tree-optimization/82264
2011 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
2013 (vn_phi_lookup): Likewise.
2014 (vn_phi_insert): Likewise.
2016 2017-09-20 Jakub Jelinek <jakub@redhat.com>
2018 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
2019 that fits into uhwi or shwi, add DW_AT_const_value regardless
2020 of early_dwarf without going through RTL, using add_AT_unsigned
2023 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
2024 (DEBUG_LTO_ABBREV_SECTION): Likewise.
2025 (DEBUG_LTO_MACINFO_SECTION): Likewise.
2026 (DEBUG_MACRO_SECTION): Likewise.
2027 (DEBUG_LTO_MACRO_SECTION): Likewise.
2028 (DEBUG_STR_DWO_SECTION): Likewise.
2029 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
2030 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
2031 (DEBUG_LTO_DWO_LINE_SECTION): Define.
2032 (DEBUG_LTO_LINE_STR_SECTION): Define.
2033 (init_sections_and_labels): Initialize debug_line_str_section
2034 variable. Initialize debug_loc_section for -gdwarf-5 to
2035 DEBUG_LOCLISTS_SECTION. Formatting fixes.
2037 2017-09-20 Richard Biener <rguenther@suse.de>
2039 * graphite-sese-to-poly.c (extract_affine): Properly handle
2040 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
2042 2017-09-20 Richard Biener <rguenther@suse.de>
2044 PR tree-optimization/81373
2045 * graphite-scop-detection.c (build_cross_bb_scalars_def):
2046 Force SESE live-out defs to be handled even if they are
2049 2017-09-19 Jeff Law <law@redhat.com>
2051 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
2052 nothing for stack adjustments with REG_STACK_CHECK.
2053 * sched-deps.c (parse_add_or_inc): Reject insns with
2054 REG_STACK_CHECK from dependency breaking.
2055 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
2056 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
2057 * reg-notes.def (STACK_CHECK): New note.
2059 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
2060 (ix86_expand_prologue): Dump stack clash info as needed.
2061 Call ix86_adjust_stack_and_probe_stack_clash as needed.
2063 * function.c (dump_stack_clash_frame_info): New function.
2064 * function.h (dump_stack_clash_frame_info): Prototype.
2065 (enum stack_clash_probes): New enum.
2067 * config/alpha/alpha.c (alpha_expand_prologue): Also check
2068 flag_stack_clash_protection.
2069 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
2070 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
2071 (arm_frame_pointer_required): Likewise.
2072 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
2073 (ia64_expand_prologue): Likewise.
2074 * config/mips/mips.c (mips_expand_prologue): Likewise.
2075 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
2076 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
2077 (sparc_flat_expand_prologue): Likewise.
2078 * config/spu/spu.c (spu_expand_prologue): Likewise.
2080 * explow.c: Include "params.h".
2081 (anti_adjust_stack_and_probe_stack_clash): New function.
2082 (get_stack_check_protect): Likewise.
2083 (compute_stack_clash_protection_loop_data): Likewise.
2084 (emit_stack_clash_protection_loop_start): Likewise.
2085 (emit_stack_clash_protection_loop_end): Likewise.
2086 (allocate_dynamic_stack_space): Use get_stack_check_protect.
2087 Use anti_adjust_stack_and_probe_stack_clash.
2088 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
2089 (emit_stack_clash_protection_loop_start): Likewise.
2090 (emit_stack_clash_protection_loop_end): Likewise.
2091 * rtl.h (get_stack_check_protect): Prototype.
2092 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
2093 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
2094 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
2096 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
2098 * doc/tm.texi: Rebuilt.
2099 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
2100 get_stack_check_protect.
2101 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
2102 * config/arm/arm.c (arm_expand_prologue): Likewise.
2103 (arm_frame_pointer_required): Likewise.
2104 * config/i386/i386.c (ix86_expand_prologue): Likewise.
2105 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
2106 * config/mips/mips.c (mips_expand_prologue): Likewise.
2107 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
2108 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
2109 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
2110 (sparc_flat_expand_prologue): Likewise.
2112 * common.opt (-fstack-clash-protection): New option.
2113 * flag-types.h (enum stack_check_type): Note difference between
2114 -fstack-check= and -fstack-clash-protection.
2115 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
2116 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
2117 * toplev.c (process_options): Issue warnings/errors for cases
2118 not handled with -fstack-clash-protection.
2119 * doc/invoke.texi (-fstack-clash-protection): Document new option.
2120 (-fstack-check): Note additional problem with -fstack-check=generic.
2121 Note that -fstack-check is primarily for Ada and refer users
2122 to -fstack-clash-protection for stack-clash-protection.
2123 Document new params for stack clash protection.
2125 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
2127 * config/i386/i386.c (ix86_split_long_move): Do not handle
2128 address used for LEA in a special way.
2130 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
2132 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
2134 2017-09-19 Martin Sebor <msebor@redhat.com>
2137 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
2138 of incompatible types.
2140 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
2142 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
2143 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
2144 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2145 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
2147 2017-09-19 Richard Biener <rguenther@suse.de>
2149 PR tree-optimization/82244
2150 * tree-vrp.c (remove_range_assertions): Do not propagate
2151 a constant to abnormals but replace the assert with a copy.
2153 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
2155 PR rtl-optimization/57878
2156 PR rtl-optimization/68988
2157 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
2158 avoidance test involving non_reload_pseudos. Move frequency test
2159 below the general fragmentation avoidance test.
2161 2017-09-19 Richard Biener <rguenther@suse.de>
2163 PR tree-optimization/69728
2164 * graphite-sese-to-poly.c (schedule_error): New global.
2165 (add_loop_schedule): Handle empty domain by failing the
2167 (build_original_schedule): Handle schedule_error.
2169 2017-09-19 Richard Biener <rguenther@suse.de>
2171 * graphite-scop-detection.c (scop_detection::can_represent_loop):
2172 Do not iterate to sibling loops but only to siblings of inner
2175 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
2178 * config/m68k/m68k.md (moveq feeding equality comparison): Check
2179 that the registers are different.
2181 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
2183 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
2184 to processor_model and "amdfam17h" to arch_names_table.
2185 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
2187 2017-09-18 Jakub Jelinek <jakub@redhat.com>
2190 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
2192 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2193 Alan Hayward <alan.hayward@arm.com>
2194 David Sherwood <david.sherwood@arm.com>
2196 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
2198 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
2200 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
2201 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
2202 vect_schedule_slp_instance.
2203 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
2204 Update call to vect_slp_analyze_node_operations. Simplify return
2206 (vect_slp_analyze_bb_1): Update call accordingly.
2207 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
2208 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
2209 (vect_schedule_slp): Update call accordingly.
2211 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2212 Alan Hayward <alan.hayward@arm.com>
2213 David Sherwood <david.sherwood@arm.com>
2215 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
2216 with types that aren't in fact scalar.
2218 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2220 * tree-vect-slp.c (vect_record_max_nunits): New function,
2222 (vect_build_slp_tree_1): ...here.
2223 (vect_build_slp_tree_2): Call it for phis too.
2225 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2227 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
2228 to vect_get_vec_def_for_operand when getting the mask operand.
2230 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2231 Alan Hayward <alan.hayward@arm.com>
2232 David Sherwood <david.sherwood@arm.com>
2234 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
2237 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2238 Alan Hayward <alan.hayward@arm.com>
2239 David Sherwood <david.sherwood@arm.com>
2241 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
2242 calculation for vector booleans.
2244 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2245 Alan Hayward <alan.hayward@arm.com>
2246 David Sherwood <david.sherwood@arm.com>
2248 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
2250 (vect_transform_stmt): ...here.
2251 (vect_analyze_stmt): Use it instead of calling
2252 vectorizable_live_operation directly.
2254 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
2256 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
2257 non-SIMT targets in acc vector loops.
2259 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
2261 * configure.ac: Add arc and check if assembler supports gdwarf2.
2262 * configure: Regenerate.
2264 2017-09-18 Richard Biener <rguenther@suse.de>
2266 PR tree-optimization/82220
2267 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
2268 epilogue niters from the min_profitable_iters compute.
2270 2017-09-18 Jakub Jelinek <jakub@redhat.com>
2273 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
2274 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
2275 (ix86_init_pic_reg): Revert 2017-09-01 changes.
2277 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
2280 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
2281 switching to a new text section.
2283 2017-09-18 Richard Biener <rguenther@suse.de>
2285 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
2287 (build_alias_set): Reject aliases with no access function.
2289 2017-09-18 Richard Biener <rguenther@suse.de>
2291 PR tree-optimization/79622
2292 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
2294 (build_cross_bb_scalars_use): Likewise.
2296 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
2298 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
2300 2017-09-18 Alan Modra <amodra@gmail.com>
2303 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
2304 stack_pointer_rtx for count 0. Update comments. Break up
2305 large rtl expression.
2307 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
2309 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
2310 Increase to 20 bytes.
2311 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
2312 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
2313 or avx version of the stub.
2315 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
2318 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
2319 compute the minimum stack alignment. Also update preferred stack
2320 boundary for leaf functions.
2322 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
2324 PR tree-optimization/82228
2325 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
2328 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2330 * common/config/nds32/nds32-common.c
2331 (nds32_option_optimization_table): Refine formatting.
2332 (nds32_option_optimization_table): Use -fsched-pressure and
2333 -fomit-frame-pointer for specific optimization level.
2335 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2337 * config/nds32/nds32.c: Refine formatting and comments.
2338 * config/nds32/nds32.h: Likewise.
2339 * config/nds32/nds32.md: Likewise.
2340 * config/nds32/nds32-cost.c: Likewise.
2341 * config/nds32/nds32-isr.c: Likewise.
2342 * config/nds32/nds32-md-auxiliary.c: Likewise.
2343 * config/nds32/nds32-multiple.md: Likewise.
2344 * config/nds32/nds32-predicates.c: Likewise.
2346 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
2347 Jakub Jelinek <jakub@redhat.com>
2349 Add support for -std=c++2a.
2350 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
2352 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
2354 2017-09-15 Steve Ellcey <sellcey@cavium.com>
2357 * doc/extend.texi (Common Function Attributes): Add
2358 references to ARM, AArch64, and S/390 specific attributes.
2359 (Function Specific Option Pragmas): Add AArch64 and S/390
2360 to list of back ends that support the target pragma.
2362 2017-09-15 Nathan Sidwell <nathan@acm.org>
2364 * doc/standards.texi: Fix C++17 description. Update URLs for
2367 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
2369 * common.opt (Wcast-align=strict): New warning option.
2370 * doc/invoke.texi: Document -Wcast-align=strict.
2372 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
2374 * cgraph.h (cgraph_thunk_info): Add comments.
2375 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
2376 assert for VIRTUAL_* arguments stricter.
2378 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
2380 PR tree-optimization/71026
2381 * match.pd: Move RDIV patterns from fold-const.c
2382 * fold-const.c (distribute_real_division): Removed.
2383 (fold_binary_loc): Remove calls to distribute_real_divison.
2385 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2387 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
2388 c++1z and gnu++1z as deprecated. Change other references to
2389 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
2390 Change -Wc++1z-compat to -Wc++17-compat.
2391 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
2392 * dwarf2out.c (highest_c_language): Handle C++17.
2393 (gen_compile_unit_die): Likewise.
2395 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2397 PR rtl-optimization/82192
2398 * combine.c (make_extraction): Don't look through non-paradoxical
2399 SUBREGs or TRUNCATE if pos + len is or might be bigger than
2402 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2403 Alan Hayward <alan.hayward@arm.com>
2404 David Sherwood <david.sherwood@arm.com>
2406 * target.def (function_arg_offset): New hook.
2407 * targhooks.h (default_function_arg_offset): Declare.
2408 * targhooks.c (default_function_arg_offset): New function.
2409 * function.c (locate_and_pad_parm): Use
2410 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
2411 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
2412 (TARGET_FUNCTION_ARG_OFFSET): ...this.
2413 * doc/tm.texi: Regenerate.
2414 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
2415 * config/spu/spu.c (spu_function_arg_offset): New function.
2416 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
2417 * system.h (FUNCTION_ARG_OFFSET): Poison.
2419 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2420 Alan Hayard <alan.hayward@arm.com>
2421 David Sherwood <david.sherwood@arm.com>
2423 * target.def (truly_noop_truncation): New hook.
2424 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
2425 than TRULY_NOOP_TRUNCATION.
2426 * hooks.h (hook_bool_uint_uint_true): Declare.
2427 * hooks.c (hook_bool_uint_uint_true): New function.
2428 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
2429 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
2430 * doc/tm.texi: Regenerate.
2431 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
2432 rather than TRULY_NOOP_TRUNCATION in comments.
2433 (simplify_comparison): Likewise.
2434 (record_truncated_value): Likewise.
2435 * expmed.c (extract_bit_field_1): Likewise.
2436 (extract_split_bit_field): Likewise.
2437 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
2438 instead of TRULY_NOOP_TRUNCATION.
2439 * function.c (assign_parm_setup_block): Likewise.
2440 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
2441 * rtlhooks.c: Include target.h.
2442 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
2443 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
2444 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
2445 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
2446 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
2447 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
2448 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
2449 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
2450 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
2451 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
2452 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
2453 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
2454 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
2455 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
2456 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
2457 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
2458 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
2459 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
2460 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
2461 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
2462 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
2463 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
2464 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
2465 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
2466 * config/mips/mips.c (mips_truly_noop_truncation): New function.
2467 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2468 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2469 rather than TRULY_NOOP_TRUNCATION in comments.
2470 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
2471 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
2472 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
2473 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
2474 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
2475 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
2476 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
2477 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
2478 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
2479 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
2480 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
2481 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2482 rather than TRULY_NOOP_TRUNCATION in comments.
2483 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
2484 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
2485 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
2486 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
2487 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
2488 TRULY_NOOP_TRUNCATION condition.
2489 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
2490 (TRULY_NOOP_TRUNCATION): Delete.
2491 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
2492 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
2493 * config/spu/spu.c (spu_truly_noop_truncation): New function.
2494 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2495 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
2496 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
2497 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
2498 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2499 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2500 rather than TRULY_NOOP_TRUNCATION in comments.
2501 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
2502 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
2503 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
2504 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
2505 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
2506 * system.h (TRULY_NOOP_TRUNCATION): Poison.
2508 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
2511 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
2512 (*cmp_ior): Likewise.
2513 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
2514 (*ior_scc_scc_cmp): Likewise.
2515 (*and_scc_scc): Likewise.
2516 (*and_scc_scc_cmp): Likewise.
2518 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2519 Alan Hayard <alan.hayward@arm.com>
2520 David Sherwood <david.sherwood@arm.com>
2522 * target.def (can_change_mode_class): New hook.
2523 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
2524 (hard_regno_nregs): Likewise.
2525 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
2526 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
2527 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
2528 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
2529 (LOAD_EXTEND_OP): Update accordingly.
2530 * doc/tm.texi: Regenerate.
2531 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
2532 CANNOT_CHANGE_MODE_CLASS.
2533 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
2534 (REG_CAN_CHANGE_MODE_P): ...this new macro.
2535 * combine.c (simplify_set): Update accordingly.
2536 * emit-rtl.c (validate_subreg): Likewise.
2537 * recog.c (general_operand): Likewise.
2538 * regcprop.c (mode_change_ok): Likewise.
2539 * reload1.c (choose_reload_regs): Likewise.
2540 (inherit_piecemeal_p): Likewise.
2541 * rtlanal.c (simplify_subreg_regno): Likewise.
2542 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
2543 instead of CANNOT_CHANGE_MODE_CLASS.
2544 (reload_cse_simplify_operands): Likewise.
2545 * reload.c (push_reload): Use targetm.can_change_mode_class
2546 instead of CANNOT_CHANGE_MODE_CLASS.
2547 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
2548 REG_CANNOT_CHANGE_MODE_P.
2549 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2550 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
2551 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2552 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2553 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2554 (arm_can_change_mode_class): New function.
2555 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2556 than CANNOT_CHANGE_MODE_CLASS in comments.
2557 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2558 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
2559 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
2560 (ix86_can_change_mode_class): ...this new function, inverting the
2561 sense of the return value.
2562 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2563 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2564 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2565 (ia64_can_change_mode_class): New function.
2566 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2567 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
2568 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
2569 (m32c_can_change_mode_class): ...this new function, inverting the
2570 sense of the return value.
2571 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2572 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2573 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
2574 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
2575 (mips_can_change_mode_class): ...this new function, inverting the
2576 sense of the return value.
2577 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2578 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2579 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2580 (msp430_can_change_mode_class): New function.
2581 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2582 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
2583 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2584 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2585 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2586 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
2587 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2588 (pa_cannot_change_mode_class): Replace with...
2589 (pa_can_change_mode_class): ...this new function, inverting the
2590 sense of the return value.
2591 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2592 than CANNOT_CHANGE_MODE_CLASS in comments.
2593 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2594 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
2595 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2596 (pdp11_cannot_change_mode_class): Replace with...
2597 (pdp11_can_change_mode_class): ...this new function, inverting the
2598 sense of the return value.
2599 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2600 * config/powerpcspe/powerpcspe-protos.h
2601 (rs6000_cannot_change_mode_class_ptr): Delete.
2602 * config/powerpcspe/powerpcspe.c
2603 (rs6000_cannot_change_mode_class_ptr): Delete.
2604 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2605 (rs6000_option_override_internal): Assign to
2606 targetm.can_change_mode_class instead of
2607 rs6000_cannot_change_mode_class_ptr.
2608 (rs6000_cannot_change_mode_class): Replace with...
2609 (rs6000_can_change_mode_class): ...this new function, inverting the
2610 sense of the return value.
2611 (rs6000_debug_cannot_change_mode_class): Replace with...
2612 (rs6000_debug_can_change_mode_class): ...this new function.
2613 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2614 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
2615 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2616 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2617 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
2619 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
2620 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2621 (rs6000_option_override_internal): Assign to
2622 targetm.can_change_mode_class instead of
2623 rs6000_cannot_change_mode_class_ptr.
2624 (rs6000_cannot_change_mode_class): Replace with...
2625 (rs6000_can_change_mode_class): ...this new function, inverting the
2626 sense of the return value.
2627 (rs6000_debug_cannot_change_mode_class): Replace with...
2628 (rs6000_debug_can_change_mode_class): ...this new function.
2629 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2630 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
2631 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
2632 (s390_can_change_mode_class): ...this new function, inverting the
2633 sense of the return value.
2634 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2635 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2636 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
2637 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2638 (sh_cannot_change_mode_class): Replace with...
2639 (sh_can_change_mode_class): ...this new function, inverting the
2640 sense of the return value.
2641 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2642 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2643 (sparc_can_change_mode_class): New function.
2644 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2645 * config/spu/spu.c (spu_can_change_mode_class): New function.
2646 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2647 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2648 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2649 (visium_can_change_mode_class): New function.
2650 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
2652 2017-09-15 Richard Biener <rguenther@suse.de>
2654 PR tree-optimization/82217
2655 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
2656 but not undefined case.
2658 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2661 * postreload.c (reload_cse_simplify_operands): Skip
2662 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
2664 2017-09-15 Richard Biener <rguenther@suse.de>
2666 PR tree-optimization/68823
2667 * graphite-scop-detection.c (build_alias_set): If we have a
2668 possible dependence check whether we can handle them by just
2669 looking at the DRs DR_ACCESS_FNs.
2670 (build_scops): If build_alias_set fails, fail the SCOP.
2672 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2674 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
2675 to support float128 built-in functions that require the ISA 3.0
2677 (BU_FLOAT128_3_HW): Likewise.
2678 (SQRTF128): Add support for the IEEE 128-bit square root and fma
2680 (FMAF128): Likewise.
2682 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
2683 support for built-in functions that need the ISA 3.0 IEEE 128-bit
2684 floating point instructions.
2685 (rs6000_invalid_builtin): Likewise.
2686 (rs6000_builtin_mask_names): Likewise.
2687 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
2688 (RS6000_BTM_FLOAT128_HW): Likewise.
2689 (RS6000_BTM_COMMON): Likewise.
2690 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
2692 * doc/extend.texi (RS/6000 built-in functions): Document the
2693 IEEE 128-bit floating point square root and fused multiply-add
2696 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
2698 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
2699 reg (r2) isn't in the set of registers defined in the prologue.
2701 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2702 Alan Hayward <alan.hayward@arm.com>
2703 David Sherwood <david.sherwood@arm.com>
2705 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
2706 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
2707 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
2708 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
2709 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
2711 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
2712 max_vectorization_factor.
2713 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
2715 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2716 Alan Hayward <alan.hayward@arm.com>
2717 David Sherwood <david.sherwood@arm.com>
2719 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
2720 (vect_worthwhile_without_simd_p): Declare.
2721 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
2722 (vectorizable_reduction): Use it.
2723 * tree-vect-stmts.c (vectorizable_shift): Likewise.
2724 (vectorizable_operation): Likewise.
2726 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2727 Alan Hayward <alan.hayward@arm.com>
2728 David Sherwood <david.sherwood@arm.com>
2730 * tree-vectorizer.h (vect_get_num_copies): New function.
2731 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
2732 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2733 (vectorizable_induction): Likewise.
2734 (vectorizable_live_operation): Likewise.
2735 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
2736 (vectorizable_bswap): Likewise.
2737 (vectorizable_call): Likewise.
2738 (vectorizable_conversion): Likewise.
2739 (vectorizable_assignment): Likewise.
2740 (vectorizable_shift): Likewise.
2741 (vectorizable_operation): Likewise.
2742 (vectorizable_store): Likewise.
2743 (vectorizable_load): Likewise.
2744 (vectorizable_condition): Likewise.
2745 (vectorizable_comparison): Likewise.
2746 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
2748 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2749 Alan Hayward <alan.hayward@arm.com>
2750 David Sherwood <david.sherwood@arm.com>
2752 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
2753 of vect_init_vector.
2755 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2756 Alan Hayward <alan.hayward@arm.com>
2757 David Sherwood <david.sherwood@arm.com>
2759 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
2760 an inline wrapper that provides a location.
2761 (gimple_build_vector): Likewise.
2762 * gimple-fold.c (gimple_build_vector_from_val): New function.
2763 (gimple_build_vector): Likewise.
2764 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
2765 functions to build the initial value. Always return a gimple value.
2766 (get_initial_defs_for_reduction): Likewise. Only compute
2768 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
2769 vect_init_vector on the results from get_initial_def(s)_for_reduction.
2770 (vectorizable_induction): Use gimple_build_vector rather than
2773 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2774 Alan Hayward <alan.hayward@arm.com>
2775 David Sherwood <david.sherwood@arm.com>
2777 * target.h (vec_perm_indices): New typedef.
2778 (auto_vec_perm_indices): Likewise.
2779 * optabs-query.h: Include target.h
2780 (can_vec_perm_p): Take a vec_perm_indices *.
2781 * optabs-query.c (can_vec_perm_p): Likewise.
2782 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
2783 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2784 * tree-vect-generic.c (lower_vec_perm): Likewise.
2785 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
2786 (vect_grouped_load_supported): Likewise.
2787 (vect_shift_permute_load_chain): Likewise.
2788 (vect_permute_store_chain): Use auto_vec_perm_indices.
2789 (vect_permute_load_chain): Likewise.
2790 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
2791 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
2792 Update uses of can_vec_perm_p.
2793 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
2794 mode with a number of elements. Take a vec_perm_indices *.
2795 (vect_create_epilog_for_reduction): Update accordingly.
2796 Use auto_vec_perm_indices.
2797 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
2798 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
2799 (vect_transform_slp_perm_load): Likewise.
2800 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
2801 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
2802 (vect_gen_perm_mask_checked): Likewise.
2803 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
2804 (vect_gen_perm_mask_checked): Likewise.
2805 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
2806 (vectorizable_store): Likewise.
2807 (vectorizable_load): Likewise.
2808 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
2809 (vectorizable_bswap): Likewise.
2811 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2812 Alan Hayward <alan.hayward@arm.com>
2813 David Sherwood <david.sherwood@arm.com>
2815 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
2816 * tree.c (build_vector): Likewise.
2817 (build_vector_from_ctor): Update accordingly.
2818 (build_vector_from_val): Likewise.
2819 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
2820 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
2821 * tree-vect-generic.c (add_rshift): Likewise.
2822 (expand_vector_divmod): Likewise.
2823 (optimize_vector_constructor): Likewise.
2824 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2825 (vect_transform_slp_perm_load): Likewise.
2826 (vect_schedule_slp_instance): Likewise.
2827 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
2828 (vectorizable_call): Likewise.
2829 (vect_gen_perm_mask_any): Likewise. Add elements in order.
2830 * expmed.c (make_tree): Likewise.
2831 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
2832 a vector passed to build_vector.
2833 (fold_convert_const): Likewise.
2834 (exact_inverse): Likewise.
2835 (fold_ternary_loc): Likewise.
2836 (fold_relational_const): Likewise.
2837 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
2838 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
2839 (const_unop): Likewise. Store the reduction accumulator in a
2840 variable rather than an array.
2841 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
2842 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
2843 the new vector, rather than constructing it after the input arrays.
2844 (native_interpret_vector): Use auto_vec<tree> when building
2845 a vector passed to build_vector. Add elements in order.
2846 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
2847 auto_vec<tree> when building a vector passed to build_vector.
2848 (vect_create_epilog_for_reduction): Likewise.
2849 (vectorizable_induction): Likewise.
2850 (get_initial_def_for_reduction): Likewise. Fix indentation of
2852 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
2854 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
2855 passed to build_vector.
2857 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2858 Alan Hayward <alan.hayward@arm.com>
2859 David Sherwood <david.sherwood@arm.com>
2861 * tree-core.h (tree_base::u): Add an "nelts" field.
2862 (tree_vector): Use VECTOR_CST_NELTS as the length.
2863 * tree.c (tree_size): Likewise.
2864 (make_vector): Initialize VECTOR_CST_NELTS.
2865 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
2866 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
2867 TYPE_VECTOR_SUBPARTS.
2868 * expr.c (const_vector_mask_from_tree): Consistently use "units"
2869 as the number of units, setting it from VECTOR_CST_NELTS.
2870 (const_vector_from_tree): Likewise.
2871 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
2872 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
2873 (fold_negate_expr_1): Likewise.
2874 (fold_convert_const): Likewise.
2875 (const_binop): Likewise. Differentiate the number of output and
2877 (const_unop): Likewise.
2878 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
2879 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
2880 in cases that did the opposite.
2882 2017-09-14 Richard Biener <rguenther@suse.de>
2884 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
2887 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
2889 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
2891 2017-09-14 Jakub Jelinek <jakub@redhat.com>
2894 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
2895 if and where to split a bb, except for splitting before debug insn
2896 sequences followed by non-label real insn. Delete debug insns
2897 in between basic blocks.
2899 * combine.c (make_compound_operation_int): Formatting fixes.
2901 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
2902 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2903 * config/netbsd.h (LINK_EH_SPEC): Likewise.
2904 * config/sol2.h (LINK_EH_SPEC): Likewise.
2905 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2906 * config/s390/linux.h (LINK_SPEC): Likewise.
2907 * config/freebsd.h (LINK_EH_SPEC): Likewise.
2908 * config/openbsd.h (LINK_EH_SPEC): Likewise.
2909 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2910 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
2911 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
2912 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
2913 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
2914 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
2916 2017-09-13 Jakub Jelinek <jakub@redhat.com>
2918 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
2920 (ENDFILE_LINUX_SPEC): Likewise.
2921 (LINK_EH_SPEC): Likewise.
2922 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
2923 (LINK_OS_LINUX_SPEC32): Likewise.
2924 (LINK_OS_LINUX_SPEC64): Likewise.
2925 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
2926 (LINK_OS_LINUX_SPEC): Likewise.
2928 2017-09-13 Martin Liska <mliska@suse.cz>
2931 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
2932 CASE_HIGH is NULL_TREE.
2934 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
2935 Alan Hayward <alan.hayward@arm.com>
2936 David Sherwood <david.sherwood@arm.com>
2938 * target.def (secondary_memory_needed): New hook.
2939 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
2940 instead of SECONDARY_MEMORY_NEEDED.
2941 (secondary_memory_needed_mode): Likewise.
2942 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
2943 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
2944 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
2945 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
2946 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
2947 * doc/tm.texi: Regenerate.
2948 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
2949 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
2950 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2951 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
2952 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
2953 * config/i386/i386.c (inline_secondary_memory_needed): Put the
2954 mode argument first and change the reg_class arguments to reg_class_t.
2955 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
2956 Make static. Update the call to inline_secondary_memory_needed.
2957 (ix86_register_move_cost): Update the call to
2958 inline_secondary_memory_needed.
2959 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2960 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
2962 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
2963 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2965 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
2966 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
2967 * config/mips/mips.c (mips_secondary_memory_needed): Make static
2968 and match hook interface. Add comment from mips.h.
2969 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2970 * config/mmix/mmix.md (truncdfsf2): Refer to
2971 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
2973 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
2974 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
2975 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2976 (pa_secondary_memory_needed): New function.
2977 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
2978 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
2979 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2980 (pdp11_secondary_memory_needed): Make static and match hook interface.
2981 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
2982 * config/powerpcspe/powerpcspe-protos.h
2983 (rs6000_secondary_memory_needed_ptr): Delete.
2984 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
2986 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2987 (rs6000_option_override_internal): Assign to
2988 targetm.secondary_memory_needed rather than
2989 rs6000_secondary_memory_needed_ptr.
2990 (rs6000_secondary_memory_needed): Match hook interface.
2991 (rs6000_debug_secondary_memory_needed): Likewise.
2992 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
2993 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
2994 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
2995 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
2996 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
2997 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
2999 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
3000 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3001 (rs6000_option_override_internal): Assign to
3002 targetm.secondary_memory_needed rather than
3003 rs6000_secondary_memory_needed_ptr.
3004 (rs6000_secondary_memory_needed): Match hook interface.
3005 (rs6000_debug_secondary_memory_needed): Likewise.
3006 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
3007 * config/s390/s390.c (s390_secondary_memory_needed): New function.
3008 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3009 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
3010 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3011 (sparc_secondary_memory_needed): New function.
3012 * lra-constraints.c (check_and_process_move): Refer to
3013 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3015 (curr_insn_transform): Likewise.
3016 (process_alt_operands): Use targetm.secondary_memory_needed
3017 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3018 (check_secondary_memory_needed_p): Likewise.
3019 (choose_split_class): Likewise.
3020 * reload.c: Unconditionally include code that was previously
3021 conditional on SECONDARY_MEMORY_NEEDED.
3022 (push_secondary_reload): Use targetm.secondary_memory_needed
3023 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3024 (push_reload): Likewise.
3025 * reload1.c: Unconditionally include code that was previously
3026 conditional on SECONDARY_MEMORY_NEEDED.
3027 (choose_reload_regs): Use targetm.secondary_memory_needed
3028 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3029 (gen_reload): Likewise.
3030 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
3032 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
3033 Alan Hayward <alan.hayward@arm.com>
3034 David Sherwood <david.sherwood@arm.com>
3036 * target.def (secondary_memory_needed_mode): New hook:
3037 * targhooks.c (default_secondary_memory_needed_mode): Declare.
3038 * targhooks.h (default_secondary_memory_needed_mode): New function.
3039 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
3040 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
3041 * doc/tm.texi: Regenerate.
3042 * lra-constraints.c (check_and_process_move): Use
3043 targetm.secondary_memory_needed_mode instead of
3044 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
3045 (curr_insn_transform): Likewise.
3046 * reload.c (get_secondary_mem): Likewise.
3047 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3048 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
3050 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3051 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3052 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
3053 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3054 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
3056 * config/powerpcspe/powerpcspe-protos.h
3057 (rs6000_secondary_memory_needed_mode): Delete.
3058 * config/powerpcspe/powerpcspe.c
3059 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3060 (rs6000_secondary_memory_needed_mode): Make static.
3061 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3062 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
3064 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
3066 (rs6000_secondary_memory_needed_mode): Make static.
3067 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3068 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
3069 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3070 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3071 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
3073 (sparc_secondary_memory_needed_mode): New function.
3074 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
3076 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
3078 * config/aarch64/constraints.md (Umq): New constraint.
3079 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
3081 (mov<mode>): Update condition.
3083 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3085 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
3086 when bitposition is the same.
3088 2017-09-13 Richard Biener <rguenther@suse.de>
3090 * dwarf2out.c (output_die_symbol): Remove.
3091 (output_die): Do not output a DIEs symbol.
3093 2017-09-13 Richard Biener <rguenther@suse.de>
3096 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
3097 default-def to avoid breaking iterator update with the weird
3098 interaction with cgraph_update_edges_for_call_stmt_node.
3100 2017-09-13 Richard Biener <rguenther@suse.de>
3102 * tree-cfg.c (verify_gimple_assign_binary): Add verification
3103 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
3104 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
3105 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
3107 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
3109 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
3110 Disable pc relative literal load irrespective of
3111 TARGET_FIX_ERR_A53_84341 for default.
3113 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
3115 * config/sparc/sparc.c (output_return): Output the source location of
3116 the insn in the delay slot, if any.
3117 (output_sibcall): Likewise.
3119 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
3122 * common.opt (-static-pie): New alias.
3123 (shared): Negate static-pie.
3124 (-no-pie): Update help text.
3126 (static-pie): New option.
3127 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
3128 -static-pie support.
3129 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
3130 (LINK_EH_SPEC): Likewise.
3131 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3132 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
3133 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
3134 * gcc.c (LINK_COMMAND_SPEC): Likewise.
3135 (init_gcc_specs): Likewise.
3136 (init_spec): Likewise.
3137 (display_help): Update help message for -pie.
3138 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
3141 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
3143 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
3144 (movdi_aarch64): Likewise.
3145 (movti_aarch64): Likewise.
3147 2017-09-12 Simon Wright <simon@pushface.org>
3150 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
3151 calculation of the minor version, always output as 0.
3153 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3156 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
3157 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
3158 on it early, rather than manual conversion late. For
3159 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
3160 instead of performing manual conversion.
3162 2017-09-12 Carl Love <cel@us.ibm.com>
3164 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
3165 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
3167 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
3168 VMULOSW): Add definitions.
3169 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
3170 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
3171 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
3172 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
3173 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
3175 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3177 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
3179 (movti_aarch64): Likewise.
3180 (movdf_aarch64): Likewise.
3181 (movtf_aarch64): Likewise.
3182 (load_pairdi): Likewise.
3183 (store_pairdi): Likewise.
3184 (load_pairdf): Likewise.
3185 (store_pairdf): Likewise.
3186 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
3187 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
3188 (ldr_got_small_<mode>): Likewise.
3189 (ldr_got_small_28k_<mode>): Likewise.
3190 (ldr_got_tiny): Likewise.
3191 * config/aarch64/iterators.md (ldst_sz): New.
3192 (ldpstp_sz): Likewise.
3193 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
3195 (thunderx_load): Split load_8 to load_16.
3196 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
3198 (thunderx2t99_storepair_basic): Split store_8 to store_16.
3199 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
3200 (xgene1_store_pair): Split store_8 to store_16.
3201 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
3202 (falkor_st_0_st_sd): Split store_8 to store_16.
3204 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3206 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
3207 and store1/2/3/4 to store_4/8/12/16.
3208 * config/aarch64/aarch64.md: Update for rename.
3209 * config/arm/arm.md: Likewise.: Likewise.
3210 * config/arm/arm.c: Likewise.
3211 * config/arm/thumb1.md: Likewise.
3212 * config/arm/thumb2.md: Likewise.
3213 * config/arm/vfp.md: Likewise.
3214 * config/arm/arm-generic.md: Likewise.
3215 * config/arm/arm1020e.md: Likewise.
3216 * config/arm/arm1026ejs.md: Likewise.
3217 * config/arm/arm1136jfs.md: Likewise.
3218 * config/arm/arm926ejs.md: Likewise.
3219 * config/arm/cortex-a15.md: Likewise.
3220 * config/arm/cortex-a17.md: Likewise.
3221 * config/arm/cortex-a5.md: Likewise.
3222 * config/arm/cortex-a53.md: Likewise.
3223 * config/arm/cortex-a57.md: Likewise.
3224 * config/arm/cortex-a7.md: Likewise.
3225 * config/arm/cortex-a8.md: Likewise.
3226 * config/arm/cortex-a9.md: Likewise.
3227 * config/arm/cortex-m4.md: Likewise.
3228 * config/arm/cortex-m7.md: Likewise.
3229 * config/arm/cortex-r4.md: Likewise.
3230 * config/arm/exynos-m1.md: Likewise.
3231 * config/arm/fa526.md: Likewise.
3232 * config/arm/fa606te.md: Likewise.
3233 * config/arm/fa626te.md: Likewise.
3234 * config/arm/fa726te.md: Likewise.
3235 * config/arm/fmp626.md: Likewise.
3236 * config/arm/iwmmxt.md: Likewise.
3237 * config/arm/ldmstm.md: Likewise.
3238 * config/arm/marvell-pj4.md: Likewise.
3239 * config/arm/xgene1.md: Likewise.
3240 * config/aarch64/thunderx.md: Likewise.
3241 * config/aarch64/thunderx2t99.md: Likewise.
3242 * config/aarch64/falkor.md: Likewise.
3244 2017-09-12 Martin Liska <mliska@suse.cz>
3246 * attribs.c (private_lookup_attribute): New function.
3247 * attribs.h (private_lookup_attribute): Declared here.
3248 (lookup_attribute): Called from this place.
3250 2017-09-12 Richard Biener <rguenther@suse.de>
3252 PR tree-optimization/82157
3253 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
3254 stmts with side-effects.
3256 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3257 Alan Hayward <alan.hayward@arm.com>
3258 David Sherwood <david.sherwood@arm.com>
3260 * target.def (hard_regno_nregs): New hook.
3261 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
3262 * targhooks.h (default_hard_regno_nregs): Declare.
3263 * targhooks.c (default_hard_regno_nregs): New function.
3264 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
3265 (TARGET_HARD_REGNO_NREGS): ...this hook.
3266 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
3267 (CLASS_MAX_NREGS): Likewise.
3268 * doc/tm.texi: Regenerate.
3269 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
3270 instead of HARD_REGNO_NREGS.
3271 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
3272 HARD_REGNO_NREGS in the comment.
3273 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
3274 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
3275 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
3276 Return an unsigned int.
3277 (TARGET_HARD_REGNO_NREGS): Redefine.
3278 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
3279 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
3280 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
3281 (arc_hard_regno_nregs): New function.
3282 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
3283 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
3284 (arm_hard_regno_nregs): New function.
3285 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
3286 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
3287 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
3288 (TARGET_HARD_REGNO_NREGS): Redefine.
3289 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
3290 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
3291 (HARD_REGNO_NREGS): Delete.
3292 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
3293 (cr16_hard_regno_nregs): New function.
3294 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
3295 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
3296 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
3297 (cris_hard_regno_nregs): New function.
3298 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
3299 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
3300 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3301 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
3302 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
3303 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
3304 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
3305 (frv_hard_regno_nregs): Make static. Take and return an
3307 (frv_class_max_nregs): Remove outdated copy of documentation.
3308 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
3309 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
3310 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
3311 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
3312 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
3313 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
3314 (TARGET_HARD_REGNO_NREGS): Redefine.
3315 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
3316 (CLASS_MAX_NREGS): Update comment.
3317 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
3318 (ia64_hard_regno_nregs): New function.
3319 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
3320 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
3321 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
3322 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
3323 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
3325 (m32c_hard_regno_nregs): Likewise. Make static.
3326 (TARGET_HARD_REGNO_NREGS): Redefine.
3327 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
3328 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
3329 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
3330 (m68k_hard_regno_nregs): New function.
3331 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
3332 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
3333 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
3334 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
3335 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
3336 Take and return an unsigned int.
3337 (TARGET_HARD_REGNO_NREGS): Redefine.
3338 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
3339 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3340 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
3341 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
3342 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
3343 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
3344 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
3345 (msp430_hard_regno_nregs): Make static. Take and return an
3347 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
3348 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
3349 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
3350 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
3351 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
3352 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
3353 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
3354 (TARGET_HARD_REGNO_NREGS): Redefine.
3355 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
3356 (PA_HARD_REGNO_NREGS): ...this.
3357 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
3358 (PA_HARD_REGNO_NREGS): ...this.
3359 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3360 (pa_hard_regno_nregs): New function.
3361 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
3362 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
3363 (pdp11_hard_regno_nregs): New function.
3364 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
3365 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
3366 (rs6000_hard_regno_nregs_hook): New function.
3367 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
3368 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3369 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
3370 Take and return an unsigned int. Move earlier in file.
3371 (TARGET_HARD_REGNO_NREGS): Redefine.
3372 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
3373 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
3374 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
3375 (rl78_hard_regno_nregs): Make static. Take and return an
3377 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
3378 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
3379 (rs6000_hard_regno_nregs_hook): New function.
3380 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
3381 * config/rx/rx.c (rx_hard_regno_nregs): New function.
3382 (TARGET_HARD_REGNO_NREGS): Redefine.
3383 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
3384 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
3385 instead of HARD_REGNO_NREGS.
3386 (s390_hard_regno_nregs): New function.
3387 (s390_hard_regno_mode_ok): Add comment from s390.h.
3388 (TARGET_HARD_REGNO_NREGS): Redefine.
3389 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
3390 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
3391 (sh_hard_regno_nregs): New function.
3392 (sh_pass_in_reg_p): Use it.
3393 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
3394 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
3395 (sparc_hard_regno_nregs): New function.
3396 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
3397 * config/spu/spu.c (spu_hard_regno_nregs): New function.
3398 (spu_function_arg_advance): Use it, supplying a valid register number.
3399 (TARGET_HARD_REGNO_NREGS): Redefine.
3400 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
3401 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
3402 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
3403 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
3404 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
3405 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
3406 (CLASS_MAX_NREGS): Remove copy of old documentation.
3407 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
3408 (visium_hard_regno_nregs): New function.
3409 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
3410 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
3411 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3412 xtensa_hard_regno_nregs): New function.
3413 * system.h (HARD_REGNO_NREGS): Poison.
3415 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3417 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
3418 hard_regno_nregs instead of HARD_REGNO_NREGS.
3419 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
3420 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
3421 (c6x_expand_epilogue): Likewise.
3422 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
3423 (frv_read_iacc_argument): Likewise.
3424 * config/sh/sh.c: Include regs.h.
3425 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3426 (regs_used): Likewise.
3427 (output_stack_adjust): Likewise.
3428 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
3429 * expmed.c: Include regs.h.
3430 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3431 * ree.c: Include regs.h.
3432 (combine_reaching_defs): Use hard_regno_nregs instead of
3434 (add_removable_extension): Likewise.
3436 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3438 * regs.h (hard_regno_nregs): Turn into a function.
3439 (end_hard_regno): Update accordingly.
3440 * caller-save.c (setup_save_areas): Likewise.
3441 (save_call_clobbered_regs): Likewise.
3442 (replace_reg_with_saved_mem): Likewise.
3443 (insert_restore): Likewise.
3444 (insert_save): Likewise.
3445 * combine.c (can_change_dest_mode): Likewise.
3446 (move_deaths): Likewise.
3447 (distribute_notes): Likewise.
3448 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
3449 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
3450 (rs6000_split_multireg_move): Likewise.
3451 (rs6000_register_move_cost): Likewise.
3452 (rs6000_memory_move_cost): Likewise.
3453 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
3454 (rs6000_split_multireg_move): Likewise.
3455 (rs6000_register_move_cost): Likewise.
3456 (rs6000_memory_move_cost): Likewise.
3457 * cselib.c (cselib_reset_table): Likewise.
3458 (cselib_lookup_1): Likewise.
3459 * emit-rtl.c (set_mode_and_regno): Likewise.
3460 * function.c (aggregate_value_p): Likewise.
3461 * ira-color.c (setup_profitable_hard_regs): Likewise.
3462 (check_hard_reg_p): Likewise.
3463 (calculate_saved_nregs): Likewise.
3464 (assign_hard_reg): Likewise.
3465 (improve_allocation): Likewise.
3466 (calculate_spill_cost): Likewise.
3467 * ira-emit.c (modify_move_list): Likewise.
3468 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
3469 (ira_hard_reg_in_set_p): Likewise.
3470 * ira.c (setup_reg_mode_hard_regset): Likewise.
3471 (clarify_prohibited_class_mode_regs): Likewise.
3472 (check_allocation): Likewise.
3473 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3474 (lra_setup_reg_renumber): Likewise.
3475 (setup_try_hard_regno_pseudos): Likewise.
3476 (spill_for): Likewise.
3477 (assign_hard_regno): Likewise.
3478 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
3479 * lra-constraints.c (in_class_p): Likewise.
3480 (lra_constraint_offset): Likewise.
3481 (simplify_operand_subreg): Likewise.
3482 (lra_constraints): Likewise.
3483 (split_reg): Likewise.
3484 (split_if_necessary): Likewise.
3485 (invariant_p): Likewise.
3486 (inherit_in_ebb): Likewise.
3487 * lra-lives.c (process_bb_lives): Likewise.
3488 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
3489 (get_hard_regs): Likewise.
3490 (do_remat): Likewise.
3491 * lra-spills.c (assign_spill_hard_regs): Likewise.
3492 * mode-switching.c (create_pre_exit): Likewise.
3493 * postreload.c (reload_combine_recognize_pattern): Likewise.
3494 * recog.c (peep2_find_free_register): Likewise.
3495 * regcprop.c (kill_value_regno): Likewise.
3496 (set_value_regno): Likewise.
3497 (copy_value): Likewise.
3498 (maybe_mode_change): Likewise.
3499 (find_oldest_value_reg): Likewise.
3500 (copyprop_hardreg_forward_1): Likewise.
3501 * regrename.c (check_new_reg_p): Likewise.
3502 (regrename_do_replace): Likewise.
3503 * reload.c (push_reload): Likewise.
3504 (combine_reloads): Likewise.
3505 (find_dummy_reload): Likewise.
3506 (operands_match_p): Likewise.
3507 (find_reloads): Likewise.
3508 (find_equiv_reg): Likewise.
3509 (reload_adjust_reg_for_mode): Likewise.
3510 * reload1.c (count_pseudo): Likewise.
3511 (count_spilled_pseudo): Likewise.
3512 (find_reg): Likewise.
3513 (clear_reload_reg_in_use): Likewise.
3514 (free_for_value_p): Likewise.
3515 (allocate_reload_reg): Likewise.
3516 (choose_reload_regs): Likewise.
3517 (reload_adjust_reg_for_temp): Likewise.
3518 (emit_reload_insns): Likewise.
3519 (delete_output_reload): Likewise.
3520 * rtlanal.c (subreg_get_info): Likewise.
3521 * sched-deps.c (sched_analyze_reg): Likewise.
3522 * sel-sched.c (init_regs_for_mode): Likewise.
3523 (mark_unavailable_hard_regs): Likewise.
3524 (choose_best_reg_1): Likewise.
3525 (verify_target_availability): Likewise.
3526 * valtrack.c (dead_debug_insert_temp): Likewise.
3527 * var-tracking.c (track_loc_p): Likewise.
3528 (emit_note_insn_var_location): Likewise.
3529 * varasm.c (make_decl_rtl): Likewise.
3530 * reginfo.c (choose_hard_reg_mode): Likewise.
3531 (init_reg_modes_target): Refer directly to
3532 this_target_regs->x_hard_regno_nregs.
3534 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3536 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
3537 instead of hard_regno_nregs.
3539 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3541 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
3542 end_hard_regno instead of hard_regno_nregs.
3543 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
3544 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
3545 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
3546 * ira-color.c (improve_allocation): Likewise.
3547 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3548 * lra-lives.c (mark_regno_live): Likewise.
3549 (mark_regno_dead): Likewise.
3550 * lra-remat.c (operand_to_remat): Likewise.
3551 * lra.c (collect_non_operand_hard_regs): Likewise.
3552 * postreload.c (reload_combine_note_store): Likewise.
3553 (move2add_valid_value_p): Likewise.
3554 * reload.c (regno_clobbered_p): Likewise.
3556 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3558 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
3560 * config/v850/v850.c (v850_reorg): Likewise.
3561 * reload.c (refers_to_regno_for_reload_p): Likewise.
3562 (find_equiv_reg): Likewise.
3563 * reload1.c (reload_reg_reaches_end_p): Likewise.
3565 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3567 * caller-save.c (add_used_regs): Use REG_NREGS instead of
3569 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
3570 * config/arm/arm.c (output_move_neon): Likewise.
3571 (arm_attr_length_move_neon): Likewise.
3572 (neon_split_vcombine): Likewise.
3573 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
3574 (c6x_mark_reg_written): Likewise.
3575 (c6x_dwarf_register_span): Likewise.
3576 * config/i386/i386.c (ix86_save_reg): Likewise.
3577 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
3578 (rws_access_reg): Likewise.
3579 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
3580 * mode-switching.c (create_pre_exit): Likewise.
3581 * ree.c (combine_reaching_defs): Likewise.
3582 (add_removable_extension): Likewise.
3583 * regcprop.c (find_oldest_value_reg): Likewise.
3584 (copyprop_hardreg_forward_1): Likewise.
3585 * reload.c (reload_inner_reg_of_subreg): Likewise.
3586 (push_reload): Likewise.
3587 (combine_reloads): Likewise.
3588 (find_dummy_reload): Likewise.
3589 (reload_adjust_reg_for_mode): Likewise.
3590 * reload1.c (find_reload_regs): Likewise.
3591 (forget_old_reloads_1): Likewise.
3592 (reload_reg_free_for_value_p): Likewise.
3593 (reload_adjust_reg_for_temp): Likewise.
3594 (emit_reload_insns): Likewise.
3595 (delete_output_reload): Likewise.
3596 * sel-sched.c (choose_best_reg_1): Likewise.
3597 (choose_best_pseudo_reg): Likewise.
3599 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3600 Alan Hayward <alan.hayward@arm.com>
3601 David Sherwood <david.sherwood@arm.com>
3603 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
3604 * target.def (slow_unaligned_access): New hook.
3605 * targhooks.h (default_slow_unaligned_access): Declare.
3606 * targhooks.c (default_slow_unaligned_access): New function.
3607 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
3608 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
3609 * doc/tm.texi: Regenerate.
3610 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
3611 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
3612 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
3614 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
3615 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
3617 (rs6000_slow_unaligned_access): New function.
3618 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3619 (expand_block_compare): Likewise.
3620 (expand_strn_compare): Likewise.
3621 (rs6000_rtx_costs): Likewise.
3622 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
3623 (riscv_slow_unaligned_access): Likewise.
3624 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
3625 (riscv_slow_unaligned_access_p): ...this and make static.
3626 (riscv_option_override): Update accordingly.
3627 (riscv_slow_unaligned_access): New function.
3628 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3629 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
3630 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3631 (rs6000_slow_unaligned_access): New function.
3632 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3633 (rs6000_rtx_costs): Likewise.
3634 * config/rs6000/rs6000-string.c (expand_block_compare)
3635 (expand_strn_compare): Use targetm.slow_unaligned_access instead
3636 of SLOW_UNALIGNED_ACCESS.
3637 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
3638 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
3639 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
3640 of SLOW_UNALIGNED_ACCESS.
3641 * expmed.c (simple_mem_bitfield_p): Likewise.
3642 * expr.c (alignment_for_piecewise_move): Likewise.
3643 (emit_group_load_1): Likewise.
3644 (emit_group_store): Likewise.
3645 (copy_blkmode_from_reg): Likewise.
3646 (emit_push_insn): Likewise.
3647 (expand_assignment): Likewise.
3648 (store_field): Likewise.
3649 (expand_expr_real_1): Likewise.
3650 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
3651 * lra-constraints.c (simplify_operand_subreg): Likewise.
3652 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3653 * gimple-ssa-store-merging.c: Likewise in block comment at start
3655 * tree-ssa-strlen.c: Include target.h.
3656 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
3657 of SLOW_UNALIGNED_ACCESS.
3658 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
3660 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3662 PR rtl-optimization/82185
3663 * expmed.c (emit_store_flag_int): Only test tem if it has been
3666 2017-09-12 Richard Biener <rguenther@suse.de>
3669 * match.pd ((FTYPE) N CMP CST): Fix typo.
3671 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
3673 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
3675 (mips_near_type_p): Add 'short_call' attribute as a synonym
3677 * doc/extend.texi (short_call): Document new function attribute.
3679 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3682 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
3683 assertion check that in the condition.
3684 (get_atomic_generic_size): Likewise. Before testing if parameter
3685 has pointer type, if it has array type, call for C++
3686 default_conversion to perform array-to-pointer conversion.
3688 2017-09-12 Richard Biener <rguenther@suse.de>
3690 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
3691 for operations we cannot scalarize.
3693 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
3695 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
3696 vectors heap vectors. Clean up comments.
3697 Make visited_bbs a reference.
3698 (profitable_jump_thread_path): Make GC
3699 vectors heap vectors. Clean up comments.
3701 (convert_and_register_jump_thread_path): Make GC vectors heap
3703 (check_subpath_and_update_thread_path): Same. Clean up comments.
3704 Make visited_bbs a reference.
3705 (handle_phi): Abstract common code to to
3706 register_jump_thread_path_if_profitable.
3707 Rename VAR_BB to DEF_BB.
3709 Make GC vectors heap vectors.
3710 Make visited_bbs a reference.
3711 (handle_assignment): Same.
3712 (register_jump_thread_path_if_profitable): New.
3713 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
3715 Make GC vectors heap vectors. Clean up comments.
3716 Make visited_bbs a reference.
3717 (find_jump_threads_backwards): Make visited_bbs live in the stack.
3718 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
3721 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
3724 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
3725 words of E_DImode object are reachable by xtensa_uimm8x4 access.
3727 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
3729 Revert r251800 and r251799.
3731 2017-09-11 Martin Jambor <mjambor@suse.cz>
3734 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
3735 arguments in advance.
3736 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
3737 use it to find predecessor edges.
3738 (naive_outof_ssa): Collect vector of predecessors.
3740 2017-09-08 Jason Merrill <jason@redhat.com>
3742 PR c++/70029 - ICE with ref-qualifier and -flto
3743 * langhooks.h (struct lang_hooks_for_types): Add
3744 copy_lang_qualifiers.
3745 * attribs.c (build_type_attribute_qual_variant): Use it.
3746 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
3748 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
3749 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
3751 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
3754 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
3755 (*mulsi3_sp64): New instruction.
3756 (mulsi3): New expander.
3758 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
3760 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
3762 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3764 * sancov.c: Include memmodel.h.
3766 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
3769 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
3772 2017-09-07 Carl Love <cel@us.ibm.com>
3774 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
3775 the sldi instruction.
3777 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
3779 * sancov.c: Include tm_p.h.
3781 2017-09-07 Jakub Jelinek <jakub@redhat.com>
3784 * output.h (switch_to_other_text_partition): New declaration.
3785 * varasm.c (switch_to_other_text_partition): New function.
3786 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
3787 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
3788 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
3789 to the other text partition before emitting LCL label and switch back
3790 after emitting the word after it.
3792 2017-09-07 Richard Biener <rguenther@suse.de>
3794 * passes.def (pass_split_crit_edges): Remove instance before PRE.
3795 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
3796 critical edges here, after loop init.
3797 (pass_data_pre): Remove PROP_no_crit_edges flags.
3798 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
3799 for valueization of call args to avoid leaking VN_TOP.
3800 (visit_use): Assert we do not visit default defs.
3801 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
3802 Use error_mark_node to more easily detect leaking VN_TOP.
3803 All default-defs are varying, not VN_TOP. Mark them visited.
3804 (run_scc_vn): Make code match comment.
3806 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3808 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
3809 OPTION_MASK_FLOAT128_KEYWORD.
3810 (POWERPC_MASKS): Likewise.
3811 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
3812 support for the -mfloat128-type option, and make -mfloat128
3813 default on PowerPC Linux systems. Define or undefine
3814 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
3815 Define __float128 to be __ieee128 if IEEE 128-bit support is
3816 enabled, or undefine it.
3817 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
3818 Delete defining __FLOAT128_TYPE__.
3819 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
3820 -mfloat128-type option and make -mfloat128 default on PowerPC
3822 (TARGET_FLOAT128_TYPE): Likewise.
3823 (-mfloat128-type): Likewise.
3824 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3825 Delete the -mfloat128-type option and make -mfloat128 default on
3826 PowerPC Linux systems. Always use __ieee128 to be the keyword for
3827 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
3828 128-bit floating point is enabled. Change tests from using
3829 -mfloat128-type to -mfloat128.
3830 (rs6000_mangle_type): Use the correct mangling for the __float128
3831 type even if normal long double is restricted to 64-bits.
3832 (floatn_mode): Enable the _Float128 type by default on VSX Linux
3834 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
3835 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
3836 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
3838 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
3839 documentation for -mfloat128.
3841 2017-09-06 Olivier Hainque <hainque@adacore.com>
3843 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
3845 2017-09-06 Wish Wu <wishwu007@gmail.com>
3846 Jakub Jelinek <jakub@redhat.com>
3848 * asan.c (initialize_sanitizer_builtins): Add
3849 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
3850 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
3851 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
3852 BT_FN_VOID_UINT64_PTR variables.
3853 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
3854 (BT_FN_VOID_UINT16_UINT16): Likewise.
3855 (BT_FN_VOID_UINT32_UINT32): Likewise.
3856 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
3857 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
3858 (BT_FN_VOID_UINT64_PTR): Likewise.
3859 * common.opt (flag_sanitize_coverage): New variable.
3860 (fsanitize-coverage=trace-pc): Remove.
3861 (fsanitize-coverage=): Add.
3862 * flag-types.h (enum sanitize_coverage_code): New enum.
3863 * fold-const.c (fold_range_test): Disable non-short-circuit
3864 optimization if flag_sanitize_coverage.
3865 (fold_truth_andor): Likewise.
3866 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
3867 * opts.c (COVERAGE_SANITIZER_OPT): Define.
3868 (coverage_sanitizer_opts): New array.
3869 (get_closest_sanitizer_option): Add OPTS argument, handle also
3870 OPT_fsanitize_coverage_.
3871 (parse_sanitizer_options): Adjusted to also handle
3872 OPT_fsanitize_coverage_.
3873 (common_handle_option): Add OPT_fsanitize_coverage_.
3874 * sancov.c (instrument_comparison, instrument_switch): New function.
3875 (sancov_pass): Add trace-cmp support.
3876 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
3877 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
3878 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
3879 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
3880 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
3881 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
3882 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
3883 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
3884 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
3885 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
3887 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3889 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
3890 error. Only quit immediately if parsing is complete.
3891 (BEGIN): Initialize fatal_err and parse_done.
3892 (begin fpu, end fpu): Check number of arguments.
3893 (begin arch, end arch): Likewise.
3894 (begin cpu, end cpu): Likewise.
3895 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
3896 (optalias): Likewise.
3898 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
3900 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
3901 * config/arm/arm-isa.h: Delete. Move definitions to ...
3902 * arm-cpus.in: ... here. Use new feature and fgroup values.
3903 * config/arm/arm.c (arm_option_override): Use lower case for feature
3905 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
3906 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
3907 * config/arm/parsecpu.awk (END): Add new command 'isa'.
3909 (print_isa_bits_for): New function.
3910 (gen_isa): New function.
3911 (gen_comm_data): Use print_isa_bits_for.
3912 (define feature): New keyword.
3913 (define fgroup): New keyword.
3914 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
3915 (arm-isa.h): Add rule to generate file.
3916 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
3917 case for feature bit names.
3919 2017-09-06 Richard Biener <rguenther@suse.de>
3921 * tree-ssa-pre.c (NECESSARY): Remove.
3922 (create_expression_by_pieces): Do not touch pass-local flags.
3923 (insert_into_preds_of_block): Likewise.
3924 (do_pre_regular_insertion): Likewise.
3925 (eliminate_insert): Likewise.
3926 (eliminate_dom_walker::before_dom_children): Likewise.
3927 (fini_eliminate): Do not look at inserted_exprs.
3928 (mark_operand_necessary): Remove.
3929 (remove_dead_inserted_code): Replace with simple work-list
3930 algorithm based on inserted_exprs and SSA uses.
3931 (pass_pre::execute): Re-order fini_eliminate and
3932 remove_dead_inserted_code.
3934 2017-09-06 Olivier Hainque <hainque@adacore.com>
3936 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
3937 for VxWorks 7. Adjust surrounding comments.
3939 2017-09-06 Richard Biener <rguenther@suse.de>
3941 * gimple-ssa-strength-reduction.c
3942 (find_candidates_dom_walker::before_dom_children): Also allow
3945 2017-09-06 Richard Biener <rguenther@suse.de>
3947 PR tree-optimization/82108
3948 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
3949 for gap in the non-permutation SLP case.
3951 2017-09-06 Martin Jambor <mjambor@suse.cz>
3953 PR tree-optimization/82078
3954 * tree-sra.c (sort_and_splice_var_accesses): Move call to
3955 add_access_to_work_queue...
3956 (build_accesses_from_assign): ...here.
3957 (propagate_all_subaccesses): Make sure racc is the group
3958 representative, if there is one.
3960 2017-09-06 Jakub Jelinek <jakub@redhat.com>
3963 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
3966 2017-09-06 Richard Biener <rguenther@suse.de>
3968 * gimple-ssa-strength-reduction.c
3969 (find_candidates_dom_walker::before_doom_children): Use a
3970 type and not a mode check.
3972 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
3975 * config/arm/predicates.md (arm_general_adddi_operand): Create new
3977 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
3979 2017-09-05 Jeff Law <law@redhat.com>
3981 PR tree-optimization/64910
3982 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
3983 cases where we have 3 or more operands.
3985 2017-09-05 Jakub Jelinek <jakub@redhat.com>
3988 * omp-low.c (lower_omp_for): Recompute tree invariant if
3989 gimple_omp_for_initial/final is ADDR_EXPR.
3992 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
3993 into gimple val before gimplification fo the COND_EXPR.
3995 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
3997 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
3998 REGION_COPY argument.
3999 (thread_through_all_blocks): Remove unused argument to
4000 duplicate_thread_path.
4002 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4003 Alan Hayward <alan.hayward@arm.com>
4004 David Sherwood <david.sherwood@arm.com>
4006 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
4007 Take a scalar_mode rather than a machine_mode.
4008 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
4009 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
4010 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
4011 (aarch64_gen_adjusted_ldpstp): Likewise.
4012 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
4014 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4015 Alan Hayward <alan.hayward@arm.com>
4016 David Sherwood <david.sherwood@arm.com>
4018 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
4019 Take a scalar_int_mode instead of a machine_mode.
4020 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
4021 (aarch64_output_scalar_simd_mov_immediate): Likewise.
4022 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
4023 (aarch64_simd_attr_length_rglist): Delete.
4024 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
4025 a scalar_int_mode instead of a machine_mode.
4026 (aarch64_add_offset): Likewise.
4027 (aarch64_internal_mov_immediate): Likewise
4028 (aarch64_add_constant_internal): Likewise.
4029 (aarch64_add_constant): Likewise.
4030 (aarch64_movw_imm): Likewise.
4031 (aarch64_rtx_arith_op_extract_p): Likewise.
4032 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
4033 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
4034 Remove assert that the mode isn't a vector.
4035 (aarch64_output_scalar_simd_mov_immediate): Likewise.
4036 (aarch64_expand_mov_immediate): Update calls after above changes.
4037 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
4038 (aarch64_and_bitmask_imm): Check for scalar integer modes.
4039 (aarch64_move_imm): Likewise.
4040 (aarch64_can_const_movi_rtx_p): Likewise.
4041 (aarch64_strip_extend): Likewise.
4042 (aarch64_extr_rtx_p): Likewise.
4043 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
4044 a CONST_INT when the mode parameter is VOIDmode.
4045 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
4047 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4049 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
4050 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4051 (bitwise_type_for_mode): Update accordingly.
4053 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4055 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
4056 * stor-layout.c (mode_for_size_tree): Likewise.
4057 (mode_for_array): Update accordingly.
4058 (layout_decl): Likewise.
4059 (compute_record_mode): Likewise. Only set the mode once.
4061 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4063 * target.def (get_mask_mode): Change return type to opt_mode.
4065 * doc/tm.texi: Regenerate.
4066 * targhooks.h (default_get_mask_mode): Return an opt_mode.
4067 * targhooks.c (default_get_mask_mode): Likewise.
4068 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4069 * optabs-query.c (can_vec_mask_load_store_p): Update use of
4070 targetm.get_mask_mode.
4071 * tree.c (build_truth_vector_type): Likewise.
4073 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4075 * machmode.h (mode_for_vector): Return an opt_mode.
4076 * stor-layout.c (mode_for_vector): Likewise.
4077 (mode_for_int_vector): Update accordingly.
4078 (layout_type): Likewise.
4079 * config/i386/i386.c (emit_memmov): Likewise.
4080 (ix86_expand_set_or_movmem): Likewise.
4081 (ix86_expand_vector_init): Likewise.
4082 (ix86_get_mask_mode): Likewise.
4083 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
4085 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
4086 * expmed.c (extract_bit_field_1): Likewise.
4087 * expr.c (expand_expr_real_2): Likewise.
4088 * optabs-query.c (can_vec_perm_p): Likewise.
4089 (can_vec_mask_load_store_p): Likewise.
4090 * optabs.c (expand_vec_perm): Likewise.
4091 * targhooks.c (default_get_mask_mode): Likewise.
4092 * tree-vect-stmts.c (vectorizable_store): Likewise.
4093 (vectorizable_load): Likewise.
4094 (get_vectype_for_scalar_type_and_size): Likewise.
4096 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4098 * machmode.h (mode_for_int_vector): New function.
4099 * stor-layout.c (mode_for_int_vector): Likewise.
4100 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
4101 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
4102 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4103 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4104 (s390_expand_vcond): Likewise.
4106 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4108 * machmode.h (opt_machine_mode): New type.
4109 (opt_mode<T>): Allow construction from anything that can be
4111 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
4112 (mode_for_size): Return an opt_machine_mode.
4113 * stor-layout.c (mode_for_size): Likewise.
4114 (mode_for_size_tree): Update call accordingly.
4115 (bitwise_mode_for_mode): Likewise.
4116 (make_fract_type): Likewise.
4117 (make_accum_type): Likewise.
4118 * caller-save.c (replace_reg_with_saved_mem): Update call
4120 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4121 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4122 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4123 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4124 * expmed.c (extract_bit_field_1): Likewise.
4125 * reload.c (get_secondary_mem): Likewise.
4126 * varasm.c (assemble_integer): Likewise.
4127 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
4130 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4132 * machmode.h (decimal_float_mode_for_size): New function.
4133 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
4134 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
4135 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
4136 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
4137 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
4138 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
4140 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4142 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
4143 (get_builtin_sync_mode): Likewise.
4144 (expand_ifn_atomic_compare_exchange): Likewise.
4145 (expand_builtin_atomic_clear): Likewise.
4146 (expand_builtin_atomic_test_and_set): Likewise.
4147 (fold_builtin_atomic_always_lock_free): Likewise.
4148 * calls.c (compute_argument_addresses): Likewise.
4149 (emit_library_call_value_1): Likewise.
4150 (store_one_arg): Likewise.
4151 * combine.c (combine_instructions): Likewise.
4152 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
4153 * config/arm/arm.c (arm_function_value): Likewise.
4154 (aapcs_allocate_return_reg): Likewise.
4155 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
4156 * config/i386/i386.c (construct_container): Likewise.
4157 (ix86_gimplify_va_arg): Likewise.
4158 (ix86_expand_sse_cmp): Likewise.
4159 (emit_memmov): Likewise.
4160 (emit_memset): Likewise.
4161 (expand_small_movmem_or_setmem): Likewise.
4162 (ix86_expand_pextr): Likewise.
4163 (ix86_expand_pinsr): Likewise.
4164 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
4165 * config/microblaze/microblaze.c (microblaze_block_move_straight):
4167 * config/mips/mips.c (mips_function_value_1) Likewise.
4168 (mips_block_move_straight): Likewise.
4169 (mips_expand_ins_as_unaligned_store): Likewise.
4170 * config/powerpcspe/powerpcspe.c
4171 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4172 (rs6000_darwin64_record_arg_flush): Likewise.
4173 * config/rs6000/rs6000.c
4174 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4175 (rs6000_darwin64_record_arg_flush): Likewise.
4176 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
4177 (sparc_function_value_1): Likewise.
4178 * config/spu/spu.c (adjust_operand): Likewise.
4179 (spu_emit_branch_or_set): Likewise.
4180 (arith_immediate_p): Likewise.
4181 * emit-rtl.c (gen_lowpart_common): Likewise.
4182 * expr.c (expand_expr_real_1): Likewise.
4183 * function.c (assign_parm_setup_block): Likewise.
4184 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
4185 * reload1.c (alter_reg): Likewise.
4186 * stor-layout.c (mode_for_vector): Likewise.
4187 (layout_type): Likewise.
4189 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4191 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
4192 (spu_convert_move): Likewise.
4193 * lower-subreg.c (resolve_simple_move): Likewise.
4195 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4198 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
4199 define_insn to a define_expand.
4200 (altivec_vsum2sws_direct): New define_insn.
4201 (altivec_vsumsws): Convert from a define_insn to a define_expand.
4203 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4205 * config/arm/arm.c (arm_option_params_internal): Improve setting of
4208 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
4213 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
4214 realign stack if stack alignment needed is less than incoming
4217 2017-09-05 Marek Polacek <polacek@redhat.com>
4220 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
4223 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4225 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
4227 2017-09-05 Richard Biener <rguenther@suse.de>
4229 PR tree-optimization/82084
4230 * fold-const.c (can_native_encode_string_p): Handle wide characters.
4232 2017-09-05 Richard Biener <rguenther@suse.de>
4234 PR tree-optimization/82102
4235 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
4237 2017-09-05 Martin Liska <mliska@suse.cz>
4239 PR tree-optimization/82032
4240 * tree-cfg.c (generate_range_test): New function.
4241 * tree-cfg.h (generate_range_test): Declared here.
4242 * tree-cfgcleanup.c (convert_single_case_switch): New function.
4243 (cleanup_control_expr_graph): Use it.
4244 * tree-switch-conversion.c (try_switch_expansion): Remove
4246 (emit_case_nodes): Use generate_range_test.
4248 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
4251 * config/i386/i386.md (*<btsc><mode>_mask): Add
4252 TARGET_USE_BT to insn constraint.
4253 (*btr<mode>_mask): Ditto.
4255 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
4257 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
4258 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
4260 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4263 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
4264 TARGET_NEON and TARGET_IWMMXT.
4265 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
4266 TARGET_NEON and TARGET_IWMMXT.
4267 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
4269 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
4271 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
4272 (ix86_rewrite_tls_address): Ditto.
4273 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
4274 (ix86_rewrite_tls_address_1): Ditto.
4275 (ix86_rewrite_tls_address): Ditto.
4276 * config/i386/predicates.md (tls_address_pattern): New predicate.
4277 * config/i386/i386.md (TLS address splitter): New splitter.
4279 2017-09-04 Richard Biener <rguenther@suse.de>
4281 PR tree-optimization/82084
4282 * fold-const.h (can_native_encode_string_p): Declare.
4283 * fold-const.c (can_native_encode_string_p): Factor out from ...
4284 (native_encode_string): ... here.
4285 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
4286 vectorizing stores from constants we later cannot handle.
4288 2017-09-04 Marek Polacek <polacek@redhat.com>
4291 * doc/invoke.texi: Update -Wtautological-compare documentation.
4293 2017-09-04 Jeff Law <law@redhat.com>
4295 PR tree-optimization/64910
4296 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
4297 swap the first and last operand if the last is a constant.
4299 2017-09-04 Marek Polacek <polacek@redhat.com>
4302 * convert.c (do_narrow): When sanitizing signed integer overflows,
4303 bail out for signed types.
4304 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
4306 2017-09-04 Richard Biener <rguenther@suse.de>
4308 PR tree-optimization/82060
4309 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
4310 Move devirtualization after stmt folding and before EH/AB/noreturn
4311 cleanup to get the stmt refs canonicalized. Use a bool instead
4312 of gimple_modified_p since that doesn't work for NOPs. Schedule
4313 NOPs generated by folding for removal.
4315 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4316 Alan Hayward <alan.hayward@arm.com>
4317 David Sherwood <david.sherwood@arm.com>
4319 * coretypes.h (pad_direction): New enum.
4320 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
4321 (FUNCTION_ARG_PADDING): Likewise.
4322 * target.def (function_arg_padding): New hook.
4323 * targhooks.h (default_function_arg_padding): Declare.
4324 * targhooks.c (default_function_arg_padding): New function.
4325 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
4326 (TARGET_FUNCTION_ARG_PADDING): ...this.
4327 * doc/tm.texi: Regenerate.
4328 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
4329 instead of direction.
4330 (compute_argument_addresses): Likewise.
4331 (load_register_parameters): Likewise.
4332 (emit_library_call_value_1): Likewise.
4333 (store_one_arg): Use targetm.calls.function_arg_padding instead
4334 of FUNCTION_ARG_PADDING.
4335 (must_pass_in_stack_var_size_or_pad): Likewise.
4336 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
4337 (emit_group_store): Likewise.
4338 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
4339 instead of FUNCTION_ARG_PADDING.
4340 (emit_push_insn): Likewise, and propagate enum change throughout
4342 * function.h (direction): Delete.
4343 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
4345 * function.c (assign_parm_find_stack_rtl): Likewise.
4346 (assign_parm_setup_block_p): Likewise.
4347 (assign_parm_setup_block): Likewise.
4348 (gimplify_parameters): Likewise.
4349 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
4350 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
4352 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
4353 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4354 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
4355 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
4356 (aarch64_function_arg_padding): ...this new function.
4357 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
4358 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4359 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
4360 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4361 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
4362 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4363 (arm_pad_arg_upward): Replace with...
4364 (arm_function_arg_padding): ...this new function.
4365 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
4367 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
4368 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
4369 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4370 (ia64_hpux_function_arg_padding): Replace with...
4371 (ia64_function_arg_padding): ...this new function. Use pad_direction
4372 instead of direction. Check for TARGET_HPUX.
4373 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
4374 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4375 (iq2000_function_arg_padding): New function.
4376 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
4377 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
4378 (mips_function_arg_padding): ...this new function.
4379 (mips_pad_reg_upward): Update accordingly.
4380 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4381 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
4382 targetm.calls.function_arg_padding.
4383 (FUNCTION_ARG_PADDING): Delete.
4384 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4385 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
4386 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4387 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
4388 (nios2_block_reg_padding): Return pad_direction instead of direction.
4389 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
4390 instead of direction.
4391 (nios2_function_arg_padding): Likewise. Make static.
4392 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4393 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
4394 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
4395 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
4396 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4397 (pa_function_arg_padding): Make static. Return pad_direction instead
4399 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
4400 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4401 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
4402 instead of direction. Use targetm.calls.function_arg_padding.
4403 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
4404 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4405 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
4406 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
4407 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
4409 (function_arg_padding): Rename to...
4410 (rs6000_function_arg_padding): ...this. Make static. Return
4411 pad_direction instead of direction.
4412 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4413 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
4414 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4415 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
4416 instead of direction. Use targetm.calls.function_arg_padding.
4417 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
4418 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4419 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
4420 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
4421 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4422 (function_arg_padding): Rename to...
4423 (rs6000_function_arg_padding): ...this. Make static. Return
4424 pad_direction instead of direction.
4425 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4426 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
4427 * config/s390/s390.c (s390_function_arg_padding): New function.
4428 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4429 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
4430 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
4431 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4432 (function_arg_padding): Rename to...
4433 (sparc_function_arg_padding): ...this. Make static. Return
4434 pad_direction instead of direction.
4435 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
4436 * config/spu/spu.c (spu_function_arg_padding): New function.
4437 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4438 * system.h (FUNCTION_ARG_PADDING): Poison.
4440 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4441 Alan Hayward <alan.hayward@arm.com>
4442 David Sherwood <david.sherwood@arm.com>
4444 * target.def (modes_tieable_p): New hook.
4445 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
4446 (TARGET_MODES_TIEABLE_P): ...this.
4447 * doc/tm.texi.in: Regenerate.
4448 * hooks.h (hook_bool_mode_mode_true): Declare.
4449 * hooks.c (hook_bool_mode_mode_true): New function.
4450 * combine.c (subst): Use targetm.modes_tieable_p instead of
4452 * dse.c (find_shift_sequence): Likewise.
4453 * expmed.c (extract_low_bits): Likewise.
4454 * lower-subreg.c: Include target.h.
4455 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
4457 * rtlanal.c (rtx_cost): Likewise.
4458 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
4459 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
4460 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
4461 (TARGET_MODES_TIEABLE_P): Redefine.
4462 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
4463 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
4464 (TARGET_MODES_TIEABLE_P): Redefine.
4465 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
4466 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
4467 (arc_modes_tieable_p): New function.
4468 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
4469 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
4470 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
4471 (arm_modes_tieable_p): Make static.
4472 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
4473 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
4474 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
4475 (TARGET_MODES_TIEABLE_P): Redefine.
4476 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
4477 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
4478 (TARGET_MODES_TIEABLE_P): Redefine.
4479 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
4480 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
4481 (cr16_modes_tieable_p): New function.
4482 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
4483 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
4484 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
4485 (TRULY_NOOP_TRUNCATION): Update comment.
4486 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
4487 (TRULY_NOOP_TRUNCATION): Update comment.
4488 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
4489 (frv_modes_tieable_p): New function.
4490 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
4491 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
4492 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
4493 (TARGET_MODES_TIEABLE_P): Redefine.
4494 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
4495 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
4496 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
4497 (TARGET_MODES_TIEABLE_P): Redefine.
4498 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
4499 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
4500 (ia64_modes_tieable_p): New function.
4501 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
4502 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
4503 (iq2000_modes_tieable_p): New function.
4504 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
4505 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
4506 (lm32_modes_tieable_p): New function.
4507 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
4508 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
4509 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
4510 (TARGET_MODES_TIEABLE_P): Redefine.
4511 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
4512 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
4513 (m32r_modes_tieable_p): New function.
4514 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
4515 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
4516 (m68k_modes_tieable_p): New function.
4517 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
4518 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
4519 (mcore_modes_tieable_p): New function.
4520 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
4521 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
4523 (TARGET_MODES_TIEABLE_P): Redefine.
4524 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
4525 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
4526 * config/mips/mips.c (mips_modes_tieable_p): Make static.
4527 (TARGET_MODES_TIEABLE_P): Redefine.
4528 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
4529 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
4530 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
4531 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
4532 (mn10300_modes_tieable_p): ...this and make static.
4533 (TARGET_MODES_TIEABLE_P): Redefine.
4534 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
4535 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
4536 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
4537 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
4538 (msp430_modes_tieable_p): Make static.
4539 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
4540 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
4541 (TARGET_MODES_TIEABLE_P): Redefine.
4542 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
4543 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
4544 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
4545 (TARGET_MODES_TIEABLE_P): Redefine.
4546 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
4547 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
4548 * config/pa/pa.c (pa_modes_tieable_p): Make static.
4549 (TARGET_MODES_TIEABLE_P): Redefine.
4550 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
4551 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
4552 (pdp11_modes_tieable_p): New function.
4553 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
4554 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
4555 (rs6000_modes_tieable_p): New function.
4556 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4557 * config/powerpcspe/powerpcspe.md: Update comment.
4558 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
4559 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
4560 (TARGET_MODES_TIEABLE_P): Redefine.
4561 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
4562 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
4563 (rl78_modes_tieable_p): New function.
4564 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
4565 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
4566 (rs6000_modes_tieable_p): New function.
4567 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4568 * config/rs6000/rs6000.md: Update comment.
4569 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
4570 * config/rx/rx.c (rx_modes_tieable_p): New function.
4571 (TARGET_MODES_TIEABLE_P): Redefine.
4572 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
4573 * config/s390/s390.c (s390_modes_tieable_p): New function.
4574 (TARGET_MODES_TIEABLE_P): Redefine.
4575 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
4576 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
4577 (sh_modes_tieable_p): New function.
4578 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
4579 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
4580 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
4581 (sparc_modes_tieable_p): Make static.
4582 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
4583 * config/spu/spu.c (spu_modes_tieable_p): New function.
4584 (TARGET_MODES_TIEABLE_P): Redefine.
4585 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
4586 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
4587 (TARGET_MODES_TIEABLE_P): Redefine.
4588 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
4589 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
4590 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
4591 * config/v850/v850.c (v850_modes_tieable_p): New function.
4592 (TARGET_MODES_TIEABLE_P): Redefine.
4593 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
4594 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
4595 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
4596 (visium_modes_tieable_p): New function.
4597 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
4598 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
4599 (xtensa_modes_tieable_p): New function.
4600 * system.h (MODES_TIEABLE_P): Poison.
4602 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4603 Alan Hayward <alan.hayward@arm.com>
4604 David Sherwood <david.sherwood@arm.com>
4606 * target.def (hard_regno_mode_ok): New hook.
4607 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
4608 (TARGET_HARD_REGNO_MODE_OK): ...this.
4609 * doc/tm.texi.in: Regenerate.
4610 * hooks.h (hook_bool_uint_mode_true): Declare.
4611 * hooks.c (hook_bool_uint_mode_true): New function.
4612 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
4614 * genpreds.c (write_insn_preds_c): Add an include of target.h.
4615 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
4616 instead of HARD_REGNO_MODE_OK.
4617 * caller-save.c: Include target.h.
4618 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
4620 * combine.c (can_combine_p): Likewise.
4621 (combinable_i3pat): Likewise.
4622 (can_change_dest_mode): Likewise.
4623 * expr.c (init_expr_target): Likewise.
4624 (convert_move): Likewise.
4625 (convert_modes): Likewise.
4626 * ira.c (setup_prohibited_class_mode_regs): Likewise.
4627 (setup_prohibited_mode_move_regs): Likewise.
4628 * ira.h (target_ira): Likewise.
4629 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4630 * lra-constraints.c (process_alt_operands): Likewise.
4631 (split_reg): Likewise.
4632 * recog.c (peep2_find_free_register): Likewise.
4633 * ree.c (combine_reaching_defs): Likewise.
4634 * regcprop.c (maybe_mode_change): Likewise.
4635 * reginfo.c (init_reg_sets_1): Likewise.
4636 (choose_hard_reg_mode): Likewise.
4637 (simplifiable_subregs): Likewise.
4638 * regrename.c (check_new_reg_p): Likewise.
4639 * reload.c (find_valid_class): Likewise.
4640 (find_valid_class_1): Likewise.
4641 (reload_inner_reg_of_subreg): Likewise.
4642 (push_reload): Likewise.
4643 (combine_reloads): Likewise.
4644 (find_dummy_reload): Likewise.
4645 (find_reloads): Likewise.
4646 * reload1.c (find_reg): Likewise.
4647 (set_reload_reg): Likewise.
4648 (allocate_reload_reg): Likewise.
4649 (choose_reload_regs): Likewise.
4650 (reload_adjust_reg_for_temp): Likewise.
4651 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
4652 (simplify_subreg_regno): Likewise.
4653 * sel-sched.c (init_regs_for_mode): Likewise.
4654 * varasm.c (make_decl_rtl): Likewise.
4655 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
4656 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
4658 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
4659 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
4660 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4661 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
4662 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
4663 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4664 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
4665 (arc_mode_class): Delete.
4666 (HARD_REGNO_MODE_OK): Delete.
4667 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4668 (arc_hard_regno_mode_ok): Rename old array to...
4669 (arc_hard_regno_mode_ok_modes): ...this.
4670 (arc_conditional_register_usage): Update accordingly.
4671 (arc_mode_class): Make static.
4672 (arc_hard_regno_mode_ok): New function.
4673 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
4674 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
4675 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4676 (arm_hard_regno_mode_ok): Make static.
4677 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
4679 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
4680 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
4681 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
4683 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4684 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
4685 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
4686 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
4687 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
4688 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4689 * config/bfin/predicates.md (valid_reg_operand): Use
4690 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
4691 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
4692 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
4693 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4694 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
4695 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
4696 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4697 (cr16_hard_regno_mode_ok): Make static and return a bool.
4698 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
4699 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4700 (cris_hard_regno_mode_ok): New function.
4701 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
4702 (epiphany_mode_class): Delete.
4703 (HARD_REGNO_MODE_OK): Delete.
4704 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
4705 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4706 (hard_regno_mode_ok): Rename to...
4707 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
4708 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
4709 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
4711 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
4712 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
4713 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4714 (frv_hard_regno_mode_ok): Make static and return a bool.
4715 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
4717 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
4718 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
4719 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
4720 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
4722 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4723 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
4724 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
4725 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
4727 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4728 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
4729 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4730 (ia64_hard_regno_mode_ok): New function.
4731 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
4732 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4733 (iq2000_hard_regno_mode_ok): New function.
4734 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
4735 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4736 (lm32_hard_regno_mode_ok): New function.
4737 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
4738 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
4739 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
4740 instead of HARD_REGNO_MODE_OK.
4741 (m32c_hard_regno_ok): Rename to...
4742 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
4743 (m32c_cannot_change_mode_class): Update accordingly.
4744 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4745 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
4746 (m32r_mode_class): Delete.
4747 (HARD_REGNO_MODE_OK): Delete.
4748 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4749 (m32r_hard_regno_mode_ok): Rename to...
4750 (m32r_hard_regno_modes): ...this.
4751 (m32r_mode_class): Make static.
4752 (m32r_hard_regno_mode_ok): New function.
4753 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
4754 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
4755 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4756 (m68k_hard_regno_mode_ok): Make static.
4757 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
4758 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4759 (mcore_hard_regno_mode_ok): New function.
4760 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
4761 (HARD_REGNO_MODE_OK): Delete.
4762 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
4764 (microblaze_hard_regno_mode_ok_p): ...this and make static.
4765 (microblaze_hard_regno_mode_ok): New function.
4766 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4767 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
4768 (mips_hard_regno_mode_ok): Delete.
4769 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
4770 (mips_hard_regno_mode_ok_p): ...this and make static.
4771 (mips_hard_regno_mode_ok_p): Rename to...
4772 (mips_hard_regno_mode_ok_uncached): ...this.
4773 (mips_hard_regno_mode_ok): New function.
4774 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
4775 of HARD_REGNO_MODE_OK.
4776 (mips_option_override): Update after above name changes.
4777 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4778 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
4779 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
4780 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
4781 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
4782 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4783 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
4784 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
4785 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
4786 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4787 (msp430_hard_regno_mode_ok): Make static and return a bool.
4788 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
4789 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
4790 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
4792 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4793 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
4794 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
4795 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
4796 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
4797 (PA_HARD_REGNO_MODE_OK): ...this
4798 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
4799 (PA_HARD_REGNO_MODE_OK): ...this.
4800 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4801 (pa_hard_regno_mode_ok): New function.
4802 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
4803 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4804 (pdp11_hard_regno_mode_ok): New function.
4805 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
4806 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
4808 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
4810 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4811 (rs6000_hard_regno_mode_ok): Rename to...
4812 (rs6000_hard_regno_mode_ok_uncached): ...this.
4813 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4814 (rs6000_hard_regno_mode_ok): New function.
4815 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
4816 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
4817 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
4818 (riscv_hard_regno_mode_ok): ...this and make static.
4819 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4820 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
4821 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
4822 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4823 (rl78_hard_regno_mode_ok): Make static and return bool.
4824 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
4825 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
4827 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
4828 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4829 (rs6000_hard_regno_mode_ok): Rename to...
4830 (rs6000_hard_regno_mode_ok_uncached): ...this.
4831 (rs6000_init_hard_regno_mode_ok): Update accordingly.
4832 (rs6000_hard_regno_mode_ok): New function.
4833 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
4834 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
4835 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4836 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
4837 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
4838 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
4839 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4840 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
4841 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
4842 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4843 (sh_hard_regno_mode_ok): Make static.
4844 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
4845 instead of HARD_REGNO_MODE_OK.
4846 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
4847 (sparc_mode_class): Delete.
4848 (HARD_REGNO_MODE_OK): Delete.
4849 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4850 (hard_regno_mode_classes): Make static.
4851 (sparc_mode_class): Likewise.
4852 (sparc_hard_regno_mode_ok): New function.
4853 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
4854 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
4855 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
4857 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4858 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
4859 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
4860 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
4861 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
4862 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4863 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
4864 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
4865 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4866 (visium_hard_regno_mode_ok): New function.
4867 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
4868 instead of HARD_REGNO_MODE_OK.
4869 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
4870 (HARD_REGNO_MODE_OK): Delete.
4871 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
4872 (xtensa_hard_regno_mode_ok_p): ...this and make static.
4873 (xtensa_option_override): Update accordingly.
4874 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4875 (xtensa_hard_regno_mode_ok): New function.
4876 * system.h (HARD_REGNO_MODE_OK): Poison.
4878 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4879 Alan Hayward <alan.hayward@arm.com>
4880 David Sherwood <david.sherwood@arm.com>
4882 * target.def (hard_regno_call_part_clobbered): New hook.
4883 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
4884 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
4885 * doc/tm.texi: Regenerate.
4886 * hooks.h (hook_bool_uint_mode_false): Declare.
4887 * hooks.c (hook_bool_uint_mode_false): New function.
4888 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4889 * cselib.c (cselib_process_insn): Use
4890 targetm.hard_regno_call_part_clobbered instead of
4891 HARD_REGNO_CALL_PART_CLOBBERED.
4892 * ira-conflicts.c (ira_build_conflicts): Likewise.
4893 * ira-costs.c (ira_tune_allocno_costs): Likewise.
4894 * lra-constraints.c (need_for_call_save_p): Likewise.
4895 * lra-lives.c: Include target.h.
4896 (check_pseudos_live_through_calls): Use
4897 targetm.hard_regno_call_part_clobbered instead of
4898 HARD_REGNO_CALL_PART_CLOBBERED.
4899 * regcprop.c: Include target.h.
4900 (copyprop_hardreg_forward_1): Use
4901 targetm.hard_regno_call_part_clobbered instead of
4902 HARD_REGNO_CALL_PART_CLOBBERED.
4903 * reginfo.c (choose_hard_reg_mode): Likewise.
4904 * regrename.c (check_new_reg_p): Likewise.
4905 * reload.c (find_equiv_reg): Likewise.
4906 * reload1.c (emit_reload_insns): Likewise.
4907 * sched-deps.c (deps_analyze_insn): Likewise.
4908 * sel-sched.c (init_regs_for_mode): Likewise.
4909 (mark_unavailable_hard_regs): Likewise.
4910 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
4911 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4912 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
4914 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4915 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4916 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
4918 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
4920 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4921 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4922 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
4924 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4925 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4926 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
4928 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4929 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
4931 * config/powerpcspe/powerpcspe.c
4932 (rs6000_hard_regno_call_part_clobbered): New function.
4933 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4934 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4935 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
4937 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4938 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4939 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
4941 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
4942 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
4943 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
4945 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4946 Alan Hayward <alan.hayward@arm.com>
4947 David Sherwood <david.sherwood@arm.com>
4949 * rtl.h (subreg_memory_offset): Declare.
4950 * emit-rtl.c (subreg_memory_offset): New function.
4951 * expmed.c (store_bit_field_1): Use it.
4952 * expr.c (undefined_operand_subword_p): Likewise.
4953 * simplify-rtx.c (simplify_subreg): Likewise.
4955 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
4957 PR rtl-optimization/57448
4960 * optabs.c (expand_atomic_load): Place compiler memory barriers if
4961 using atomic_load pattern.
4962 (expand_atomic_store): Likewise.
4964 2017-09-04 Jakub Jelinek <jakub@redhat.com>
4967 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
4968 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
4969 handling. Use replace_call_with_value with NULL instead of
4970 gsi_replace, unlink_stmt_vdef and release_defs.
4972 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
4975 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
4977 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4980 * rtl.h (emit_library_call_value_1): Declare.
4981 (emit_library_call): Replace declaration with a series of overloads.
4982 Remove the parameter count argument.
4983 (emit_library_call_value): Likewise.
4984 * calls.c (emit_library_call_value_1): Make global. Replace varargs
4985 with an "rtx_mode_t *".
4986 (emit_library_call_value): Delete.
4987 (emit_library_call): Likewise.
4988 * asan.c (asan_emit_stack_protection): Update calls accordingly.
4989 (asan_emit_allocas_unpoison): Likewise.
4990 * builtins.c (expand_builtin_powi): Likewise.
4991 (expand_asan_emit_allocas_unpoison): Likewise.
4992 * cfgexpand.c (expand_main_function): Likewise.
4993 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
4994 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
4995 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
4996 * config/arm/arm.c (arm_trampoline_init): Likewise.
4997 (arm_call_tls_get_addr): Likewise.
4998 (arm_expand_divmod_libfunc): Likewise.
4999 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
5000 (smulsi3_highpart): Likewise.
5001 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
5002 (c6x_expand_compare): Likewise.
5003 (c6x_expand_movmem): Likewise.
5004 * config/frv/frv.c (frv_trampoline_init): Likewise.
5005 * config/i386/i386.c (ix86_trampoline_init): Likewise.
5006 (ix86_expand_divmod_libfunc): Likewise.
5007 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
5008 (ia64_expand_compare): Likewise.
5009 (ia64_profile_hook): Likewise.
5010 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
5011 (nonlocal_goto): Likewise.
5012 (restore_stack_nonlocal): Likewise.
5013 * config/m32r/m32r.c (block_move_call): Likewise.
5014 (m32r_trampoline_init): Likewise.
5015 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
5016 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
5017 (m68k_call_m68k_read_tp): Likewise.
5018 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
5019 (microblaze_expand_divide): Likewise.
5020 * config/mips/mips.h (mips_args): Likewise.
5021 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
5022 (MIPS_ICACHE_SYNC): Likewise.
5023 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
5024 (nios2_trampoline_init): Likewise.
5025 * config/pa/pa.c (hppa_tls_call): Likewise.
5026 (pa_trampoline_init): Likewise.
5027 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
5028 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
5029 (expand_strn_compare): Likewise.
5030 (rs6000_generate_compare): Likewise.
5031 (rs6000_expand_float128_convert): Likewise.
5032 (output_profile_hook): Likewise.
5033 (rs6000_trampoline_init): Likewise.
5034 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
5035 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
5036 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
5037 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
5038 (rs6000_generate_compare): Likewise.
5039 (rs6000_expand_float128_convert): Likewise.
5040 (output_profile_hook): Likewise.
5041 (rs6000_trampoline_init): Likewise.
5042 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
5043 * config/sh/sh.c (sh_trampoline_init): Likewise.
5044 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
5045 (sparc_emit_float_lib_cmp): Likewise.
5046 (sparc32_initialize_trampoline): Likewise.
5047 (sparc64_initialize_trampoline): Likewise.
5048 (sparc_profile_hook): Likewise.
5049 * config/spu/spu.c (ea_load_store): Likewise.
5050 * config/spu/spu.md (floatunssidf2): Likewise.
5051 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
5052 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
5053 * config/visium/visium.c (expand_block_move_4): Likewise.
5054 (expand_block_move_2): Likewise.
5055 (expand_block_move_1): Likewise.
5056 (expand_block_set_4): Likewise.
5057 (expand_block_set_2): Likewise.
5058 (expand_block_set_1): Likewise.
5059 (visium_trampoline_init): Likewise.
5060 (visium_profile_hook): Likewise.
5061 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
5062 (xtensa_setup_frame_addresses): Likewise.
5063 (xtensa_trampoline_init): Likewise.
5064 * except.c (sjlj_emit_function_enter): Likewise.
5065 (sjlj_emit_function_exit): Likewise.
5066 * explow.c (allocate_dynamic_stack_space): Likewise.
5067 (probe_stack_range): Likewise.
5068 * expr.c (convert_mode_scalar): Likewise.
5069 * optabs.c (expand_binop): Likewise.
5070 (expand_twoval_binop_libfunc): Likewise.
5071 (expand_unop): Likewise.
5072 (prepare_cmp_insn): Likewise.
5073 (prepare_float_lib_cmp): Likewise.
5074 (expand_float): Likewise.
5075 (expand_fix): Likewise.
5076 (expand_fixed_convert): Likewise.
5077 (maybe_emit_sync_lock_test_and_set): Likewise.
5078 (expand_atomic_compare_and_swap): Likewise.
5079 (expand_mem_thread_fence): Likewise.
5080 (expand_atomic_fetch_op): Likewise.
5082 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
5084 * doc/generic.texi (OpenACC): Adjust URL.
5085 * doc/invoke.texi (C Dialect Options): Ditto.
5087 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
5089 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
5090 predicate for operand 1. Add (m,<S>) constraint.
5091 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
5092 Prevent memory operand 1 with register operand 2.
5094 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
5096 PR rtl-optimization/82024
5097 * combine.c (try_combine): If the combination result is a PARALLEL,
5098 and we only need to retain the SET in there that would be placed
5099 at I2, check that we can place that at I3 instead, before doing so.
5101 2017-09-01 Jakub Jelinek <jakub@redhat.com>
5104 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
5106 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
5107 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
5110 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
5111 Jeff Law <law@redhat.com>
5113 * varasm.c (bss_initializer_p): Do not put constants into .bss
5114 (categorize_decl_for_section): Handle bss_initializer_p returning
5115 false when DECL_INITIAL is NULL.
5117 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5120 * config/s390/s390.c (s390_can_inline_p): New function.
5122 2017-09-01 Jeff Law <law@redhat.com>
5124 PR tree-optimization/82052
5125 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
5126 Always initialize the returned slot after a hash table miss
5127 when INSERT is true.
5129 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
5131 * config/s390/s390.md (mem_signal_fence): Remove.
5132 * doc/md.texi (mem_signal_fence): Remove.
5133 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
5135 * target-insns.def (mem_signal_fence): Remove.
5137 2017-09-01 Jakub Jelinek <jakub@redhat.com>
5140 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
5143 * asan.c (create_odr_indicator): Strip name encoding from assembler
5144 name before appending it after __odr_asan_.
5146 2017-09-01 Martin Liska <mliska@suse.cz>
5148 PR tree-optimization/82059
5149 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
5150 frequency only when an edge is redirected.
5152 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5154 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
5155 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
5156 (arc_conditional_register_usage): Remove ARC600 lp_count
5158 (arc_file_start): Emit Tag_ARC_CPU_variation.
5159 (arc_can_use_doloop_p): New conditions to use ZOLs.
5160 (hwloop_fail): New function.
5161 (hwloop_optimize): Likewise.
5162 (hwloop_pattern_reg): Likewise.
5163 (arc_doloop_hooks): New struct, to be used with reorg_loops.
5164 (arc_reorg_loops): New function, calls reorg_loops.
5165 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
5166 (arc600_corereg_hazard): Remove ZOL checking, case handled by
5168 (arc_loop_hazard): Remove function, functionality moved into
5170 (arc_hazard): Remove arc_loop_hazard call.
5171 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
5172 into hwloop_optimize.
5173 (arc_label_align): Remove ZOL handling.
5174 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
5175 * config/arc/arc.md (doloop_begin): Remove pattern.
5176 (doloop_begin_i): Likewise.
5177 (doloop_end_i): Likewise.
5178 (doloop_fallback): Likewise.
5179 (doloop_fallback_m): Likewise.
5180 (doloop_end): Reimplement expand.
5181 (arc_lp): New pattern for LP instruction.
5182 (loop_end): New pattern.
5183 (loop_fail): Likewise.
5184 (decrement_and_branch_until_zero): Likewise.
5185 * config/arc/arc.opt (mlpc-width): New option.
5186 * doc/invoke.texi (mlpc-width): Document option.
5188 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5190 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
5191 (arc_ccfsm_advance): Fix checking for delay slots.
5192 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
5194 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5196 * config/arc/arc.md (movqi_insn): Add stores to save constant long
5198 (movhi_insn): Update store instruction constraint which are saving
5199 6-bit short immediates.
5200 (movsi_insn): Consider also short scaled load operations.
5201 (zero_extendhisi2_i): Use Usd constraint instead of T.
5202 (extendhisi2_i): Add q constraint.
5203 (arc_clzsi2): Add type and length attributes.
5204 (arc_ctzsi2): Likewise.
5205 * config/arc/constraints.md (Usc): Update constraint, the
5206 assembler can parse two relocations for a single instruction.
5208 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5210 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
5211 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
5213 2017-08-31 Olivier Hainque <hainque@adacore.com>
5215 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
5216 match as powerpc-wrs-vxworks*.
5218 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
5220 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
5221 register constraint for by-element operand.
5222 (aarch64_mls_elt_merge<mode>): Likewise.
5224 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5226 * config/arc/arc.c (arc_can_follow_jump): Check for short
5229 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5231 * config.gcc: Use g.opt for arc.
5232 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
5233 functionality moved to ...
5234 (legitimate_scaled_address_p): New function, ...here.
5235 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
5236 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
5237 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
5239 (arc_override_options): Handle G option.
5240 (arc_output_pic_addr_const): Correct function definition.
5241 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
5242 (arc_decl_anon_ns_mem_p): Delete.
5243 (arc_in_small_data_p): Overhaul this function to take into
5244 consideration the value given via G option.
5245 (arc_rewrite_small_data_1): Renamed and corrected old
5246 arc_rewrite_small_data function.
5247 (arc_rewrite_small_data): New function.
5248 (small_data_pattern): Don't use pic_offset_table_rtx.
5249 * config/arc/arc.h (CC1_SPEC): Recognize G option.
5250 * config/arc/simdext.md (movmisalignv2hi): Use
5251 prepare_move_operands function.
5253 (movmisalign*): Likewise.
5254 * doc/invoke.texi (ARC options): Document -G option.
5256 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5258 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
5260 * config/arc/arc.c (arc_print_operand): Output scalled address for
5261 sdata whenever is possible.
5262 (arc_in_small_data_p): Allow sdata for 64bit datum when double
5263 load/stores are available.
5264 (compact_sda_memory_operand): Check for the alignment required by
5265 code density instructions.
5266 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
5268 * config/arc/constraints.md (Usd): Update constraint.
5269 (Us0): New constraint.
5270 (Usc): Update constraint.
5272 2017-08-31 Richard Biener <rguenther@suse.de>
5275 * dwarf2out.c (dwarf2out_early_global_decl): Process each
5278 2017-08-31 Tamar Christina <tamar.christina@arm.com>
5280 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
5281 Resize type_signature.
5283 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5284 Alan Hayward <alan.hayward@arm.com>
5285 David Sherwood <david.sherwood@arm.com>
5287 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
5288 subregs whose inner modes can be stored in GPRs.
5289 (aarch64_classify_index): Likewise.
5291 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5292 Alan Hayward <alan.hayward@arm.com>
5293 David Sherwood <david.sherwood@arm.com>
5295 * config/aarch64/iterators.md (V_cmp_result): Rename to...
5296 (V_INT_EQUIV): ...this.
5297 (v_cmp_result): Rename to...
5298 (v_int_equiv): ...this.
5299 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
5300 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
5301 (copysign<mode>3): Likewise.
5302 (aarch64_simd_bsl<mode>_internal): Likewise.
5303 (aarch64_simd_bsl<mode>): Likewise.
5304 (vec_cmp<mode><mode>): Likewise.
5305 (vcond<mode><mode>): Likewise.
5306 (vcond<v_cmp_mixed><mode>): Likewise.
5307 (vcondu<mode><v_cmp_mixed>): Likewise.
5308 (aarch64_cm<optab><mode>): Likewise.
5309 (aarch64_cmtst<mode>): Likewise.
5310 (aarch64_fac<optab><mode>): Likewise.
5311 (vec_perm_const<mode>): Likewise.
5312 (vcond_mask_<mode><v_cmp_result>): Rename to...
5313 (vcond_mask_<mode><v_int_equiv>): ...this.
5314 (vec_cmp<mode><v_cmp_result>): Rename to...
5315 (vec_cmp<mode><v_int_equiv>): ...this.
5317 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5318 Alan Hayward <alan.hayward@arm.com>
5319 David Sherwood <david.sherwood@arm.com>
5321 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
5323 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
5324 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
5325 (UNSPEC_LD4_DREG): New unspecs.
5326 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
5327 (aarch64_ld2<mode>_dreg_be): Replace with...
5328 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
5330 (aarch64_ld3<mode>_dreg_le)
5331 (aarch64_ld3<mode>_dreg_be): Replace with...
5332 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
5334 (aarch64_ld4<mode>_dreg_le)
5335 (aarch64_ld4<mode>_dreg_be): Replace with...
5336 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
5339 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5341 PR tree-optimization/81987
5342 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
5343 insert an initializer in a location not dominated by the stride
5346 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5348 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
5349 on the entire header of the finally block in the fallthru case.
5351 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5353 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
5355 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
5357 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
5358 rs6000_emit_move_from_cr and call renamed function.
5359 (rs6000_emit_prologue): Call renamed functions.
5360 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
5361 movesi_from_cr, remove volatile CRs.
5363 2017-08-30 Jon Beniston <jon@beniston.com>
5364 Richard Biener <rguenther@suse.de>
5366 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
5367 of VECTOR_MODE_P check.
5368 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
5369 element vector types.
5371 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5373 * df.h (df_read_modify_subreg_p): Remove in favor of...
5374 * rtl.h (read_modify_subreg_p): ...this new function. Take a
5375 const_rtx instead of an rtx.
5376 * cprop.c (local_cprop_find_used_regs): Update accordingly.
5377 * df-problems.c (df_word_lr_mark_ref): Likewise.
5378 * ira-lives.c (mark_pseudo_reg_live): Likewise.
5379 (mark_pseudo_reg_dead): Likewise.
5380 (mark_ref_dead): Likewise.
5381 * reginfo.c (init_subregs_of_mode): Likewise.
5382 * sched-deps.c (sched_analyze_1): Likewise.
5383 * df-scan.c (df_def_record_1): Likewise.
5384 (df_uses_record): Likewise.
5385 (df_read_modify_subreg_p): Remove in favor of...
5386 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
5387 const_rtx instead of an rtx.
5389 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5390 Alan Hayward <alan.hayward@arm.com>
5391 David Sherwood <david.sherwood@arm.com>
5393 * rtl.h (partial_subreg_p): New function.
5394 * caller-save.c (save_call_clobbered_regs): Use it.
5395 * calls.c (expand_call): Likewise.
5396 * combine.c (combinable_i3pat): Likewise.
5397 (simplify_set): Likewise.
5398 (make_extraction): Likewise.
5399 (make_compound_operation_int): Likewise.
5400 (gen_lowpart_or_truncate): Likewise.
5401 (force_to_mode): Likewise.
5402 (make_field_assignment): Likewise.
5403 (reg_truncated_to_mode): Likewise.
5404 (record_truncated_value): Likewise.
5405 (move_deaths): Likewise.
5406 * cse.c (record_jump_cond): Likewise.
5407 (cse_insn): Likewise.
5408 * cselib.c (cselib_lookup_1): Likewise.
5409 * expmed.c (extract_bit_field_using_extv): Likewise.
5410 * function.c (assign_parm_setup_reg): Likewise.
5411 * ifcvt.c (noce_convert_multiple_sets): Likewise.
5412 * ira-build.c (create_insn_allocnos): Likewise.
5413 * lra-coalesce.c (merge_pseudos): Likewise.
5414 * lra-constraints.c (match_reload): Likewise.
5415 (simplify_operand_subreg): Likewise.
5416 (curr_insn_transform): Likewise.
5417 * lra-lives.c (process_bb_lives): Likewise.
5418 * lra.c (new_insn_reg): Likewise.
5419 (lra_substitute_pseudo): Likewise.
5420 * regcprop.c (mode_change_ok): Likewise.
5421 (maybe_mode_change): Likewise.
5422 (copyprop_hardreg_forward_1): Likewise.
5423 * reload.c (push_reload): Likewise.
5424 (find_reloads): Likewise.
5425 (find_reloads_subreg_address): Likewise.
5426 * reload1.c (alter_reg): Likewise.
5427 (eliminate_regs_1): Likewise.
5428 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5430 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
5432 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
5433 back to if statements, including unpack.
5435 2017-08-30 Martin Liska <mliska@suse.cz>
5438 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
5440 (func_checker::compare_asm_inputs_outputs): ... this function.
5441 (func_checker::compare_gimple_asm): Use the function to compare
5442 also ASM constrains.
5443 * ipa-icf-gimple.h: Rename the function.
5445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5446 Alan Hayward <alan.hayward@arm.com>
5447 David Sherwood <david.sherwood@arm.com>
5449 * coretypes.h (complex_mode): New type.
5450 * gdbhooks.py (build_pretty_printer): Handle it.
5451 * machmode.h (complex_mode): New class.
5452 (complex_mode::includes_p): New function.
5453 (is_complex_int_mode): Likewise.
5454 (is_complex_float_mode): Likewise.
5455 * genmodes.c (get_mode_class): Handle complex mode classes.
5456 * function.c (expand_function_end): Use is_complex_int_mode.
5458 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5459 Alan Hayward <alan.hayward@arm.com>
5460 David Sherwood <david.sherwood@arm.com>
5462 * coretypes.h (scalar_mode_pod): New typedef.
5463 * gdbhooks.py (build_pretty_printer): Handle it.
5464 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
5465 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
5466 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
5467 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
5470 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5471 Alan Hayward <alan.hayward@arm.com>
5472 David Sherwood <david.sherwood@arm.com>
5474 * machmode.h (mode_for_vector): Take a scalar_mode instead
5476 * stor-layout.c (mode_for_vector): Likewise.
5477 * explow.c (promote_mode): Use as_a <scalar_mode>.
5478 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
5480 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5481 Alan Hayward <alan.hayward@arm.com>
5482 David Sherwood <david.sherwood@arm.com>
5484 * target.def (preferred_simd_mode): Take a scalar_mode
5485 instead of a machine_mode.
5486 * targhooks.h (default_preferred_simd_mode): Likewise.
5487 * targhooks.c (default_preferred_simd_mode): Likewise.
5488 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
5489 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
5490 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
5491 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
5492 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
5493 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
5494 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
5495 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
5497 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
5498 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
5499 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
5500 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
5501 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
5502 * doc/tm.texi: Regenerate.
5503 * optabs-query.c (can_vec_mask_load_store_p): Return false for
5506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5507 Alan Hayward <alan.hayward@arm.com>
5508 David Sherwood <david.sherwood@arm.com>
5510 * target.def (scalar_mode_supported_p): Take a scalar_mode
5511 instead of a machine_mode.
5512 * targhooks.h (default_scalar_mode_supported_p): Likewise.
5513 * targhooks.c (default_scalar_mode_supported_p): Likewise.
5514 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
5515 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5516 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
5517 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
5518 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
5519 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
5520 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
5521 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
5522 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
5523 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
5524 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
5525 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
5527 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
5528 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
5529 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5530 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5531 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
5533 * doc/tm.texi: Regenerate.
5535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5536 Alan Hayward <alan.hayward@arm.com>
5537 David Sherwood <david.sherwood@arm.com>
5539 * coretypes.h (opt_scalar_mode): New typedef.
5540 * gdbhooks.py (build_pretty_printers): Handle it.
5541 * machmode.h (mode_iterator::get_2xwider): Add overload for
5543 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
5545 * expr.c (convert_mode_scalar): Likewise.
5546 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5547 * optabs.c (expand_float): Likewise.
5548 (expand_fix): Likewise.
5549 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5551 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5552 Alan Hayward <alan.hayward@arm.com>
5553 David Sherwood <david.sherwood@arm.com>
5555 * optabs.c (expand_float): Explicitly check for scalars before
5556 using a branching expansion.
5557 (expand_fix): Likewise.
5559 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5560 Alan Hayward <alan.hayward@arm.com>
5561 David Sherwood <david.sherwood@arm.com>
5563 * expr.c (convert_mode): Split scalar handling out into...
5564 (convert_mode_scalar): ...this new function. Treat the modes
5567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5568 Alan Hayward <alan.hayward@arm.com>
5569 David Sherwood <david.sherwood@arm.com>
5571 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
5573 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
5575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5576 Alan Hayward <alan.hayward@arm.com>
5577 David Sherwood <david.sherwood@arm.com>
5579 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
5580 rather than a machine_mode.
5581 (fixed_from_string): Likewise.
5582 (fixed_convert): Likewise.
5583 (fixed_convert_from_int): Likewise.
5584 (fixed_convert_from_real): Likewise.
5585 (real_convert_from_fixed): Likewise.
5586 * fixed-value.c (fixed_from_double_int): Likewise.
5587 (fixed_from_string): Likewise.
5588 (fixed_convert): Likewise.
5589 (fixed_convert_from_int): Likewise.
5590 (fixed_convert_from_real): Likewise.
5591 (real_convert_from_fixed): Likewise.
5592 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
5594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5595 Alan Hayward <alan.hayward@arm.com>
5596 David Sherwood <david.sherwood@arm.com>
5598 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
5599 of separate mode class checks. Do not allow vector modes here.
5600 (immed_wide_int_const): Use as_a <scalar_mode>.
5601 * explow.c (trunc_int_for_mode): Likewise.
5602 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
5603 (wi::shwi): Likewise.
5604 (wi::min_value): Likewise.
5605 (wi::max_value): Likewise.
5606 * dwarf2out.c (loc_descriptor): Likewise.
5607 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
5610 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5611 Alan Hayward <alan.hayward@arm.com>
5612 David Sherwood <david.sherwood@arm.com>
5614 * tree.h (SCALAR_TYPE_MODE): New macro.
5615 * expr.c (expand_expr_addr_expr_1): Use it.
5616 (expand_expr_real_2): Likewise.
5617 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
5618 (fold_convert_const_fixed_from_int): Likewise.
5619 (fold_convert_const_fixed_from_real): Likewise.
5620 (native_encode_fixed): Likewise
5621 (native_encode_complex): Likewise
5622 (native_encode_vector): Likewise.
5623 (native_interpret_fixed): Likewise.
5624 (native_interpret_real): Likewise.
5625 (native_interpret_complex): Likewise.
5626 (native_interpret_vector): Likewise.
5627 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
5628 (simd_clone_adjust_argument_types): Likewise.
5629 (simd_clone_init_simd_arrays): Likewise.
5630 (simd_clone_adjust): Likewise.
5631 * stor-layout.c (layout_type): Likewise.
5632 * tree.c (build_minus_one_cst): Likewise.
5633 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5634 * tree-inline.c (estimate_move_cost): Likewise.
5635 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
5636 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
5637 (vectorizable_reduction): Likewise.
5638 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
5639 (vect_recog_mixed_size_cond_pattern): Likewise.
5640 (check_bool_pattern): Likewise.
5641 (adjust_bool_pattern): Likewise.
5642 (search_type_for_mask_1): Likewise.
5643 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
5644 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5645 (vectorizable_load): Likewise.
5646 (vectorizable_store): Likewise.
5647 * ubsan.c (ubsan_encode_value): Likewise.
5648 * varasm.c (output_constant): Likewise.
5650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5651 Alan Hayward <alan.hayward@arm.com>
5652 David Sherwood <david.sherwood@arm.com>
5654 * coretypes.h (scalar_mode): New class.
5655 * machmode.h (scalar_mode): Likewise.
5656 (scalar_mode::includes_p): New function.
5657 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
5658 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
5659 * genmodes.c (get_mode_class): Handle remaining scalar modes.
5660 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
5661 * expmed.c (store_bit_field_1): Likewise.
5662 (extract_bit_field_1): Likewise.
5663 * expr.c (write_complex_part): Likewise.
5664 (read_complex_part): Likewise.
5665 (emit_move_complex_push): Likewise.
5666 (expand_expr_real_2): Likewise.
5667 * function.c (assign_parm_setup_reg): Likewise.
5668 (assign_parms_unsplit_complex): Likewise.
5669 * optabs.c (expand_binop): Likewise.
5670 * rtlanal.c (subreg_get_info): Likewise.
5671 * simplify-rtx.c (simplify_immed_subreg): Likewise.
5672 * varasm.c (output_constant_pool_2): Likewise.
5674 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5675 Alan Hayward <alan.hayward@arm.com>
5676 David Sherwood <david.sherwood@arm.com>
5678 * expmed.c (extract_high_half): Use scalar_int_mode and remove
5680 (expmed_mult_highpart_optab): Likewise.
5681 (expmed_mult_highpart): Likewise.
5683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5684 Alan Hayward <alan.hayward@arm.com>
5685 David Sherwood <david.sherwood@arm.com>
5687 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
5688 instead of a machine_mode.
5689 (builtin_memset_read_str): Likewise.
5690 * builtins.c (c_readstr): Likewise.
5691 (builtin_memcpy_read_str): Likewise.
5692 (builtin_strncpy_read_str): Likewise.
5693 (builtin_memset_read_str): Likewise.
5694 (builtin_memset_gen_str): Likewise.
5695 (expand_builtin_signbit): Use scalar_int_mode for local variables.
5696 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
5697 instead of a machine_mode.
5698 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
5700 (make_extraction): Likewise.
5701 (try_widen_shift_mode): Take and return scalar_int_modes instead
5703 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
5704 a scalar_int_mode instead of a machine_mode.
5705 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
5706 (avr_addr_space_pointer_mode): Likewise.
5707 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
5708 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
5709 (msp430_unwind_word_mode): Likewise.
5710 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
5711 (spu_addr_space_pointer_mode): Likewise.
5712 (spu_addr_space_address_mode): Likewise.
5713 (spu_libgcc_cmp_return_mode): Likewise.
5714 (spu_libgcc_shift_count_mode): Likewise.
5715 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
5716 (rl78_addr_space_pointer_mode): Likewise.
5717 (fl78_unwind_word_mode): Likewise.
5718 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
5720 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
5721 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
5722 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
5723 (mips_valid_pointer_mode): Likewise.
5724 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
5725 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
5726 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
5728 (ft32_addr_space_address_mode): Likewise.
5729 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
5730 scalar_int_mode instead of a machine_mode.
5731 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
5733 (m32c_addr_space_address_mode): Likewise.
5734 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
5735 (rs6000_eh_return_filter_mode): Likewise.
5736 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
5737 (rs6000_eh_return_filter_mode): Likewise.
5738 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
5739 (s390_libgcc_shift_count_mode): Likewise.
5740 (s390_unwind_word_mode): Likewise.
5741 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
5743 * target.def (mode_rep_extended): Likewise.
5744 (valid_pointer_mode): Likewise.
5745 (addr_space.valid_pointer_mode): Likewise.
5746 (eh_return_filter_mode): Return a scalar_int_mode rather than
5748 (libgcc_cmp_return_mode): Likewise.
5749 (libgcc_shift_count_mode): Likewise.
5750 (unwind_word_mode): Likewise.
5751 (addr_space.pointer_mode): Likewise.
5752 (addr_space.address_mode): Likewise.
5753 * doc/tm.texi: Regenerate.
5754 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
5756 (do_jump): Use scalar_int_mode for local variables.
5757 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
5758 rather than a machine_mode.
5759 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
5760 (scompare_loc_descriptor_wide): Likewise.
5761 (scompare_loc_descriptor_narrow): Likewise.
5762 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
5764 * except.c (sjlj_emit_dispatch_table): Likewise.
5765 (expand_builtin_eh_copy_values): Likewise.
5766 * explow.c (convert_memory_address_addr_space_1): Likewise.
5767 Take a scalar_int_mode rather than a machine_mode.
5768 (convert_memory_address_addr_space): Take a scalar_int_mode rather
5769 than a machine_mode.
5770 (memory_address_addr_space): Use scalar_int_mode for local variables.
5771 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
5772 rather than a machine_mode.
5773 * expmed.c (mask_rtx): Likewise.
5774 (init_expmed_one_conv): Likewise.
5775 (expand_mult_highpart_adjust): Likewise.
5776 (extract_high_half): Likewise.
5777 (expmed_mult_highpart_optab): Likewise.
5778 (expmed_mult_highpart): Likewise.
5779 (expand_smod_pow2): Likewise.
5780 (expand_sdiv_pow2): Likewise.
5781 (emit_store_flag_int): Likewise.
5782 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
5784 (extract_low_bits): Likewise.
5785 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
5787 * expr.c (pieces_addr::adjust): Likewise.
5788 (can_store_by_pieces): Likewise.
5789 (store_by_pieces): Likewise.
5790 (clear_by_pieces_1): Likewise.
5791 (expand_expr_addr_expr_1): Likewise.
5792 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
5793 (expand_expr_real_1): Likewise.
5794 (try_casesi): Likewise.
5795 * final.c (shorten_branches): Likewise.
5796 * fold-const.c (fold_convert_const_int_from_fixed): Change the
5797 type of "mode" to machine_mode.
5798 * internal-fn.c (expand_arith_overflow_result_store): Take a
5799 scalar_int_mode rather than a machine_mode.
5800 (expand_mul_overflow): Use scalar_int_mode for local variables.
5801 * loop-doloop.c (doloop_modify): Likewise.
5802 (doloop_optimize): Likewise.
5803 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
5804 than a machine_mode.
5805 (expand_doubleword_shift_condmove): Likewise.
5806 (expand_doubleword_shift): Likewise.
5807 (expand_doubleword_clz): Likewise.
5808 (expand_doubleword_popcount): Likewise.
5809 (expand_doubleword_parity): Likewise.
5810 (expand_absneg_bit): Use scalar_int_mode for local variables.
5811 (prepare_float_lib_cmp): Likewise.
5812 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
5813 rather than a machine_mode.
5814 (convert_memory_address_addr_space): Likewise.
5815 (get_mode_bounds): Likewise.
5816 (get_address_mode): Return a scalar_int_mode rather than a
5818 * rtlanal.c (get_address_mode): Likewise.
5819 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
5820 than a machine_mode.
5821 * targhooks.c (default_mode_rep_extended): Likewise.
5822 (default_valid_pointer_mode): Likewise.
5823 (default_addr_space_valid_pointer_mode): Likewise.
5824 (default_eh_return_filter_mode): Return a scalar_int_mode rather
5825 than a machine_mode.
5826 (default_libgcc_cmp_return_mode): Likewise.
5827 (default_libgcc_shift_count_mode): Likewise.
5828 (default_unwind_word_mode): Likewise.
5829 (default_addr_space_pointer_mode): Likewise.
5830 (default_addr_space_address_mode): Likewise.
5831 * targhooks.h (default_eh_return_filter_mode): Likewise.
5832 (default_libgcc_cmp_return_mode): Likewise.
5833 (default_libgcc_shift_count_mode): Likewise.
5834 (default_unwind_word_mode): Likewise.
5835 (default_addr_space_pointer_mode): Likewise.
5836 (default_addr_space_address_mode): Likewise.
5837 (default_mode_rep_extended): Take a scalar_int_mode rather than
5839 (default_valid_pointer_mode): Likewise.
5840 (default_addr_space_valid_pointer_mode): Likewise.
5841 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
5843 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
5844 rather than a machine_mode.
5845 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
5846 for local variables.
5847 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
5848 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
5849 than a machine_mode.
5851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5852 Alan Hayward <alan.hayward@arm.com>
5853 David Sherwood <david.sherwood@arm.com>
5855 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
5856 the mode argument to scalar_int_mode.
5857 (do_jump_by_parts_zero_rtx): Likewise.
5858 (do_jump_by_parts_equality_rtx): Likewise.
5859 (do_jump_by_parts_greater): Take a mode argument.
5860 (do_jump_by_parts_equality): Likewise.
5861 (do_jump_1): Update calls accordingly.
5863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5864 Alan Hayward <alan.hayward@arm.com>
5865 David Sherwood <david.sherwood@arm.com>
5867 * is-a.h (safe_dyn_cast): New function.
5868 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
5869 (jump_table_for_label): Likewise.
5870 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
5871 instead of an rtx_insn *.
5872 (shorten_branches): Use dyn_cast instead of LABEL_P and
5873 JUMP_TABLE_DATA_P. Use jump_table_for_label and
5874 rtx_jump_table_data::get_data_mode.
5875 (final_scan_insn): Likewise.
5877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5878 Alan Hayward <alan.hayward@arm.com>
5879 David Sherwood <david.sherwood@arm.com>
5881 * combine.c (try_combine): Use is_a <scalar_int_mode> when
5882 trying to combine a full-register integer set with a subreg
5885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5886 Alan Hayward <alan.hayward@arm.com>
5887 David Sherwood <david.sherwood@arm.com>
5889 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
5890 that is always either address_mode or pointer_mode.
5892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5893 Alan Hayward <alan.hayward@arm.com>
5894 David Sherwood <david.sherwood@arm.com>
5896 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
5897 when the two are known to be equal.
5899 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5900 Alan Hayward <alan.hayward@arm.com>
5901 David Sherwood <david.sherwood@arm.com>
5903 * simplify-rtx.c (simplify_const_unary_operation): Use
5904 is_a <scalar_int_mode> instead of checking for a nonzero
5905 precision. Forcibly convert op_mode to a scalar_int_mode
5906 in that case. More clearly differentiate the operand and
5907 result modes and use the former when deciding what the value
5908 of a count-bits operation should be. Use is_int_mode instead
5909 of checking for a MODE_INT. Remove redundant check for whether
5910 this mode has a zero precision.
5912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5913 Alan Hayward <alan.hayward@arm.com>
5914 David Sherwood <david.sherwood@arm.com>
5916 * optabs.c (widen_leading): Change the type of the mode argument
5917 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
5918 (widen_bswap): Likewise.
5919 (expand_parity): Likewise.
5920 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
5921 (expand_ffs): Likewise.
5922 (epand_unop): Check for scalar integer modes before calling the
5925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5926 Alan Hayward <alan.hayward@arm.com>
5927 David Sherwood <david.sherwood@arm.com>
5929 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
5931 (expand_expr_real_1): Update call accordingly.
5933 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5934 Alan Hayward <alan.hayward@arm.com>
5935 David Sherwood <david.sherwood@arm.com>
5937 * expmed.c (store_bit_field_using_insv): Add op0_mode and
5938 value_mode arguments. Use scalar_int_mode internally.
5939 (store_bit_field_1): Rename the new integer mode from imode
5940 to op0_mode and use it instead of GET_MODE (op0). Update calls
5941 to store_split_bit_field, store_bit_field_using_insv and
5942 store_fixed_bit_field.
5943 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
5944 Use scalar_int_mode internally. Use a bit count rather than a mode
5945 when calculating the largest bit size for get_best_mode.
5946 Update calls to store_split_bit_field and store_fixed_bit_field_1.
5947 (store_fixed_bit_field_1): Add mode and value_mode arguments.
5948 Remove assertion that OP0 has a scalar integer mode.
5949 (store_split_bit_field): Add op0_mode and value_mode arguments.
5950 Update calls to extract_fixed_bit_field.
5951 (extract_bit_field_using_extv): Add an op0_mode argument.
5952 Use scalar_int_mode internally.
5953 (extract_bit_field_1): Rename the new integer mode from imode to
5954 op0_mode and use it instead of GET_MODE (op0). Update calls to
5955 extract_split_bit_field, extract_bit_field_using_extv and
5956 extract_fixed_bit_field.
5957 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
5958 to extract_split_bit_field and extract_fixed_bit_field_1.
5959 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
5960 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
5962 (extract_split_bit_field): Add an op0_mode argument. Update call
5963 to extract_fixed_bit_field.
5965 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5966 Alan Hayward <alan.hayward@arm.com>
5967 David Sherwood <david.sherwood@arm.com>
5969 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
5970 * explow.c (hard_function_value): Likewise.
5971 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
5972 convert_to_mode call outside the loop.
5973 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
5974 for the mode iterator. Require the mode specified by max_pieces
5976 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
5978 (copy_blkmode_to_reg): Likewise.
5979 (set_storage_via_setmem): Likewise.
5980 * optabs.c (prepare_cmp_insn): Likewise.
5981 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
5982 * stor-layout.c (finish_bitfield_representative): Likewise.
5984 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5985 Alan Hayward <alan.hayward@arm.com>
5986 David Sherwood <david.sherwood@arm.com>
5988 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
5989 * expr.c (convert_move): Use them.
5990 (convert_modes): Likewise.
5991 (store_expr_with_bounds): Likewise.
5993 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5994 Alan Hayward <alan.hayward@arm.com>
5995 David Sherwood <david.sherwood@arm.com>
5997 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
5998 parameter for the mode of "x". Remove the "known_x", "known_mode"
5999 and "known_ret" arguments. Change the type of the mode argument
6001 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
6002 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
6003 (reg_num_sign_bit_copies_for_combine): Likewise.
6004 * rtlanal.c (nonzero_bits1): Likewise.
6005 (num_sign_bit_copies1): Likewise.
6006 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
6007 (reg_num_sign_bit_copies_general): Likewise.
6008 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
6009 (reg_nonzero_bits_general): Likewise.
6011 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6012 Alan Hayward <alan.hayward@arm.com>
6013 David Sherwood <david.sherwood@arm.com>
6015 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
6016 than in subroutines. Return 1 for non-integer modes.
6017 (cached_num_sign_bit_copies): Change the type of the mode parameter
6019 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
6020 classes. Handle CONST_INT_P first and then check whether X also
6021 has a scalar integer mode. Check the same thing for inner registers
6022 of a SUBREG and for values that are being extended or truncated.
6024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6025 Alan Hayward <alan.hayward@arm.com>
6026 David Sherwood <david.sherwood@arm.com>
6028 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
6029 in subroutines. Return the mode mask for non-integer modes.
6030 (cached_nonzero_bits): Change the type of the mode parameter
6032 (nonzero_bits1): Likewise. Remove early exit for other mode
6033 classes. Handle CONST_INT_P first and then check whether X
6034 also has a scalar integer mode.
6036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6037 Alan Hayward <alan.hayward@arm.com>
6038 David Sherwood <david.sherwood@arm.com>
6040 * expr.c (widest_int_mode_for_size): Make the comment match the code.
6041 Return a scalar_int_mode and assert that the size is greater than
6043 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
6044 (op_by_pieces_d::op_by_pieces_d): Likewise.
6045 (op_by_pieces_d::run): Likewise.
6046 (can_store_by_pieces): Likewise.
6048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6049 Alan Hayward <alan.hayward@arm.com>
6050 David Sherwood <david.sherwood@arm.com>
6052 * combine.c (extract_left_shift): Add a mode argument and update
6054 (make_compound_operation_int): Change the type of the mode parameter
6055 to scalar_int_mode and update the call to extract_left_shift.
6057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6058 Alan Hayward <alan.hayward@arm.com>
6059 David Sherwood <david.sherwood@arm.com>
6061 * combine.c (simplify_and_const_int): Change the type of the mode
6062 parameter to scalar_int_mode.
6063 (simplify_and_const_int_1): Likewise. Update recursive call.
6065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6066 Alan Hayward <alan.hayward@arm.com>
6067 David Sherwood <david.sherwood@arm.com>
6069 * combine.c (simplify_compare_const): Check that the mode is a
6070 scalar_int_mode (rather than VOIDmode) before testing its
6072 (simplify_comparison): Move COMPARISON_P handling out of the
6073 loop and restrict the latter part of the loop to scalar_int_modes.
6074 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
6075 and when considering SUBREG_REGs. Use is_int_mode instead of
6076 checking GET_MODE_CLASS against MODE_INT.
6078 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6079 Alan Hayward <alan.hayward@arm.com>
6080 David Sherwood <david.sherwood@arm.com>
6082 * combine.c (try_widen_shift_mode): Move check for equal modes to...
6083 (simplify_shift_const_1): ...here. Use scalar_int_mode for
6084 shift_unit_mode and for modes involved in scalar shifts.
6086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6087 Alan Hayward <alan.hayward@arm.com>
6088 David Sherwood <david.sherwood@arm.com>
6090 * combine.c (force_int_to_mode): New function, split out from...
6091 (force_to_mode): ...here. Keep xmode up-to-date and use it
6092 instead of GET_MODE (x).
6094 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6095 Alan Hayward <alan.hayward@arm.com>
6096 David Sherwood <david.sherwood@arm.com>
6098 * optabs-query.h (extraction_insn::struct_mode): Change type to
6099 opt_scalar_int_mode and update comment.
6100 (extraction_insn::field_mode): Change type to scalar_int_mode.
6101 (extraction_insn::pos_mode): Likewise.
6102 * combine.c (make_extraction): Update accordingly.
6103 * optabs-query.c (get_traditional_extraction_insn): Likewise.
6104 (get_optab_extraction_insn): Likewise.
6105 * recog.c (simplify_while_replacing): Likewise.
6106 * expmed.c (narrow_bit_field_mem): Change the type of the mode
6107 parameter to opt_scalar_int_mode.
6109 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6110 Alan Hayward <alan.hayward@arm.com>
6111 David Sherwood <david.sherwood@arm.com>
6113 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
6114 to a scalar_int_mode instead of a machine_mode.
6115 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
6116 (get_best_mode): Return a boolean and use a pointer argument to store
6117 the selected mode. Replace the limit mode parameter with a bit limit.
6118 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
6119 for the values returned by bit_field_mode_iterator::next_mode.
6120 (store_bit_field): Update call to get_best_mode.
6121 (store_fixed_bit_field): Likewise.
6122 (extract_fixed_bit_field): Likewise.
6123 * expr.c (optimize_bitfield_assignment_op): Likewise.
6124 * fold-const.c (optimize_bit_field_compare): Likewise.
6125 (fold_truth_andor_1): Likewise.
6126 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
6127 Update for new type of m_mode.
6128 (get_best_mode): As above.
6130 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6131 Alan Hayward <alan.hayward@arm.com>
6132 David Sherwood <david.sherwood@arm.com>
6134 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
6135 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
6136 (store_bit_field): Check is_a <scalar_int_mode> before calling
6137 strict_volatile_bitfield_p.
6138 (extract_bit_field): Likewise.
6140 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6141 Alan Hayward <alan.hayward@arm.com>
6142 David Sherwood <david.sherwood@arm.com>
6144 * target.def (cstore_mode): Return a scalar_int_mode.
6145 * doc/tm.texi: Regenerate.
6146 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
6147 * targhooks.h (default_cstore_mode): Likewise.
6148 * targhooks.c (default_cstore_mode): Likewise, using a forced
6150 * expmed.c (emit_cstore): Expect the target of the cstore to be
6153 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6154 Alan Hayward <alan.hayward@arm.com>
6155 David Sherwood <david.sherwood@arm.com>
6157 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
6159 (niter_desc): Likewise mode.
6160 (iv_analyze): Add a mode parameter.
6162 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
6163 and change its type to scalar_int_mode.
6164 * loop-iv.c: Update commentary at head of file.
6165 (iv_constant): Pass the mode paraeter before the rtx it describes
6166 and change its type to scalar_int_mode. Remove VOIDmode handling.
6167 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
6168 (iv_extend): Likewise.
6169 (shorten_into_mode): Likewise.
6170 (iv_add): Use scalar_int_mode.
6171 (iv_mult): Likewise.
6172 (iv_shift): Likewise.
6173 (canonicalize_iv_subregs): Likewise.
6174 (get_biv_step_1): Pass the outer_mode parameter before the rtx
6175 it describes and change its mode to scalar_int_mode. Also change
6176 the type of the returned inner_mode to scalar_int_mode.
6177 (get_biv_step): Likewise, turning outer_mode from a pointer
6178 into a direct parameter. Update call to get_biv_step_1.
6179 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
6180 iv_constant and get_biv_step.
6181 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
6182 and change its type to scalar_int_mode. Don't initialise iv->mode
6183 to VOIDmode and remove later checks for its still being VOIDmode.
6184 Update calls to iv_analyze_op and iv_analyze_expr. Check
6185 is_a <scalar_int_mode> when changing the mode under consideration.
6186 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
6187 Update call to iv_analyze_expr.
6188 (iv_analyze_op): Add a mode parameter. Reject subregs whose
6189 inner register is not also a scalar_int_mode. Update call to
6191 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
6192 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
6193 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
6194 separate mode class checks. Update calls to iv_analyze. Remove
6195 fix-up of VOIDmodes after iv_analyze_biv.
6196 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
6197 don't have a scalar_int_mode. Update call to biv_p.
6199 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6200 Alan Hayward <alan.hayward@arm.com>
6201 David Sherwood <david.sherwood@arm.com>
6203 * cfgexpand.c (convert_debug_memory_address): Use
6204 as_a <scalar_int_mode>.
6205 * combine.c (expand_compound_operation): Likewise.
6206 (make_extraction): Likewise.
6207 (change_zero_ext): Likewise.
6208 (simplify_comparison): Likewise.
6209 * cse.c (cse_insn): Likewise.
6210 * dwarf2out.c (minmax_loc_descriptor): Likewise.
6211 (mem_loc_descriptor): Likewise.
6212 (loc_descriptor): Likewise.
6213 * expmed.c (init_expmed_one_mode): Likewise.
6214 (synth_mult): Likewise.
6215 (emit_store_flag_1): Likewise.
6216 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
6217 of a comparison with size.
6218 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
6219 (reduce_to_bit_field_precision): Likewise.
6220 * function.c (expand_function_end): Likewise.
6221 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
6222 * loop-doloop.c (doloop_modify): Likewise.
6223 * optabs.c (expand_binop): Likewise.
6224 (expand_unop): Likewise.
6225 (expand_copysign_absneg): Likewise.
6226 (prepare_cmp_insn): Likewise.
6227 (maybe_legitimize_operand): Likewise.
6228 * recog.c (const_scalar_int_operand): Likewise.
6229 * rtlanal.c (get_address_mode): Likewise.
6230 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6231 (simplify_cond_clz_ctz): Likewise.
6232 * tree-nested.c (get_nl_goto_field): Likewise.
6233 * tree.c (build_vector_type_for_mode): Likewise.
6234 * var-tracking.c (use_narrower_mode): Likewise.
6236 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6237 Alan Hayward <alan.hayward@arm.com>
6238 David Sherwood <david.sherwood@arm.com>
6240 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
6241 * builtins.c (expand_builtin_signbit): Use it.
6242 * cfgexpand.c (expand_debug_expr): Likewise.
6243 * dojump.c (do_jump): Likewise.
6244 (do_compare_and_jump): Likewise.
6245 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
6246 * expmed.c (make_tree): Likewise.
6247 * expr.c (expand_expr_real_2): Likewise.
6248 (expand_expr_real_1): Likewise.
6249 (try_casesi): Likewise.
6250 * fold-const-call.c (fold_const_call_ss): Likewise.
6251 * fold-const.c (unextend): Likewise.
6252 (extract_muldiv_1): Likewise.
6253 (fold_single_bit_test): Likewise.
6254 (native_encode_int): Likewise.
6255 (native_encode_string): Likewise.
6256 (native_interpret_int): Likewise.
6257 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
6258 * internal-fn.c (expand_addsub_overflow): Likewise.
6259 (expand_neg_overflow): Likewise.
6260 (expand_mul_overflow): Likewise.
6261 (expand_arith_overflow): Likewise.
6262 * match.pd: Likewise.
6263 * stor-layout.c (layout_type): Likewise.
6264 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
6265 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
6266 * tree-ssanames.c (get_range_info): Likewise.
6267 * tree-switch-conversion.c (array_value_type) Likewise.
6268 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
6269 (vect_recog_divmod_pattern): Likewise.
6270 (vect_recog_mixed_size_cond_pattern): Likewise.
6271 * tree-vrp.c (extract_range_basic): Likewise.
6272 (simplify_float_conversion_using_ranges): Likewise.
6273 * tree.c (int_fits_type_p): Likewise.
6274 * ubsan.c (instrument_bool_enum_load): Likewise.
6275 * varasm.c (mergeable_string_section): Likewise.
6276 (narrowing_initializer_constant_valid_p): Likewise.
6277 (output_constant): Likewise.
6279 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6280 Alan Hayward <alan.hayward@arm.com>
6281 David Sherwood <david.sherwood@arm.com>
6283 * machmode.h (NARROWEST_INT_MODE): New macro.
6284 * expr.c (alignment_for_piecewise_move): Use it instead of
6285 GET_CLASS_NARROWEST_MODE (MODE_INT).
6286 (push_block): Likewise.
6287 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
6289 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6292 Alan Hayward <alan.hayward@arm.com>
6293 David Sherwood <david.sherwood@arm.com>
6295 * postreload.c (move2add_valid_value_p): Change the type of the
6296 mode parameter to scalar_int_mode.
6297 (move2add_use_add2_insn): Add a mode parameter and use it instead
6299 (move2add_use_add3_insn): Likewise.
6300 (reload_cse_move2add): Update accordingly.
6302 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6303 Alan Hayward <alan.hayward@arm.com>
6304 David Sherwood <david.sherwood@arm.com>
6306 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
6308 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
6309 * optabs.c (expand_unop): Likewise.
6311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6312 Alan Hayward <alan.hayward@arm.com>
6313 David Sherwood <david.sherwood@arm.com>
6315 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
6316 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
6317 (popcount_loc_descriptor): Likewise.
6318 (bswap_loc_descriptor): Likewise.
6319 (rotate_loc_descriptor): Likewise.
6320 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
6321 calling the functions above.
6323 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6324 Alan Hayward <alan.hayward@arm.com>
6325 David Sherwood <david.sherwood@arm.com>
6327 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
6329 (try_combine): Likewise.
6330 (simplify_if_then_else): Likewise.
6331 * cse.c (cse_insn): Likewise.
6332 * dwarf2out.c (mem_loc_descriptor): Likewise.
6333 * emit-rtl.c (gen_lowpart_common): Likewise.
6334 * simplify-rtx.c (simplify_truncation): Likewise.
6335 (simplify_binary_operation_1): Likewise.
6336 (simplify_const_relational_operation): Likewise.
6337 (simplify_ternary_operation): Likewise.
6338 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
6340 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6341 Alan Hayward <alan.hayward@arm.com>
6342 David Sherwood <david.sherwood@arm.com>
6344 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
6345 * reload.c (push_reload): Likewise.
6346 (find_reloads): Likewise.
6348 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6349 Alan Hayward <alan.hayward@arm.com>
6350 David Sherwood <david.sherwood@arm.com>
6352 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
6353 (make_compound_operation_int): Likewise.
6354 (change_zero_ext): Likewise.
6355 * expr.c (convert_move): Likewise.
6356 (convert_modes): Likewise.
6357 * fwprop.c (forward_propagate_subreg): Likewise.
6358 * loop-iv.c (get_biv_step_1): Likewise.
6359 * optabs.c (widen_operand): Likewise.
6360 * postreload.c (move2add_valid_value_p): Likewise.
6361 * recog.c (simplify_while_replacing): Likewise.
6362 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6363 (simplify_binary_operation_1): Likewise. Remove redundant
6364 mode equality check.
6366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6367 Alan Hayward <alan.hayward@arm.com>
6368 David Sherwood <david.sherwood@arm.com>
6370 * combine.c (combine_simplify_rtx): Add checks for
6371 is_a <scalar_int_mode>.
6372 (simplify_if_then_else): Likewise.
6373 (make_field_assignment): Likewise.
6374 (simplify_comparison): Likewise.
6375 * ifcvt.c (noce_try_bitop): Likewise.
6376 * loop-invariant.c (canonicalize_address_mult): Likewise.
6377 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6379 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6380 Alan Hayward <alan.hayward@arm.com>
6381 David Sherwood <david.sherwood@arm.com>
6383 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
6384 is_a <scalar_int_mode> instead of != BLKmode.
6386 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6387 Alan Hayward <alan.hayward@arm.com>
6388 David Sherwood <david.sherwood@arm.com>
6390 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
6391 instead of != VOIDmode.
6392 * combine.c (if_then_else_cond): Likewise.
6393 (change_zero_ext): Likewise.
6394 * dwarf2out.c (mem_loc_descriptor): Likewise.
6395 (loc_descriptor): Likewise.
6396 * rtlanal.c (canonicalize_condition): Likewise.
6397 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
6399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6400 Alan Hayward <alan.hayward@arm.com>
6401 David Sherwood <david.sherwood@arm.com>
6403 * simplify-rtx.c (simplify_binary_operation_1): Use
6404 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
6406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6407 Alan Hayward <alan.hayward@arm.com>
6408 David Sherwood <david.sherwood@arm.com>
6410 * wide-int.h (int_traits<unsigned char>) New class.
6411 (int_traits<unsigned short>) Likewise.
6412 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
6413 Use GET_MODE_UNIT_PRECISION and remove redundant test for
6415 * combine.c (set_nonzero_bits_and_sign_copies): Use
6416 is_a <scalar_int_mode>.
6417 (find_split_point): Likewise.
6418 (combine_simplify_rtx): Likewise.
6419 (simplify_logical): Likewise.
6420 (expand_compound_operation): Likewise.
6421 (expand_field_assignment): Likewise.
6422 (make_compound_operation): Likewise.
6423 (extended_count): Likewise.
6424 (change_zero_ext): Likewise.
6425 (simplify_comparison): Likewise.
6426 * dwarf2out.c (scompare_loc_descriptor): Likewise.
6427 (ucompare_loc_descriptor): Likewise.
6428 (minmax_loc_descriptor): Likewise.
6429 (mem_loc_descriptor): Likewise.
6430 (loc_descriptor): Likewise.
6431 * expmed.c (init_expmed_one_mode): Likewise.
6432 * lra-constraints.c (lra_constraint_offset): Likewise.
6433 * optabs.c (prepare_libcall_arg): Likewise.
6434 * postreload.c (move2add_note_store): Likewise.
6435 * reload.c (operands_match_p): Likewise.
6436 * rtl.h (load_extend_op): Likewise.
6437 * rtlhooks.c (gen_lowpart_general): Likewise.
6438 * simplify-rtx.c (simplify_truncation): Likewise.
6439 (simplify_unary_operation_1): Likewise.
6440 (simplify_binary_operation_1): Likewise.
6441 (simplify_const_binary_operation): Likewise.
6442 (simplify_const_relational_operation): Likewise.
6443 (simplify_subreg): Likewise.
6444 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6445 * var-tracking.c (adjust_mems): Likewise.
6446 (prepare_call_arguments): Likewise.
6448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6449 Alan Hayward <alan.hayward@arm.com>
6450 David Sherwood <david.sherwood@arm.com>
6452 * machmode.h (is_int_mode): New fuction.
6453 * combine.c (find_split_point): Use it.
6454 (combine_simplify_rtx): Likewise.
6455 (simplify_if_then_else): Likewise.
6456 (simplify_set): Likewise.
6457 (simplify_shift_const_1): Likewise.
6458 (simplify_comparison): Likewise.
6459 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
6460 * cse.c (notreg_cost): Likewise.
6461 (cse_insn): Likewise.
6462 * cselib.c (cselib_lookup_1): Likewise.
6463 * dojump.c (do_jump_1): Likewise.
6464 (do_compare_rtx_and_jump): Likewise.
6465 * dse.c (get_call_args): Likewise.
6466 * dwarf2out.c (rtl_for_decl_init): Likewise.
6467 (native_encode_initializer): Likewise.
6468 * expmed.c (emit_store_flag_1): Likewise.
6469 (emit_store_flag): Likewise.
6470 * expr.c (convert_modes): Likewise.
6471 (store_field): Likewise.
6472 (expand_expr_real_1): Likewise.
6473 * fold-const.c (fold_read_from_constant_string): Likewise.
6474 * gimple-ssa-sprintf.c (get_format_string): Likewise.
6475 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
6476 * optabs.c (expand_binop): Likewise.
6477 (expand_unop): Likewise.
6478 (expand_abs_nojump): Likewise.
6479 (expand_one_cmpl_abs_nojump): Likewise.
6480 * simplify-rtx.c (mode_signbit_p): Likewise.
6481 (val_signbit_p): Likewise.
6482 (val_signbit_known_set_p): Likewise.
6483 (val_signbit_known_clear_p): Likewise.
6484 (simplify_relational_operation_1): Likewise.
6485 * tree.c (vector_type_mode): Likewise.
6487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6488 Alan Hayward <alan.hayward@arm.com>
6489 David Sherwood <david.sherwood@arm.com>
6491 * machmode.h (smallest_mode_for_size): Fix formatting.
6492 (smallest_int_mode_for_size): New function.
6493 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
6494 instead of smallest_mode_for_size.
6495 * combine.c (make_extraction): Likewise.
6496 * config/arc/arc.c (arc_expand_movmem): Likewise.
6497 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
6498 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6499 * config/s390/s390.c (s390_expand_insv): Likewise.
6500 * config/sparc/sparc.c (assign_int_registers): Likewise.
6501 * config/spu/spu.c (spu_function_value): Likewise.
6502 (spu_function_arg): Likewise.
6503 * coverage.c (get_gcov_type): Likewise.
6504 (get_gcov_unsigned_t): Likewise.
6505 * dse.c (find_shift_sequence): Likewise.
6506 * expmed.c (store_bit_field_1): Likewise.
6507 * expr.c (convert_move): Likewise.
6508 (store_field): Likewise.
6509 * internal-fn.c (expand_arith_overflow): Likewise.
6510 * optabs-query.c (get_best_extraction_insn): Likewise.
6511 * optabs.c (expand_twoval_binop_libfunc): Likewise.
6512 * stor-layout.c (layout_type): Likewise.
6513 (initialize_sizetypes): Likewise.
6514 * targhooks.c (default_get_mask_mode): Likewise.
6515 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
6517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6518 Alan Hayward <alan.hayward@arm.com>
6519 David Sherwood <david.sherwood@arm.com>
6521 * machmode.h (opt_mode::else_blk): New function.
6522 (int_mode_for_mode): Declare.
6523 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
6524 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
6526 * cfgexpand.c (expand_debug_expr): Likewise.
6527 * combine.c (gen_lowpart_or_truncate): Likewise.
6528 (gen_lowpart_for_combine): Likewise.
6529 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
6530 * config/avr/avr.c (avr_to_int_mode): Likewise.
6531 (avr_out_plus_1): Likewise.
6532 (avr_out_plus): Likewise.
6533 (avr_out_round): Likewise.
6534 * config/i386/i386.c (ix86_split_to_parts): Likewise.
6535 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6536 (s390_expand_vcond): Likewise.
6537 * config/spu/spu.c (spu_split_immediate): Likewise.
6538 (spu_expand_mov): Likewise.
6539 * dse.c (get_stored_val): Likewise.
6540 * expmed.c (store_bit_field_1): Likewise.
6541 (convert_extracted_bit_field): Use int_mode_for_mode instead of
6543 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
6544 (extract_low_bits): Likewise.
6545 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
6546 handling rather than repeating the check.
6547 (emit_group_store): Likewise.
6548 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
6549 * optabs.c (expand_absneg_bit): Likewise.
6550 (expand_copysign_absneg): Likewise.
6551 (expand_copysign_bit): Likewise.
6552 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
6553 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
6554 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
6555 * var-tracking.c (prepare_call_arguments): Likewise.
6556 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
6557 int_mode_for_mode instead of mode_for_size.
6558 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6560 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6561 Alan Hayward <alan.hayward@arm.com>
6562 David Sherwood <david.sherwood@arm.com>
6564 * machmode.h (int_mode_for_size): New function.
6565 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
6566 instead of mode_for_size.
6567 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
6569 * combine.c (expand_field_assignment): Use int_mode_for_size
6570 instead of mode_for_size.
6571 (make_extraction): Likewise.
6572 (simplify_shift_const_1): Likewise.
6573 (simplify_comparison): Likewise.
6574 * dojump.c (do_jump): Likewise.
6575 * dwarf2out.c (mem_loc_descriptor): Likewise.
6576 * emit-rtl.c (init_derived_machine_modes): Likewise.
6577 * expmed.c (flip_storage_order): Likewise.
6578 (convert_extracted_bit_field): Likewise.
6579 * expr.c (copy_blkmode_from_reg): Likewise.
6580 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
6581 * internal-fn.c (expand_mul_overflow): Likewise.
6582 * lower-subreg.c (simple_move): Likewise.
6583 * optabs-libfuncs.c (init_optabs): Likewise.
6584 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6585 * tree.c (vector_type_mode): Likewise.
6586 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
6587 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
6588 * tree-vect-generic.c (expand_vector_parallel): Likewise.
6589 * tree-vect-stmts.c (vectorizable_load): Likewise.
6590 (vectorizable_store): Likewise.
6592 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6593 Alan Hayward <alan.hayward@arm.com>
6594 David Sherwood <david.sherwood@arm.com>
6596 * coretypes.h (pod_mode): New type.
6597 (scalar_int_mode_pod): New typedef.
6598 * machmode.h (pod_mode): New class.
6599 (int_n_data_t::m): Change type to scalar_int_mode_pod.
6600 * genmodes.c (emit_mode_int_n): Update accordingly.
6601 * lower-subreg.h (target_lower_subreg): Change type to
6602 scalar_int_mode_pod.
6603 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
6604 scalar_int_mode_pod.
6606 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6607 Alan Hayward <alan.hayward@arm.com>
6608 David Sherwood <david.sherwood@arm.com>
6610 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
6611 machine_mode to scalar_int_mode.
6612 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
6613 (rs6000_option_override_internal): Remove cast to int.
6614 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
6615 machine_mode to scalar_int_mode.
6616 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
6617 (rs6000_option_override_internal): Remove cast to int.
6618 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
6619 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
6621 * config/s390/s390.c (s390_expand_builtin): Likewise.
6622 * coretypes.h (scalar_int_mode): New type.
6623 (opt_scalar_int_mode): New typedef.
6624 * machmode.h (scalar_int_mode): New class.
6625 (scalar_int_mode::includes_p): New function.
6626 (byte_mode): Change type to scalar_int_mode.
6627 (word_mode): Likewise.
6628 (ptr_mode): Likewise.
6629 * emit-rtl.c (byte_mode): Likewise.
6630 (word_mode): Likewise.
6631 (ptr_mode): Likewise.
6632 (init_derived_machine_modes): Update accordingly.
6633 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
6634 and MODE_PARTIAL_INT.
6635 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
6636 opt_scalar_int_mode.
6638 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6639 Alan Hayward <alan.hayward@arm.com>
6640 David Sherwood <david.sherwood@arm.com>
6642 * target.def (libgcc_floating_mode_supported_p): Take a
6644 * doc/tm.texi: Regenerate.
6645 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
6647 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6648 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
6651 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6652 Alan Hayward <alan.hayward@arm.com>
6653 David Sherwood <david.sherwood@arm.com>
6655 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
6656 * doc/tm.texi: Regenerate.
6657 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
6658 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
6659 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
6660 * targhooks.h (default_floatn_mode): Likewise.
6661 * targhooks.c (default_floatn_mode): Likewise.
6662 * tree.c (build_common_tree_nodes): Update accordingly.
6664 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6665 Alan Hayward <alan.hayward@arm.com>
6666 David Sherwood <david.sherwood@arm.com>
6668 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
6669 (mode_iterator::iterate_p): Likewise.
6670 (mode_iterator::get_wider): Likewise.
6671 * expr.c (init_expr_target): Use opt_scalar_float_mode.
6673 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6674 Alan Hayward <alan.hayward@arm.com>
6675 David Sherwood <david.sherwood@arm.com>
6677 * coretypes.h (opt_scalar_float_mode): New typedef.
6678 * machmode.h (float_mode_for_size): New function.
6679 * emit-rtl.c (double_mode): Delete.
6680 (init_emit_once): Use float_mode_for_size.
6681 * stor-layout.c (layout_type): Likewise.
6682 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
6684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6685 Alan Hayward <alan.hayward@arm.com>
6686 David Sherwood <david.sherwood@arm.com>
6688 * output.h (assemble_real): Take a scalar_float_mode.
6689 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
6690 * config/arm/arm.md (consttable_4): Likewise.
6691 (consttable_8): Likewise.
6692 (consttable_16): Likewise.
6693 * config/mips/mips.md (consttable_float): Likewise.
6694 * config/s390/s390.c (s390_output_pool_entry): Likewise.
6695 * varasm.c (assemble_real): Take a scalar_float_mode.
6696 (output_constant_pool_2): Update accordingly.
6697 (output_constant): Likewise.
6699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6700 Alan Hayward <alan.hayward@arm.com>
6701 David Sherwood <david.sherwood@arm.com>
6703 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
6704 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
6705 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
6706 (native_encode_real): Likewise.
6707 (native_interpret_real): Likewise.
6708 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
6709 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6712 Alan Hayward <alan.hayward@arm.com>
6713 David Sherwood <david.sherwood@arm.com>
6715 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
6716 <scalar_float_mode>. Simplify.
6717 (gen_extend_conv_libfunc): Likewise.
6719 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6720 Alan Hayward <alan.hayward@arm.com>
6721 David Sherwood <david.sherwood@arm.com>
6723 * coretypes.h (scalar_float_mode): New type.
6724 * machmode.h (mode_traits::from_int): Use machine_mode if
6725 USE_ENUM_MODES is defined.
6726 (is_a): New function.
6728 (dyn_cast): Likewise.
6729 (scalar_float_mode): New class.
6730 (scalar_float_mode::includes_p): New function.
6731 (is_float_mode): Likewise.
6732 * gdbhooks.py (MachineModePrinter): New class.
6733 (build_pretty_printer): Use it for scalar_float_mode.
6734 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
6735 (format_helper::format_helper): Turn into a template.
6736 * genmodes.c (get_mode_class): New function.
6737 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
6738 or machine_mode if none.
6739 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
6740 as_a <scalar_float_mode>.
6741 * dwarf2out.c (mem_loc_descriptor): Likewise.
6742 (insert_float): Likewise.
6743 (add_const_value_attribute): Likewise.
6744 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6745 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
6746 (expand_unop): Update accordingly.
6747 (expand_abs_nojump): Likewise.
6748 (expand_copysign_absneg): Take a scalar_float_mode.
6749 (expand_copysign_bit): Likewise.
6750 (expand_copysign): Update accordingly.
6752 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6753 Alan Hayward <alan.hayward@arm.com>
6754 David Sherwood <david.sherwood@arm.com>
6756 * coretypes.h (opt_mode): New class.
6757 * machmode.h (opt_mode): Likewise.
6758 (opt_mode::else_void): New function.
6759 (opt_mode::require): Likewise.
6760 (opt_mode::exists): Likewise.
6761 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
6762 (GET_MODE_2XWIDER_MODE): Likewise.
6763 (mode_iterator::get_wider): Update accordingly.
6764 (mode_iterator::get_2xwider): Likewise.
6765 (mode_iterator::get_known_wider): Likewise, turning into a template.
6766 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
6767 forcing a wider mode to exist.
6768 * config/cr16/cr16.h (LONG_REG_P): Likewise.
6769 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6770 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
6771 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
6772 * lower-subreg.c (init_lower_subreg): Likewise.
6773 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
6774 on the final iteration.
6775 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
6776 a wider mode exists before asking for a move pattern.
6777 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
6778 forcing a wider mode to exist.
6779 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
6780 returning false if no such mode exists.
6781 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
6782 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
6783 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
6784 Avoid checking for a MODE_INT if we already know the mode is not a
6786 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
6787 forcing a wider mode to exist.
6788 (expmed_mult_highpart_optab): Likewise.
6789 (expmed_mult_highpart): Likewise.
6790 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
6792 * lto-streamer-in.c (lto_input_mode_table): Likewise.
6793 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
6794 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
6795 * internal-fn.c (expand_mul_overflow): Update use of
6796 GET_MODE_2XWIDER_MODE.
6797 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6798 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
6799 GET_MODE_WIDER_MODE.
6800 (convert_plusminus_to_widen): Likewise.
6801 * tree-switch-conversion.c (array_value_type): Likewise.
6802 * var-tracking.c (emit_note_insn_var_location): Likewise.
6803 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6804 Return false inside rather than outside the loop if no wider mode
6806 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
6807 and GET_MODE_2XWIDER_MODE
6808 (can_compare_p): Use else_void.
6809 * gdbhooks.py (OptMachineModePrinter): New class.
6810 (build_pretty_printer): Use it for opt_mode.
6812 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6813 Alan Hayward <alan.hayward@arm.com>
6814 David Sherwood <david.sherwood@arm.com>
6816 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
6817 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
6819 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6820 Alan Hayward <alan.hayward@arm.com>
6821 David Sherwood <david.sherwood@arm.com>
6823 * machmode.h (mode_traits): New structure.
6824 (get_narrowest_mode): New function.
6825 (mode_iterator::start): Likewise.
6826 (mode_iterator::iterate_p): Likewise.
6827 (mode_iterator::get_wider): Likewise.
6828 (mode_iterator::get_known_wider): Likewise.
6829 (mode_iterator::get_2xwider): Likewise.
6830 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
6831 (FOR_EACH_MODE): Likewise.
6832 (FOR_EACH_MODE_FROM): Likewise.
6833 (FOR_EACH_MODE_UNTIL): Likewise.
6834 (FOR_EACH_WIDER_MODE): Likewise.
6835 (FOR_EACH_2XWIDER_MODE): Likewise.
6836 * builtins.c (expand_builtin_strlen): Use new mode iterators.
6837 * combine.c (simplify_comparison): Likewise
6838 * config/i386/i386.c (type_natural_mode): Likewise.
6839 * cse.c (cse_insn): Likewise.
6840 * dse.c (find_shift_sequence): Likewise.
6841 * emit-rtl.c (init_derived_machine_modes): Likewise.
6842 (init_emit_once): Likewise.
6843 * explow.c (hard_function_value): Likewise.
6844 * expmed.c (extract_fixed_bit_field_1): Likewise.
6845 (extract_bit_field_1): Likewise.
6846 (expand_divmod): Likewise.
6847 (emit_store_flag_1): Likewise.
6848 * expr.c (init_expr_target): Likewise.
6849 (convert_move): Likewise.
6850 (alignment_for_piecewise_move): Likewise.
6851 (widest_int_mode_for_size): Likewise.
6852 (emit_block_move_via_movmem): Likewise.
6853 (copy_blkmode_to_reg): Likewise.
6854 (set_storage_via_setmem): Likewise.
6855 (compress_float_constant): Likewise.
6856 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6857 * optabs-query.c (get_best_extraction_insn): Likewise.
6858 * optabs.c (expand_binop): Likewise.
6859 (expand_twoval_unop): Likewise.
6860 (expand_twoval_binop): Likewise.
6861 (widen_leading): Likewise.
6862 (widen_bswap): Likewise.
6863 (expand_parity): Likewise.
6864 (expand_unop): Likewise.
6865 (prepare_cmp_insn): Likewise.
6866 (prepare_float_lib_cmp): Likewise.
6867 (expand_float): Likewise.
6868 (expand_fix): Likewise.
6869 (expand_sfix_optab): Likewise.
6870 * postreload.c (move2add_use_add2_insn): Likewise.
6871 * reg-stack.c (reg_to_stack): Likewise.
6872 * reginfo.c (choose_hard_reg_mode): Likewise.
6873 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6874 * stor-layout.c (mode_for_size): Likewise.
6875 (smallest_mode_for_size): Likewise.
6876 (mode_for_vector): Likewise.
6877 (finish_bitfield_representative): Likewise.
6878 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
6879 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
6880 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6881 * var-tracking.c (prepare_call_arguments): Likewise.
6883 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6884 Alan Hayward <alan.hayward@arm.com>
6885 David Sherwood <david.sherwood@arm.com>
6887 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
6888 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
6889 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
6890 * machmode.h (mode_size): Move earlier in file.
6891 (mode_precision): Likewise.
6892 (mode_inner): Likewise.
6893 (mode_nunits): Likewise.
6894 (mode_unit_size): Likewise.
6895 (unit_unit_precision): Likewise.
6896 (mode_wider): Likewise.
6897 (mode_2xwider): Likewise.
6898 (machine_mode): New class.
6899 (mode_to_bytes): New function.
6900 (mode_to_bits): Likewise.
6901 (mode_to_precision): Likewise.
6902 (mode_to_inner): Likewise.
6903 (mode_to_unit_size): Likewise.
6904 (mode_to_unit_precision): Likewise.
6905 (mode_to_nunits): Likewise.
6906 (GET_MODE_SIZE): Use mode_to_bytes.
6907 (GET_MODE_BITSIZE): Use mode_to_bits.
6908 (GET_MODE_PRECISION): Use mode_to_precision.
6909 (GET_MODE_INNER): Use mode_to_inner.
6910 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
6911 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
6912 (GET_MODE_NUNITS): Use mode_to_nunits.
6913 * system.h (ALWAYS_INLINE): New macro.
6914 * config/powerpcspe/powerpcspe-c.c
6915 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
6916 int for arg1_mode and arg2_mode.
6918 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6919 Alan Hayward <alan.hayward@arm.com>
6920 David Sherwood <david.sherwood@arm.com>
6922 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6923 Prefix mode names with E_ in case statements.
6924 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6925 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
6926 (aarch64_split_simd_move): Likewise.
6927 (aarch64_gen_storewb_pair): Likewise.
6928 (aarch64_gen_loadwb_pair): Likewise.
6929 (aarch64_gen_store_pair): Likewise.
6930 (aarch64_gen_load_pair): Likewise.
6931 (aarch64_get_condition_code_1): Likewise.
6932 (aarch64_constant_pool_reload_icode): Likewise.
6933 (get_rsqrte_type): Likewise.
6934 (get_rsqrts_type): Likewise.
6935 (get_recpe_type): Likewise.
6936 (get_recps_type): Likewise.
6937 (aarch64_gimplify_va_arg_expr): Likewise.
6938 (aarch64_simd_container_mode): Likewise.
6939 (aarch64_emit_load_exclusive): Likewise.
6940 (aarch64_emit_store_exclusive): Likewise.
6941 (aarch64_expand_compare_and_swap): Likewise.
6942 (aarch64_gen_atomic_cas): Likewise.
6943 (aarch64_emit_bic): Likewise.
6944 (aarch64_emit_atomic_swap): Likewise.
6945 (aarch64_emit_atomic_load_op): Likewise.
6946 (aarch64_evpc_trn): Likewise.
6947 (aarch64_evpc_uzp): Likewise.
6948 (aarch64_evpc_zip): Likewise.
6949 (aarch64_evpc_ext): Likewise.
6950 (aarch64_evpc_rev): Likewise.
6951 (aarch64_evpc_dup): Likewise.
6952 (aarch64_gen_ccmp_first): Likewise.
6953 (aarch64_gen_ccmp_next): Likewise.
6954 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
6955 (alpha_emit_xfloating_libcall): Likewise.
6956 (emit_insxl): Likewise.
6957 (alpha_arg_type): Likewise.
6958 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
6959 (arc_preferred_simd_mode): Likewise.
6960 (arc_secondary_reload): Likewise.
6961 (get_arc_condition_code): Likewise.
6962 (arc_print_operand): Likewise.
6963 (arc_legitimate_constant_p): Likewise.
6964 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6965 * config/arc/arc.md (casesi_load): Likewise.
6966 (casesi_compact_jump): Likewise.
6967 * config/arc/predicates.md (proper_comparison_operator): Likewise.
6968 (cc_use_register): Likewise.
6969 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
6970 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6971 (arm_init_iwmmxt_builtins): Likewise.
6972 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
6973 (neon_expand_vector_init): Likewise.
6974 (arm_attr_length_move_neon): Likewise.
6975 (maybe_get_arm_condition_code): Likewise.
6976 (arm_emit_vector_const): Likewise.
6977 (arm_preferred_simd_mode): Likewise.
6978 (arm_output_iwmmxt_tinsr): Likewise.
6979 (thumb1_output_casesi): Likewise.
6980 (thumb2_output_casesi): Likewise.
6981 (arm_emit_load_exclusive): Likewise.
6982 (arm_emit_store_exclusive): Likewise.
6983 (arm_expand_compare_and_swap): Likewise.
6984 (arm_evpc_neon_vuzp): Likewise.
6985 (arm_evpc_neon_vzip): Likewise.
6986 (arm_evpc_neon_vrev): Likewise.
6987 (arm_evpc_neon_vtrn): Likewise.
6988 (arm_evpc_neon_vext): Likewise.
6989 (arm_validize_comparison): Likewise.
6990 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
6991 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
6992 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
6993 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
6994 (c6x_preferred_simd_mode): Likewise.
6995 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
6996 (epiphany_rtx_costs): Likewise.
6997 * config/epiphany/predicates.md (proper_comparison_operator):
6999 * config/frv/frv.c (condexec_memory_operand): Likewise.
7000 (frv_emit_move): Likewise.
7001 (output_move_single): Likewise.
7002 (output_condmove_single): Likewise.
7003 (frv_hard_regno_mode_ok): Likewise.
7004 (frv_matching_accg_mode): Likewise.
7005 * config/h8300/h8300.c (split_adds_subs): Likewise.
7006 (h8300_rtx_costs): Likewise.
7007 (h8300_print_operand): Likewise.
7008 (compute_mov_length): Likewise.
7009 (output_logical_op): Likewise.
7010 (compute_logical_op_length): Likewise.
7011 (compute_logical_op_cc): Likewise.
7012 (h8300_shift_needs_scratch_p): Likewise.
7013 (output_a_shift): Likewise.
7014 (compute_a_shift_length): Likewise.
7015 (compute_a_shift_cc): Likewise.
7016 (expand_a_rotate): Likewise.
7017 (output_a_rotate): Likewise.
7018 * config/i386/i386.c (classify_argument): Likewise.
7019 (function_arg_advance_32): Likewise.
7020 (function_arg_32): Likewise.
7021 (function_arg_64): Likewise.
7022 (function_value_64): Likewise.
7023 (ix86_gimplify_va_arg): Likewise.
7024 (ix86_legitimate_constant_p): Likewise.
7025 (put_condition_code): Likewise.
7026 (split_double_mode): Likewise.
7027 (ix86_avx256_split_vector_move_misalign): Likewise.
7028 (ix86_expand_vector_logical_operator): Likewise.
7029 (ix86_split_idivmod): Likewise.
7030 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
7031 (ix86_build_const_vector): Likewise.
7032 (ix86_build_signbit_mask): Likewise.
7033 (ix86_match_ccmode): Likewise.
7034 (ix86_cc_modes_compatible): Likewise.
7035 (ix86_expand_branch): Likewise.
7036 (ix86_expand_sse_cmp): Likewise.
7037 (ix86_expand_sse_movcc): Likewise.
7038 (ix86_expand_int_sse_cmp): Likewise.
7039 (ix86_expand_vec_perm_vpermi2): Likewise.
7040 (ix86_expand_vec_perm): Likewise.
7041 (ix86_expand_sse_unpack): Likewise.
7042 (ix86_expand_int_addcc): Likewise.
7043 (ix86_split_to_parts): Likewise.
7044 (ix86_vectorize_builtin_gather): Likewise.
7045 (ix86_vectorize_builtin_scatter): Likewise.
7046 (avx_vpermilp_parallel): Likewise.
7047 (inline_memory_move_cost): Likewise.
7048 (ix86_tieable_integer_mode_p): Likewise.
7049 (x86_maybe_negate_const_int): Likewise.
7050 (ix86_expand_vector_init_duplicate): Likewise.
7051 (ix86_expand_vector_init_one_nonzero): Likewise.
7052 (ix86_expand_vector_init_one_var): Likewise.
7053 (ix86_expand_vector_init_concat): Likewise.
7054 (ix86_expand_vector_init_interleave): Likewise.
7055 (ix86_expand_vector_init_general): Likewise.
7056 (ix86_expand_vector_set): Likewise.
7057 (ix86_expand_vector_extract): Likewise.
7058 (emit_reduc_half): Likewise.
7059 (ix86_emit_i387_round): Likewise.
7060 (ix86_mangle_type): Likewise.
7061 (ix86_expand_round_sse4): Likewise.
7062 (expand_vec_perm_blend): Likewise.
7063 (canonicalize_vector_int_perm): Likewise.
7064 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
7065 (expand_vec_perm_1): Likewise.
7066 (expand_vec_perm_interleave3): Likewise.
7067 (expand_vec_perm_even_odd_pack): Likewise.
7068 (expand_vec_perm_even_odd_1): Likewise.
7069 (expand_vec_perm_broadcast_1): Likewise.
7070 (ix86_vectorize_vec_perm_const_ok): Likewise.
7071 (ix86_expand_vecop_qihi): Likewise.
7072 (ix86_expand_mul_widen_hilo): Likewise.
7073 (ix86_expand_sse2_abs): Likewise.
7074 (ix86_expand_pextr): Likewise.
7075 (ix86_expand_pinsr): Likewise.
7076 (ix86_preferred_simd_mode): Likewise.
7077 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
7078 * config/i386/sse.md (*andnot<mode>3): Likewise.
7079 (<mask_codefor><code><mode>3<mask_name>): Likewise.
7080 (*<code><mode>3): Likewise.
7081 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
7082 (ia64_expand_atomic_op): Likewise.
7083 (ia64_arg_type): Likewise.
7084 (ia64_mode_to_int): Likewise.
7085 (ia64_scalar_mode_supported_p): Likewise.
7086 (ia64_vector_mode_supported_p): Likewise.
7087 (expand_vec_perm_broadcast): Likewise.
7088 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
7089 (iq2000_function_arg_advance): Likewise.
7090 (iq2000_function_arg): Likewise.
7091 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
7092 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
7093 (m68k_libcall_value): Likewise.
7094 (m68k_function_value): Likewise.
7095 (sched_attr_op_type): Likewise.
7096 * config/mcore/mcore.c (mcore_output_move): Likewise.
7097 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
7099 (microblaze_function_arg): Likewise.
7100 * config/mips/mips.c (mips16_build_call_stub): Likewise.
7101 (mips_print_operand): Likewise.
7102 (mips_mode_ok_for_mov_fmt_p): Likewise.
7103 (mips_vector_mode_supported_p): Likewise.
7104 (mips_preferred_simd_mode): Likewise.
7105 (mips_expand_vpc_loongson_even_odd): Likewise.
7106 (mips_expand_vec_unpack): Likewise.
7107 (mips_expand_vi_broadcast): Likewise.
7108 (mips_expand_vector_init): Likewise.
7109 (mips_expand_vec_reduc): Likewise.
7110 (mips_expand_msa_cmp): Likewise.
7111 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
7112 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
7113 (cc_flags_for_mode): Likewise.
7114 * config/msp430/msp430.c (msp430_print_operand): Likewise.
7115 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
7116 (nds32_output_casesi_pc_relative): Likewise.
7117 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7118 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
7119 (nvptx_gen_unpack): Likewise.
7120 (nvptx_gen_pack): Likewise.
7121 (nvptx_gen_shuffle): Likewise.
7122 (nvptx_gen_wcast): Likewise.
7123 (nvptx_preferred_simd_mode): Likewise.
7124 * config/pa/pa.c (pa_secondary_reload): Likewise.
7125 * config/pa/predicates.md (base14_operand): Likewise.
7126 * config/powerpcspe/powerpcspe-c.c
7127 (altivec_resolve_overloaded_builtin): Likewise.
7128 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
7130 (rs6000_preferred_simd_mode): Likewise.
7131 (output_vec_const_move): Likewise.
7132 (rs6000_expand_vector_extract): Likewise.
7133 (rs6000_split_vec_extract_var): Likewise.
7134 (reg_offset_addressing_ok_p): Likewise.
7135 (rs6000_legitimate_offset_address_p): Likewise.
7136 (rs6000_legitimize_address): Likewise.
7137 (rs6000_emit_set_const): Likewise.
7138 (rs6000_const_vec): Likewise.
7139 (rs6000_emit_move): Likewise.
7140 (spe_build_register_parallel): Likewise.
7141 (rs6000_darwin64_record_arg_recurse): Likewise.
7142 (swap_selector_for_mode): Likewise.
7143 (spe_init_builtins): Likewise.
7144 (paired_init_builtins): Likewise.
7145 (altivec_init_builtins): Likewise.
7146 (do_load_for_compare): Likewise.
7147 (rs6000_generate_compare): Likewise.
7148 (rs6000_expand_float128_convert): Likewise.
7149 (emit_load_locked): Likewise.
7150 (emit_store_conditional): Likewise.
7151 (rs6000_output_function_epilogue): Likewise.
7152 (rs6000_handle_altivec_attribute): Likewise.
7153 (rs6000_function_value): Likewise.
7154 (emit_fusion_gpr_load): Likewise.
7155 (emit_fusion_p9_load): Likewise.
7156 (emit_fusion_p9_store): Likewise.
7157 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
7158 (fusion_gpr_mem_load): Likewise.
7159 (fusion_addis_mem_combo_load): Likewise.
7160 (fusion_addis_mem_combo_store): Likewise.
7161 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
7162 (fusion_gpr_mem_load): Likewise.
7163 (fusion_addis_mem_combo_load): Likewise.
7164 (fusion_addis_mem_combo_store): Likewise.
7165 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7167 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
7168 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7169 (rs6000_preferred_simd_mode): Likewise.
7170 (output_vec_const_move): Likewise.
7171 (rs6000_expand_vector_extract): Likewise.
7172 (rs6000_split_vec_extract_var): Likewise.
7173 (reg_offset_addressing_ok_p): Likewise.
7174 (rs6000_legitimate_offset_address_p): Likewise.
7175 (rs6000_legitimize_address): Likewise.
7176 (rs6000_emit_set_const): Likewise.
7177 (rs6000_const_vec): Likewise.
7178 (rs6000_emit_move): Likewise.
7179 (rs6000_darwin64_record_arg_recurse): Likewise.
7180 (swap_selector_for_mode): Likewise.
7181 (paired_init_builtins): Likewise.
7182 (altivec_init_builtins): Likewise.
7183 (rs6000_expand_float128_convert): Likewise.
7184 (emit_load_locked): Likewise.
7185 (emit_store_conditional): Likewise.
7186 (rs6000_output_function_epilogue): Likewise.
7187 (rs6000_handle_altivec_attribute): Likewise.
7188 (rs6000_function_value): Likewise.
7189 (emit_fusion_gpr_load): Likewise.
7190 (emit_fusion_p9_load): Likewise.
7191 (emit_fusion_p9_store): Likewise.
7192 * config/rx/rx.c (rx_gen_move_template): Likewise.
7193 (flags_from_mode): Likewise.
7194 * config/s390/predicates.md (s390_alc_comparison): Likewise.
7195 (s390_slb_comparison): Likewise.
7196 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
7197 (s390_vector_mode_supported_p): Likewise.
7198 (s390_cc_modes_compatible): Likewise.
7199 (s390_match_ccmode_set): Likewise.
7200 (s390_canonicalize_comparison): Likewise.
7201 (s390_emit_compare_and_swap): Likewise.
7202 (s390_branch_condition_mask): Likewise.
7203 (s390_rtx_costs): Likewise.
7204 (s390_secondary_reload): Likewise.
7205 (__SECONDARY_RELOAD_CASE): Likewise.
7206 (s390_expand_cs): Likewise.
7207 (s390_preferred_simd_mode): Likewise.
7208 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
7209 * config/sh/sh.c (sh_print_operand): Likewise.
7210 (dump_table): Likewise.
7211 (sh_secondary_reload): Likewise.
7212 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7213 * config/sh/sh.md (casesi_worker_1): Likewise.
7214 (casesi_worker_2): Likewise.
7215 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
7216 (fcc_comparison_operator): Likewise.
7217 * config/sparc/sparc.c (sparc_expand_move): Likewise.
7218 (emit_soft_tfmode_cvt): Likewise.
7219 (sparc_preferred_simd_mode): Likewise.
7220 (output_cbranch): Likewise.
7221 (sparc_print_operand): Likewise.
7222 (sparc_expand_vec_perm_bmask): Likewise.
7223 (vector_init_bshuffle): Likewise.
7224 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7225 (spu_vector_mode_supported_p): Likewise.
7226 (spu_expand_insv): Likewise.
7227 (spu_emit_branch_or_set): Likewise.
7228 (spu_handle_vector_attribute): Likewise.
7229 (spu_builtin_splats): Likewise.
7230 (spu_builtin_extract): Likewise.
7231 (spu_builtin_promote): Likewise.
7232 (spu_expand_sign_extend): Likewise.
7233 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7234 (tilegx_simd_int): Likewise.
7235 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
7236 (tilepro_simd_int): Likewise.
7237 * config/v850/v850.c (const_double_split): Likewise.
7238 (v850_print_operand): Likewise.
7239 (ep_memory_offset): Likewise.
7240 * config/vax/vax.c (vax_rtx_costs): Likewise.
7241 (vax_output_int_move): Likewise.
7242 (vax_output_int_add): Likewise.
7243 (vax_output_int_subtract): Likewise.
7244 * config/visium/predicates.md (visium_branch_operator): Likewise.
7245 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
7246 (visium_print_operand_address): Likewise.
7247 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7248 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
7249 (xtensa_expand_conditional_branch): Likewise.
7250 (xtensa_copy_incoming_a7): Likewise.
7251 (xtensa_output_literal): Likewise.
7252 * dfp.c (decimal_real_maxval): Likewise.
7253 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
7255 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7256 Alan Hayward <alan.hayward@arm.com>
7257 David Sherwood <david.sherwood@arm.com>
7259 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
7260 (mode_nunits_inline): Likewise.
7261 (mode_inner_inline): Likewise.
7262 (mode_unit_size_inline): Likewise.
7263 (mode_unit_precision_inline): Likewise.
7264 (emit_insn_modes_h): Likewise. Also emit a #define of the
7266 (emit_mode_wider): Add an E_ prefix to mode names.
7267 (emit_mode_complex): Likewise.
7268 (emit_mode_inner): Likewise.
7269 (emit_mode_adjustments): Likewise.
7270 (emit_mode_int_n): Likewise.
7271 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
7272 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
7273 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
7274 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
7275 (CRC32_BUILTIN, ENTRY): Likewise.
7276 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
7277 (aarch64_pop_regs): Likewise.
7278 (aarch64_process_components): Likewise.
7279 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
7280 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
7281 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
7282 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
7283 * config/arm/arm.c (arm_init_libfuncs): Likewise.
7284 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
7286 * config/i386/i386-builtin.def (pcmpestr): Likewise.
7287 (pcmpistr): Likewise.
7288 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
7289 * config/mmix/mmix.c (mmix_output_condition): Likewise.
7290 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
7292 * config/rl78/rl78.c (mduc_regs): Likewise.
7293 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7294 (htm_expand_builtin): Likewise.
7295 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
7296 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
7297 * config/xtensa/xtensa.c (print_operand): Likewise.
7298 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
7299 (NUM_MODE_VECTOR_INT): Likewise.
7300 * genoutput.c (null_operand): Likewise.
7301 (output_operand_data): Likewise.
7302 * genrecog.c (print_parameter_value): Likewise.
7303 * lra.c (debug_operand_data): Likewise.
7305 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7307 * dwarf2out.c (scompare_loc_descriptor_wide)
7308 (scompare_loc_descriptor_narrow): New functions, split out from...
7309 (scompare_loc_descriptor): ...here.
7310 * expmed.c (emit_store_flag_int): New function, split out from...
7311 (emit_store_flag): ...here.
7313 2017-08-30 Richard Biener <rguenther@suse.de>
7315 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
7316 (dwarf2out_early_finish): Move setting of AT_pubnames from
7317 early debug output to early finish.
7319 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7321 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
7322 and -mdata-region to the assembler.
7324 2017-08-30 Richard Biener <rguenther@suse.de>
7326 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
7328 (gen_subprogram_die): Add DW_AT_object_pointer only early.
7329 (dwarf2out_early_global_decl): Only generate a DIE for the
7330 abstract origin if it doesn't already exist or is a declaration DIE.
7331 (resolve_addr): Do not add the linkage name twice when
7332 generating a stub DIE for the DW_TAG_GNU_call_site target.
7334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7336 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7337 Use machine_mode rather than int for arg1_mode.
7339 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
7342 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
7343 that the second argument of the built-in functions to unpack
7344 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
7345 switch statement instead a lot of if statements.
7346 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
7347 Allow 64-bit values to be in Altivec registers as well as
7348 traditional floating point registers.
7349 (pack<mode>, FMOVE128_VSX iterator): Likewise.
7351 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
7353 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
7354 MAX_REGS_PER_ADDRESS == 1.
7356 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
7358 * config/i386/i386.opt (flag_fentry): Do not init to -1.
7359 * config/i386/i386.c (ix86_option_override_internal): Simplify
7360 setting of opts->x_flag_entry.
7362 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7363 Jakub Jelinek <jakub@redhat.com>
7364 Richard Biener <rguenther@suse.de>
7366 PR tree-optimization/81503
7367 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
7368 folded constant fits in the target type; reorder tests for clarity.
7370 2017-08-29 Martin Liska <mliska@suse.cz>
7372 * passes.def: Include pass_lower_switch.
7373 * stmt.c (dump_case_nodes): Remove and move to
7374 tree-switch-conversion.
7375 (case_values_threshold): Likewise.
7376 (expand_switch_as_decision_tree_p): Likewise.
7377 (emit_case_decision_tree): Likewise.
7378 (expand_case): Likewise.
7379 (balance_case_nodes): Likewise.
7380 (node_has_low_bound): Likewise.
7381 (node_has_high_bound): Likewise.
7382 (node_is_bounded): Likewise.
7383 (emit_case_nodes): Likewise.
7384 (struct simple_case_node): New struct.
7385 (add_case_node): Remove.
7386 (emit_case_dispatch_table): Use vector instead of case_list.
7387 (reset_out_edges_aux): Remove.
7388 (compute_cases_per_edge): Likewise.
7389 (expand_case): Build list of simple_case_node.
7390 (expand_sjlj_dispatch_table): Use it.
7391 * tree-switch-conversion.c (struct case_node): Moved from
7392 stmt.c and adjusted.
7393 (emit_case_nodes): Likewise.
7394 (node_has_low_bound): Likewise.
7395 (node_has_high_bound): Likewise.
7396 (node_is_bounded): Likewise.
7397 (case_values_threshold): Likewise.
7398 (reset_out_edges_aux): Likewise.
7399 (compute_cases_per_edge): Likewise.
7400 (add_case_node): Likewise.
7401 (dump_case_nodes): Likewise.
7402 (balance_case_nodes): Likewise.
7403 (expand_switch_as_decision_tree_p): Likewise.
7404 (emit_jump): Likewise.
7405 (emit_case_decision_tree): Likewise.
7406 (try_switch_expansion): Likewise.
7407 (do_jump_if_equal): Likewise.
7408 (emit_cmp_and_jump_insns): Likewise.
7409 (fix_phi_operands_for_edge): New function.
7410 (record_phi_operand_mapping): Likewise.
7411 (class pass_lower_switch): New pass.
7412 (pass_lower_switch::execute): New function.
7413 (make_pass_lower_switch): Likewise.
7414 (conditional_probability):
7415 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
7416 * tree-pass.h: Add make_pass_lower_switch.
7418 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7421 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
7424 2017-08-29 Richard Biener <rguenther@suse.de>
7426 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
7427 we do not add a DW_AT_inline attribute twice.
7428 (gen_subprogram_die): Remove code setting DW_AT_inline on
7429 DECL_ABSTRACT_P nodes.
7431 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
7433 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
7434 calls to internal functions.
7435 (gimplify_modify_expr): Likewise.
7436 * tree-call-cdce.c (use_internal_fn): Likewise.
7437 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
7438 (convert_to_divmod): Set the nothrow flag.
7439 * tree-if-conv.c (predicate_mem_writes): Likewise.
7440 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7441 (vectorizable_call): Likewise.
7442 (vectorizable_store): Likewise.
7443 (vectorizable_load): Likewise.
7444 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
7445 (vect_recog_mask_conversion_pattern): Likewise.
7447 2017-08-29 Martin Liska <mliska@suse.cz>
7450 * gcc.c (driver_handle_option): Add new argument.
7451 * opts-common.c (handle_option): Pass
7452 target_option_override_hook.
7453 * opts-global.c (lang_handle_option): Add new option.
7454 (set_default_handlers): Add new argument.
7455 (decode_options): Likewise.
7456 * opts.c (target_handle_option): Likewise.
7457 (common_handle_option): Call target_option_override_hook.
7458 * opts.h (struct cl_option_handler_func): Add hook for
7459 target option override.
7460 (struct cl_option_handlers): Likewise.
7461 (set_default_handlers): Add new argument.
7462 (decode_options): Likewise.
7463 (common_handle_option): Likewise.
7464 (target_handle_option): Likewise.
7465 * toplev.c (toplev::main): Pass targetm.target_option.override
7468 2017-08-29 Richard Biener <rguenther@suse.de>
7469 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7471 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
7472 life to the active subtree.
7474 2017-08-28 Jeff Law <law@redhat.com>
7476 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
7477 derive_equivalences.
7478 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
7480 (edge_info::derive_equivalences): New private member function
7482 * tree-ssa-dom.c (class edge_info): Changed from a struct
7483 to a class. Add ctor/dtor, methods and data members.
7484 (edge_info::edge_info): Renamed from allocate_edge_info.
7485 Initialize additional members.
7486 (edge_info::~edge_info): New.
7487 (free_dom_edge_info): Delete the edge info.
7488 (record_edge_info): Use new class & associated member functions.
7489 Tighten forms for testing for edge equivalences.
7490 (record_temporary_equivalences): Iterate over the simple
7491 equivalences rather than assuming there's only one per edge.
7492 (cprop_into_successor_phis): Iterate over the simple
7493 equivalences rather than assuming there's only one per edge.
7494 (optimize_stmt): Use operand_equal_p rather than pointer
7495 equality for mini-DSE code.
7497 2017-08-28 Nathan Sidwell <nathan@acm.org>
7499 * gcc.c (execute): Fold SIGPIPE handling into switch
7500 statement. Adjust internal error message.
7502 2017-08-28 Richard Biener <rguenther@suse.de>
7505 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
7506 Do nothing for removed DIEs.
7508 2017-08-28 Richard Biener <rguenther@suse.de>
7510 PR tree-optimization/81977
7511 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
7514 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
7517 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
7518 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
7519 using targetm.gen_mem_thread_fence.
7521 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
7524 * config/i386/i386.md (*<btsc><mode>): Change operand 2
7525 predicate to register_operand. Reorder operands.
7526 (*btr<mode>): Ditto.
7527 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
7528 (*btr<mode>_mask): Ditto.
7530 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
7532 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
7533 * config/rs6000/xmmintrin.h: New file.
7534 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
7536 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7539 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
7540 parameter and_insn and return it.
7541 (recombine_lvx_pattern): Insert a copy to ensure availability of
7542 the base register of the copied masking operation at the point of
7543 the instruction replacement.
7544 (recombine_stvx_pattern): Likewise.
7546 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
7548 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
7549 undocumented switches.
7550 (-mpower9-dform-vector): Likewise.
7551 (-mpower9-dform): Likewise.
7552 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
7553 comments to delete references to -mpower9-dform* switches.
7554 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
7555 Delete reference to -mpower9-dform* switches, test for
7556 -mpower9-vector instead.
7557 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
7558 (OTHER_P9_VECTOR_MASKS): Likewise.
7559 (POWERPC_MASKS): Likewise.
7560 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
7561 tests against -mpower9-dform* to -mpower9-vector. Delete code
7562 that checked for -mpower9-dform* consistancy with other options.
7563 Add test for -mpower9-misc to enable other power9 switches.
7564 (rs6000_init_hard_regno_mode_ok): Likewise.
7565 (rs6000_option_override_internal): Likewise.
7566 (rs6000_emit_prologue): Likewise.
7567 (rs6000_emit_epilogue): Likewise.
7568 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
7569 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
7570 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
7572 (emit_fusion_p9_store): Likewise.
7573 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
7574 resetting these macros if the assembler does not support ISA 3.0
7576 (TARGET_P9_DFORM_VECTOR): Likewise.
7577 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
7578 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
7580 2017-08-25 Alan Modra <amodra@gmail.com>
7582 PR rtl-optimization/81747
7583 * cse.c (cse_extended_basic_block): Don't attempt to record
7584 equivalences for degenerate conditional branches that branch
7585 to their fall-through.
7587 2017-08-24 Martin Sebor <msebor@redhat.com>
7590 * gimple-fold.c (size_must_be_zero_p): New function.
7591 (gimple_fold_builtin_memory_op): Call it.
7593 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
7595 * config/rs6000/mm_malloc.h: New file.
7597 2017-08-24 Bin Cheng <bin.cheng@arm.com>
7599 PR tree-optimization/81913
7600 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
7601 analysis when either IVs in condition can wrap.
7603 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
7605 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
7606 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
7608 2017-08-24 Richard Biener <rguenther@suse.de>
7611 * targhooks.c (default_target_can_inline_p): Properly
7612 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
7613 is present and always compare.
7614 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
7615 infer -mfpmath=sse from TARGET_SSE_P.
7616 (ix86_can_inline_p): Properly use target_option_default_node when
7617 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
7619 2017-08-24 Richard Biener <rguenther@suse.de>
7622 * dwarf2out.c (output_die): Handle flag_generate_offload like
7624 (output_comp_unit): Likewise.
7625 (gen_array_type_die): Likewise.
7626 (dwarf2out_early_finish): Likewise.
7627 (note_variable_value_in_expr): Likewise.
7628 (dwarf2out_finish): Likewise. Adjust assert.
7629 * cgraphunit.c (symbol_table::compile): Move setting of
7630 flag_generate_offload earlier ...
7631 (symbol_table::finalize_compilation_unit): ... here, before
7632 early debug finalization.
7634 2017-08-24 Richard Biener <rguenther@suse.de>
7636 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
7637 and ipa-fnsummary.h.
7638 (ix86_can_inline_p): When ix86_fpmath flags do not match
7639 check whether the callee uses FP math at all.
7641 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
7644 * tree-ssanames.c (get_nonzero_bits): Use element_precision
7645 instead of TYPE_PRECISION.
7647 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
7648 Alan Hayward <alan.hayward@arm.com>
7649 David Sherwood <david.sherwood@arm.com>
7651 * combine.c (make_extraction): Use subreg_offset_from_lsb.
7653 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
7655 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
7657 (ix86_frame::stack_realign_allocate): New field.
7658 (struct machine_frame_state): Modify comments.
7659 (machine_frame_state::sp_realigned_fp_end): New field.
7660 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
7662 (sp_valid_at): Add assertion to assure no attempt to access invalid
7663 offset of a realigned stack.
7664 (fp_valid_at): Likewise.
7665 (choose_baseaddr): Modify comments.
7666 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
7667 ix86_expand_prologue.
7668 (ix86_expand_prologue): Modify stack realignment and allocation.
7669 (ix86_expand_epilogue): Modify comments.
7670 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
7671 avx2_runtime, avx512f, and avx512f_runtime.
7673 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
7675 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
7676 (mstackrealign): Do not init to -1.
7677 * config/i386/i386.c (ix86_option_override_internal):
7678 Check opts_set, not opts when setting default value of
7679 opts->x_ix86_force_align_arg_pointer.
7681 2017-08-23 Richard Biener <rguenther@suse.de>
7683 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
7684 lang_hooks.decl_printable_name.
7685 * print-rtl-function.c (print_rtx_function): Likewise.
7686 * tree-pretty-print.c (dump_function_header): Likewise.
7688 2017-08-23 Richard Biener <rguenther@suse.de>
7691 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
7692 -g0 at compile-time.
7694 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7697 * doc/sourcebuild.texi (Other hardware attributes):
7700 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7703 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
7704 Fix single-use check.
7706 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
7708 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
7710 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7712 * doc/install.texi: Modify to add more details on running selected
7715 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
7717 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
7718 is combined with -mabi=ms.
7719 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
7722 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7724 PR tree-optimization/81488
7725 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
7726 and cached_basis fields.
7727 (MAX_SPREAD): New constant.
7728 (alloc_cand_and_find_basis): Initialize new fields.
7729 (clear_visited): New function.
7730 (create_phi_basis_1): Rename from create_phi_basis, set visited
7731 and cached_basis fields.
7732 (create_phi_basis): New wrapper function.
7733 (phi_add_costs_1): Rename from phi_add_costs, add spread
7734 parameter, set visited field, short-circuit when limits reached.
7735 (phi_add_costs): New wrapper function.
7736 (record_phi_increments_1): Rename from record_phi_increments, set
7738 (record_phi_increments): New wrapper function.
7739 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
7740 (phi_incr_cost): New wrapper function.
7741 (all_phi_incrs_profitable_1): Rename from
7742 all_phi_incrs_profitable, set visited field.
7743 (all_phi_incrs_profitable): New wrapper function.
7745 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7746 Alan Hayward <alan.hayward@arm.com>
7747 David Sherwood <david.sherwood@arm.com>
7749 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
7750 that takes the outer and inner modes.
7751 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
7752 comparison as the canonical test for a paradoxical subreg.
7753 * combine.c (simplify_set): Use paradoxical_subreg_p.
7754 (make_extraction): Likewise.
7755 (force_to_mode): Likewise.
7756 (rtx_equal_for_field_assignment_p): Likewise.
7757 (gen_lowpart_for_combine): Likewise.
7758 (simplify_comparison): Likewise.
7759 * cse.c (equiv_constant): Likewise.
7760 * expmed.c (store_bit_field_1): Likewise.
7761 * final.c (alter_subreg): Likewise.
7762 * fwprop.c (propagate_rtx): Likewise.
7763 (forward_propagate_subreg): Likewise.
7764 * ira-conflicts.c (ira_build_conflicts): Likewise.
7765 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
7766 * lra-constraints.c (curr_insn_transform): Likewise.
7767 (split_reg): Likewise.
7768 * lra-eliminations.c (move_plus_up): Likewise.
7769 (lra_eliminate_regs_1): Likewise.
7770 * recog.c (general_operand): Likewise.
7771 * ree.c (combine_reaching_defs): Likewise.
7772 * reload.c (push_reload): Likewise.
7773 (find_reloads): Likewise.
7774 * reload1.c (elimination_effects): Likewise.
7775 (compute_reload_subreg_offset): Likewise.
7776 (choose_reload_regs): Likewise.
7777 * rtlanal.c (subreg_lsb_1): Likewise.
7778 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7779 (simplify_subreg): Likewise.
7780 * var-tracking.c (track_loc_p): Likewise.
7781 * emit-rtl.c (byte_lowpart_offset): Likewise.
7782 (paradoxical_subreg_p): Delete out-of-line definition.
7784 2017-08-22 Jeff Law <law@redhat.com>
7786 PR tree-optimization/81741
7787 PR tree-optimization/71947
7788 * tree-ssa-dom.c: Include tree-inline.h.
7789 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
7790 equivalences if one is more expensive to compute than the other.
7791 * tree-ssa-scopedtables.h (class const_or_copies): Make
7792 record_const_or_copy_raw method private.
7793 (class avail_exprs_stack): New method simplify_binary_operation.
7794 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
7795 avail_exprs_stack::simplify_binary_operation as needed.
7796 (avail_exprs_stack::simplify_binary_operation): New function.
7798 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
7800 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
7801 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
7802 (DOT_SYMBOLS): Likewise.
7803 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
7804 (RELOCATABLE_NEEDS_FIXUP): Likewise.
7805 (RS6000_ABI_NAME): Likewise.
7806 (TARGET_CMODEL): Likewise.
7807 (TOC_SECTION_ASM_OP): Likewise.
7808 (SET_CMODEL): New macro.
7809 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
7811 2017-08-22 Richard Biener <rguenther@suse.de>
7813 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
7814 to allow for free-lang-data replacements similar to verify_type_variant.
7816 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
7817 Alan Hayward <alan.hayward@arm.com>
7818 David Sherwood <david.sherwood@arm.com>
7820 * config/aarch64/aarch64.md (casesi): Use DImode rather than
7821 VOIDmode for the LABEL_REF.
7823 2017-08-22 Richard Biener <rguenther@suse.de>
7825 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
7827 2017-08-22 Richard Biener <rguenther@suse.de>
7829 * common.opt (feliminate-dwarf2-dups): Ignore.
7830 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
7831 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
7832 same_die_p_wrap, compute_section_prefix,
7833 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
7834 (comdat_symbol_id, comdat_symbol_number): Likewise.
7835 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
7837 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
7838 (output_die): Mark unreachable path unreachable.
7839 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
7840 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
7841 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
7842 (dwarf2out_early_finish): Likewise.
7844 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
7846 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
7848 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
7851 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
7852 not VAR_P. Filter attribute warnings with OPT_Wattributes.
7853 (avr_attribute_table) <io, io_low, address>: Initialize
7854 .decl_required with true.
7856 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
7858 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
7859 undocumented debugging options.
7860 (-mvsx-scalar-double): Likewise.
7861 (-mallow-df-permute): Likewise.
7862 (-mvectorize-builtins): Likewise.
7863 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7864 (rs6000_builtin_vectorized_function): Likewise.
7865 (rs6000_builtin_md_vectorized_function): Likewise.
7866 (rs6000_opt_vars): Likewise.
7868 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
7871 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
7872 (*btrq_imm): Rename from *btrq.
7873 (*btcq_imm): Rename from *btcq.
7874 (btsc): New code attribute.
7875 (*<btsc><mode>): New insn pattern.
7876 (*btr<mode>): Ditto.
7877 (*<btsc><mode>_mask): New insn_and_split pattern.
7878 (*btr<mode>_mask): Ditto.
7880 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7881 Alan Hayward <alan.hayward@arm.com>
7882 David Sherwood <david.sherwood@arm.com>
7884 * function.c (pad_below): Simplify padding calculation.
7886 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7887 Alan Hayward <alan.hayward@arm.com>
7888 David Sherwood <david.sherwood@arm.com>
7890 * target.def (function_prologue): Remove frame size argument.
7891 (function_epilogue): Likewise.
7892 * doc/tm.texi: Regenerate.
7893 * final.c (final_start_function): Update call to function_prologue.
7894 (final_end_function): Update call to function_epilogue.
7895 (default_function_pro_epilogue): Remove frame size argument.
7896 * output.h (default_function_pro_epilogue): Likewise.
7897 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
7898 (arm_output_function_prologue): Likewise.
7899 * config/frv/frv.c (frv_function_prologue): Likewise.
7900 (frv_function_epilogue): Likewise.
7901 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
7902 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
7903 (ia64_output_function_epilogue): Likewise.
7904 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
7905 (m32r_output_function_epilogue): Likewise.
7906 * config/microblaze/microblaze.c (microblaze_function_prologue)
7907 (microblaze_function_epilogue): Likewise.
7908 * config/mips/mips.c (mips_output_function_prologue): Likewise.
7909 (mips_output_function_epilogue): Likewise.
7910 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
7911 (mmix_target_asm_function_epilogue): Likewise.
7912 * config/msp430/msp430.c (msp430_start_function): Likewise.
7913 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
7914 (nds32_asm_function_epilogue): Likewise.
7915 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
7916 * config/pa/pa.c (pa_output_function_prologue): Likewise.
7917 (pa_output_function_epilogue): Likewise.
7918 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
7919 (rs6000_output_function_epilogue): Likewise.
7920 * config/rl78/rl78.c (rl78_start_function): Likewise.
7921 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
7922 (rs6000_output_function_epilogue): Likewise.
7923 * config/rx/rx.c (rx_output_function_prologue): Likewise.
7924 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
7925 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
7926 (sparc_asm_function_epilogue): Likewise.
7928 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7930 * tree.h (type_has_mode_precision_p): New function.
7931 * convert.c (convert_to_integer_1): Use it.
7932 * expr.c (expand_expr_real_2): Likewise.
7933 (expand_expr_real_1): Likewise.
7934 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
7935 * match.pd: Likewise.
7936 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
7937 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
7938 * tree-tailcall.c (process_assignment): Likewise.
7939 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7940 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
7941 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
7942 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7943 (vectorizable_assignment): Likewise.
7944 (vectorizable_shift): Likewise.
7945 (vectorizable_operation): Likewise.
7946 * tree-vrp.c (register_edge_assert_for_2): Likewise.
7948 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
7950 * match.pd: Add pow (C, x) simplification.
7952 2017-08-21 Richard Biener <rguenther@suse.de>
7954 PR tree-optimization/81900
7955 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
7956 for blocks with abnormal predecessors.
7957 (compute_antic): Do not set visited flag prematurely.
7959 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
7962 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
7964 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
7966 * stor-layout.h (vector_type_mode): Move to...
7967 * tree.h (vector_type_mode): ...here.
7968 * stor-layout.c (vector_type_mode): Move to...
7969 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
7971 2017-08-21 Richard Biener <rguenther@suse.de>
7973 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
7974 register_external_die hooks.
7975 (debug_false_tree_charstarstar_uhwistar): Declare.
7976 (debug_nothing_tree_charstar_uhwi): Likewise.
7977 * debug.c (do_nothing_debug_hooks): Adjust.
7978 (debug_false_tree_charstarstar_uhwistar): New do nothing.
7979 (debug_nothing_tree_charstar_uhwi): Likewise.
7980 * dbxout.c (dbx_debug_hooks): Adjust.
7981 (xcoff_debug_hooks): Likewise.
7982 * sdbout.c (sdb_debug_hooks): Likewise.
7983 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
7984 * dwarf2out.c (macinfo_label_base): New global.
7985 (dwarf2out_register_external_die): New function for the
7986 register_external_die hook.
7987 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
7988 (dwarf2_debug_hooks): Use them.
7989 (dwarf2_lineno_debug_hooks): Adjust.
7990 (struct die_struct): Add with_offset flag.
7991 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
7992 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
7993 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
7994 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
7995 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
7996 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
7997 defining section names for the early LTO debug variants.
7998 (reset_indirect_string): New helper.
7999 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
8000 (print_dw_val): Add support for offsetted symbol references.
8001 (get_ultimate_context): Split out from is_cxx.
8002 (is_cxx): Use get_ultimate_context.
8003 (is_fortran): Add decl overload.
8004 (compute_comp_unit_symbol): Split out worker from
8005 compute_section_prefix.
8006 (compute_section_prefix): Call compute_comp_unit_symbol and
8007 set comdat_type_p here.
8008 (output_die): Skip DIE symbol output for the LTO added one.
8009 Handle DIE symbol references with offset.
8010 (output_comp_unit): Guard section name mangling properly.
8011 For LTO debug sections emit a symbol at the section beginning
8012 which we use to refer to its DIEs.
8013 (add_abstract_origin_attribute): For DIEs registered via
8014 dwarf2out_register_external_die directly refer to the early
8015 DIE rather than indirectly through the shadow one we created.
8016 Remove obsolete call to dwarf2out_abstract_function for
8017 non-function/block origins.
8018 (gen_array_type_die): When generating early LTO debug do
8019 not emit DW_AT_string_length.
8020 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
8021 late when in LTO. As suggested place a gcc_unreachable for
8022 the DECL_ABSTRACT_P case.
8023 (gen_subprogram_die): Avoid another specification DIE
8024 for early built declarations/definitions for the late LTO case.
8025 (gen_variable_die): Add type references for late duplicated VLA dies
8027 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
8028 we have the abstract instance already.
8029 (process_scope_var): Adjust decl DIE contexts in LTO which
8030 first puts them in limbo.
8031 (gen_decl_die): Do not generate type DIEs late apart from
8032 types for VLAs or for decls we do not yet have a DIE. Do not
8033 call dwarf2out_abstract_function late.
8034 (dwarf2out_early_global_decl): Make sure to create DIEs
8035 for abstract instances of a decl first.
8036 (dwarf2out_late_global_decl): Adjust comment.
8037 (output_macinfo_op): With multiple macro sections use
8038 macinfo_label_base to distinguish labels.
8039 (output_macinfo): Likewise. Update macinfo_label_base.
8040 Pass in the line info label.
8041 (note_variable_value_in_expr): When generating LTO resolve
8042 all variable values here by generating DIEs as needed.
8043 (init_sections_and_labels): Add early LTO debug flag parameter
8044 and generate different sections and names if set. Add generation
8045 counter for the labels so we can have multiple of them.
8046 (reset_dies): Helper to allow DIEs to be output multiple times.
8047 (dwarf2out_finish): When outputting DIEs to the fat part of an
8048 LTO object first reset DIEs.
8049 (dwarf2out_early_finish): Output early DIEs when generating LTO.
8050 (modified_type_die): Check for decl_ultimate_origin being self
8052 (gen_type_die_with_usage): Likewise.
8053 (gen_typedef_die): Allow decl_ultimate_origin being self.
8054 (set_decl_abstract_flags): Remove.
8055 (set_block_abstract_flags): Likewise.
8056 (dwarf2out_abstract_function): Treat the early generated DIEs
8057 as the abstract copy and only add DW_AT_inline and
8058 DW_AT_artificial here and call set_decl_origin_self.
8059 If the DIE has an abstract origin don't do anything.
8060 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
8061 if we have none yet (Go fails to build one, PR78628).
8062 (variably_modified_type_p): Prevent endless recursion for Ada
8063 cyclic pointer types.
8064 * lto-streamer-in.c: Include debug.h.
8065 (dref_queue): New global.
8066 (lto_read_tree_1): Stream in DIE references.
8067 (lto_input_tree): Register DIE references.
8068 (input_function): Stream DECL_DEBUG_ARGS.
8069 * lto-streamer-out.c: Include debug.h.
8070 (lto_write_tree_1): Output DIE references.
8071 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
8072 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
8073 (output_function): Stream DECL_DEBUG_ARGS.
8074 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
8075 Stream DECL_ABSTRACT_ORIGIN.
8076 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
8077 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
8078 DECL_CONTEXT for file-scope decls.
8079 * lto-streamer.h (struct dref_entry): Declare.
8080 (dref_queue): Likewise.
8081 * cfgexpand.c (pass_expand::execute): Do not call the
8082 outlining_inline_function hook here.
8083 * lto-wrapper.c (debug_obj): New global.
8084 (tool_cleanup): Unlink it if required.
8085 (debug_objcopy): New function.
8086 (run_gcc): Handle early debug sections in the IL files by
8087 extracting them to separate files, partially linkin them and
8088 feeding the result back as result to the linker.
8089 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
8090 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
8091 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
8092 sections into a separate segment.
8093 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
8095 (darwin_asm_dwarf_section): Likewise.
8096 (darwin_asm_output_dwarf_offset): Likewise.
8097 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
8099 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8100 Alan Hayward <alan.hayward@arm.com>
8101 David Sherwood <david.sherwood@arm.com>
8103 * read-md.h (md_reader::record_potential_iterator_use): Replace
8104 pointer argument with an rtx and an index.
8105 * read-rtl.c (iterator_group::apply_iterator): Likewise.
8106 (apply_mode_iterator): Likewise.
8107 (apply_code_iterator): Likewise.
8108 (apply_int_iterator): Likewise.
8109 (apply_subst_iterator): Likewise.
8110 (record_iterator_use): Likewise.
8111 (record_attribute_use): Likewise.
8112 (md_reader::record_potential_iterator_use): Likewise. Update calls
8113 to record_iterator_use and apply_iterator.
8114 (iterator_use): Replace ptr with x and index.
8115 (attribute_use): Likewise.
8116 (apply_attribute_uses): Update calls to apply_iterator.
8117 (apply_iterators): Likewise. Update initialization of iterator_use.
8118 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
8119 and record_potential_iterator_use.
8120 (rtx_reader::read_rtx_operand): Likewise.
8122 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8123 Alan Hayward <alan.hayward@arm.com>
8124 David Sherwood <david.sherwood@arm.com>
8126 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
8129 2017-08-21 Richard Biener <rguenther@suse.de>
8132 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
8133 at struct end conservatively when comparing common bases.
8135 2017-08-21 Richard Biener <rguenther@suse.de>
8137 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
8138 (mem_ref_in_stmt): Remove.
8139 (determine_max_movement): Use ref index to get at the reference.
8140 (invariantness_dom_walker::before_dom_children): Deal with
8141 lim data already initialized.
8142 (gather_mem_refs_stmt): Initialize lim data and record ref index.
8144 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
8146 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
8147 (TARGET_ISA_ROUND): Ditto.
8148 (TARGET_ROUND): Ditto.
8149 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
8150 * config/i386/i386.md: Ditto.
8151 * config/i386/sse.md: Ditto.
8152 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
8153 with OPTION_MASK_ISA_SSE4_1.
8155 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
8158 * doc/extend.texi (x86 Built-in Functions): Correct the name of
8159 __builtin_ia32_lzcnt_u16.
8161 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
8164 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
8165 (rs6000_set_current_function): Rewrite function to use it.
8167 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
8170 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
8171 and TYPE_WARN_IF_NOT_ALIGN.
8172 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
8173 (handle_warn_if_not_align): New.
8174 (place_union_field): Call handle_warn_if_not_align.
8175 (place_field): Call handle_warn_if_not_align.
8176 Copy TYPE_WARN_IF_NOT_ALIGN.
8177 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
8178 (layout_type): Likewise.
8179 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
8181 (tree_decl_common): Add warn_if_not_align.
8182 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
8183 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
8184 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
8185 (DECL_WARN_IF_NOT_ALIGN): Likewise.
8186 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
8187 * doc/extend.texi: Document warn_if_not_aligned attribute.
8188 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
8190 2017-08-17 Martin Liska <mliska@suse.cz>
8193 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
8194 (get_data_dependence): Use it as pointer type.
8195 (distribute_loop): Likewise.
8197 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8199 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
8200 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
8201 (p8_vmrgow_<mode>_direct): New define_insn.
8202 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
8203 handle endianness for vmrgew and vmrgow permute patterns.
8205 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
8207 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
8208 * config/rs6000/rs6000-cpus.def: Remove comment.
8209 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
8210 (POWERPC_MASKS): Likewise.
8211 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
8212 use of TARGET_VSX_TIMODE.
8213 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
8214 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
8215 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
8216 (rs6000_option_override_internal): Remove dead code.
8217 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
8218 (rs6000_legitimize_reload_address): Likewise.
8219 (rs6000_legitimate_address_p): Likewise.
8220 (rs6000_opt_masks): Delete "vsx-timode".
8221 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
8222 from function comment.
8223 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
8224 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
8225 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
8227 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
8228 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
8229 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
8232 (define_peephole2): Likewise.
8234 2017-08-17 Martin Sebor <msebor@redhat.com>
8237 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
8238 past the end of an array.
8239 (test_pp_format): Add test cases.
8241 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
8243 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
8244 missing ECF_NOTHROW flags.
8246 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
8249 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
8250 operands residing in integer registers.
8251 (*vsx_le_perm_load_<mode>): Likewise.
8252 (*vsx_le_perm_store_<mode>): Likewise.
8253 (define_peephole2): Add peepholes to optimize the above.
8255 2017-08-17 Marek Polacek <polacek@redhat.com>
8258 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
8259 to mimic what shorten_compare did. Change the return type to bool.
8260 (fold_cond_expr_with_comparison): Update call to
8261 operand_equal_for_comparison_p.
8262 (fold_ternary_loc): Likewise.
8264 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
8266 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
8268 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
8270 2017-08-17 Richard Biener <rguenther@suse.de>
8272 * tree-ssa-structalias.c (solve_graph): When propagating
8273 to successors update the graphs succ edges and avoid duplicate work.
8275 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
8278 * config/i386/i386.c (ix86_option_override_internal): Save target
8279 specific options after ix86_stack_protector_guard_reg was changed.
8281 2017-08-17 Richard Biener <rguenther@suse.de>
8283 PR tree-optimization/81827
8284 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
8285 (new_var_info): Initialize it conservatively.
8286 (get_call_vi): Mark register vars.
8287 (new_scalar_tmp_constraint_exp): Likewise.
8288 (handle_rhs_call): Likewise.
8289 (handle_const_call): Likewise.
8290 (create_function_info_for): Likewise.
8291 (solve_constraints): Sort varinfos to separate register from
8292 non-register vars to pack points-to solution bitmaps during
8295 2017-08-17 Marek Polacek <polacek@redhat.com>
8297 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
8299 2017-08-17 Richard Biener <rguenther@suse.de>
8301 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
8302 to true when overflow is undefined and we saturated the result.
8304 2017-08-17 Alan Modra <amodra@gmail.com>
8307 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
8308 Don't use store multiple if only one reg needs saving.
8309 (interesting_frame_related_regno): New function.
8310 (rs6000_frame_related): Don't emit frame info for regs that
8312 (rs6000_emit_epilogue): Likewise.
8314 2017-08-16 Nathan Sidwell <nathan@acm.org>
8316 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
8317 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
8318 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
8319 * tree.c (free_lang_data_in_type): Use else-if chain. Always
8320 clear TYPE_LANG_1. Remove obsolete member-function stripping.
8321 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
8322 (verify_type): Adjust for TYPE_BINFO move.
8323 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
8324 process TYPE_BINFO directly.
8325 (hash_tree): Likewise.
8326 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8328 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8331 2017-08-16 David Malcolm <dmalcolm@redhat.com>
8333 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
8335 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
8338 * config/i386/i386.md (*anddi_1_btr): Change predicates of
8339 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
8340 Add ix86_binary_operator_ok to insn constraint.
8341 (*iordi_1_bts): Ditto.
8342 (*xordi_1_btc): Ditto.
8343 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
8344 Update corresponding peephole2 pattern.
8348 2017-08-16 Bin Cheng <bin.cheng@arm.com>
8350 PR tree-optimization/81832
8351 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
8352 copy loop header which has IFN_LOOP_DIST_ALIAS call.
8354 2017-08-16 Marek Polacek <polacek@redhat.com>
8357 * fold-const.c (fold_indirect_ref_1): Restore original behavior
8358 regarding size_zero_node.
8360 2017-08-16 Martin Liska <mliska@suse.cz>
8363 * config.gcc: Respect previously set extra_objs in case
8366 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
8368 PR tree-optimization/81835
8369 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
8370 the phi SSA_NAME. Check that the condition in a COND_EXPR does
8371 not depend on the phi.
8373 2017-08-16 Alan Modra <amodra@gmail.com>
8375 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
8378 2017-08-16 Alan Modra <amodra@gmail.com>
8380 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
8381 (save_reg_p): ..into this. Update all callers.
8382 (first_reg_to_save): Simplify.
8384 2017-08-16 Alan Modra <amodra@gmail.com>
8386 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
8389 2017-08-15 Joseph Myers <joseph@codesourcery.com>
8393 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
8394 constant count if that count is less than 32.
8396 2017-08-15 Nathan Sidwell <nathan@acm.org>
8398 * gcc.c (execute): Emit friendlier message if inferior is killed
8399 by an external cause.
8401 2017-08-15 Richard Biener <rguenther@suse.de>
8403 PR tree-optimization/81790
8404 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
8405 CONSTRUCTORs from simplifying and VN.
8407 2017-08-14 Martin Sebor <msebor@redhat.com>
8409 * builtin-attrs.def: Add comments.
8411 2017-08-14 Martin Sebor <msebor@redhat.com>
8414 * doc/extend.texi (attribute nonstring): Document new attribute.
8416 2017-08-14 Martin Sebor <msebor@redhat.com>
8419 * tree-diagnostic.c (default_tree_printer): Handle %G.
8420 * gimple-pretty-print.h (percent_G_format): Declare new function.
8421 * gimple-pretty-print.c (percent_G_format): Define.
8422 * tree-pretty-print.c (percent_K_format): Add argument.
8424 2017-08-14 Martin Sebor <msebor@redhat.com>
8426 PR translation/79998
8427 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
8428 Remove a stray space.
8430 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
8433 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
8434 (*iordi_1_bts): Ditto.
8435 (*xordi_1_btc): Ditto.
8437 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8440 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
8441 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8443 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
8444 quoted strings, and make more translator-friendly.
8445 (darwin_rs6000_override_options): Likewise.
8446 (rs6000_option_override_internal): Likewise.
8447 (rs6000_return_in_memory): Fix overlong line.
8448 (init_cmulative_args): Use quoted strings, and make more
8449 translator-friendly.
8450 (rs6000_pass_by_reference): Fix overlong line.
8451 (def_builtin): Use quoted strings.
8452 (altivec_expand_predicate_builtin): Use quoted strings, and make
8453 more translator-friendly.
8454 (htm_expand_builtin): Use quoted strings.
8455 (cpu_expand_builtin): Use quoted strings, and make more
8456 translator-friendly.
8457 (altivec_expand_builtin): Likewise.
8458 (paired_expand_predicate_builtin): Likewise.
8459 (rs6000_invalid_builtin): Likewise.
8460 (builtin_function_type): Use quoted strings.
8461 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
8462 more translator-friendly.
8463 (rs6000_trampoline_init): Likewise.
8464 (rs6000_handle_altivec_attribute): Likewise.
8465 (rs6000_inner_target_options): Use quoted strings.
8466 (rs6000_disable_incompatible_switches): Likewise.
8467 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
8468 strings, and make more translator-friendly.
8469 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
8471 2017-08-14 Bin Cheng <bin.cheng@arm.com>
8473 PR tree-optimization/81799
8474 * tree-loop-distribution.c (version_loop_by_alias_check): Force
8475 cond_expr to simple gimple operand.
8477 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
8480 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
8482 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
8486 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
8487 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
8488 TARGET_GASISR_PROLOGUES.
8489 * config/avr/avr.c (avr_option_override): Same.
8490 (avr_pass_pre_proep::execute): Same.
8492 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
8495 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
8496 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
8498 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
8500 * config/i386/i386.md (*load_tp_<mode>): Redefine as
8501 define_insn_and_split. Split to a memory load from 0 in
8502 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
8503 using PTR mode iterator.
8504 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
8505 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
8506 (*add_tp_<mode>): Redefine as define_insn_and_split.
8507 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
8508 address space. Merge with *add_tp_x32 using PTR mode iterator.
8509 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
8510 Split to an add with a memory load from 0 in
8511 DEFAULT_TLS_SEG_REG address space.
8513 2017-08-12 Andrew Pinski <apinski@cavium.com>
8515 * config/aarch64/aarch64-option-extensions.def (rdma):
8516 Fix feature string to what Linux prints out in /proc/cpuinfo.
8518 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
8521 * dwarf2out.c (modified_type_die): For C typedef types that have
8522 an ultimate origin, process the ultimate origin instead of the
8524 (gen_typedef_die): Assert that input DECLs have no ultimate
8526 (gen_type_die_with_usage): For typedef variants that have an
8527 ultimate origin, just call gen_decl_die on the original DECL.
8528 (process_scope_var): Avoid creating DIEs for local typedefs and
8529 concrete static variables.
8531 2017-08-12 Alan Modra <amodra@gmail.com>
8535 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
8536 match gnu-user.h startfile.
8537 (ENDFILE_LINUX_SPEC): Similarly.
8539 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
8542 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
8544 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
8546 2017-08-11 Tamar Christina <tamar.christina@arm.com>
8547 * config/aarch64/aarch64.md (mov<mode>): Change.
8548 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8549 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
8550 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
8552 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
8554 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
8555 for storage order barriers.
8557 2017-08-11 Martin Liska <mliska@suse.cz>
8560 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
8561 variables of void type.
8563 2017-08-11 Martin Liska <mliska@suse.cz>
8565 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
8566 TARGET_SUPPORTS_ALIASES.
8567 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
8568 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
8569 (optimize_weakref): Likewise.
8570 * symtab.c (symtab_node::noninterposable_alias): Likewise.
8571 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
8572 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
8574 2017-08-11 Martin Liska <mliska@suse.cz>
8577 * config/i386/i386.c (make_resolver_func): Do complete
8578 refactoring of the function.
8580 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8583 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
8584 * config/i386/i386.c (ix86_stack_protect_guard): Use
8585 ix86_stack_protect_guard_symbol_str to generate varible declaration.
8586 * doc/invoke.texi (x86 Options): Document
8587 -mstack-protector-guard-symbol= option.
8589 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8591 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
8592 * config/i386/i386.c (ix86_split_stack_guard): New function.
8593 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
8594 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
8595 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
8596 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
8597 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
8598 (split_stack_space_check): Call ix86_split_stack_guard.
8600 2017-08-10 Martin Sebor <msebor@redhat.com>
8602 * print-tree.c (print_node): Print location using the established
8604 Replace spaces with colons.
8605 (debug_raw, debug): Ditto.
8607 2017-08-10 Martin Sebor <msebor@redhat.com>
8610 * pretty-print.c (pp_format): Correct the handling of %s precision.
8612 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
8615 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8617 (ix86_finalize_stack_frame_flags): This. Also clear
8618 frame_pointer_needed if -fno-omit-frame-pointer is used without
8620 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8621 with ix86_finalize_stack_frame_flags.
8622 (ix86_expand_epilogue): Likewise.
8623 (ix86_expand_split_stack_prologue): Likewise.
8624 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
8626 2017-08-10 Martin Liska <mliska@suse.cz>
8629 * c-attribs.c (handle_target_attribute):
8630 Report warning for an empty string argument of target attribute.
8632 2017-08-09 Jakub Jelinek <jakub@redhat.com>
8635 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
8637 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
8638 or DECL_NONLOCAL labels.
8639 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
8640 or DECL_NONLOCAL labels here.
8642 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
8644 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
8645 to indicate when early gimple folding has been disabled.
8646 (rs6000_gimple_fold_builtin): Add debug content.
8647 (rs6000_invalid_builtin): Fix whitespace.
8648 (rs6000_expand_builtin): Fix whitespace.
8649 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
8651 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
8654 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
8655 SAVE_MULTIPLE if not all the registers that saves, should be saved.
8657 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
8659 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
8660 (qdf24xx): Likewise.
8661 * config/aarch64/aarch64.md: Include falkor.md.
8662 * config/aarch64/falkor.md: New.
8664 2017-08-09 Marek Polacek <polacek@redhat.com>
8667 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
8668 * diagnostic.c (emit_diagnostic): Add a comment.
8669 (emit_diagnostic_valist): New function.
8671 2017-08-09 Marek Polacek <polacek@redhat.com>
8674 * input.c (make_location): New overload.
8675 * input.h (make_location): Declare.
8677 2017-08-08 Alan Modra <amodra@gmail.com>
8678 H.J. Lu <hongjiu.lu@intel.com>
8681 * gcc.c (NO_PIE_SPEC): Delete.
8682 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
8684 (LINK_PIE_SPEC): ..to here.
8685 (LINK_COMMAND_SPEC): Support -no-pie.
8686 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
8687 chain of crtbegin*.o selection, update for PIE_SPEC changes and
8689 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
8690 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
8691 (ENDFILE_CRTEND_SPEC): Similarly.
8693 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
8696 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
8697 (mstack-protector-guard-offset=): Ditto.
8698 * config/i386/i386.c (ix86_option_override): Handle
8699 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
8701 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
8702 ix86_stack_protect_guard_offset variables.
8703 (TARGET_STACK_PROTECT_GUARD): Always define.
8704 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
8705 and -mstack-protector-guard-offset= options.
8707 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8709 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
8710 boundary case for the last candidate.
8712 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8714 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
8715 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
8717 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8720 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
8721 * config/aarch64/aarch64-builtins.c
8722 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
8723 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
8724 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
8726 2017-08-08 Tamar Christina <tamar.christina@arm.com>
8727 Andrew Pinski <pinskia@gmail.com>
8730 * internal-fn.def (XORSIGN): New.
8731 * optabs.def (xorsign_optab): New.
8732 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
8733 (convert_expand_mult_copysign): New.
8734 (pass_optimize_widening_mul::execute): Call
8735 convert_expand_mult_copysign.
8737 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8739 PR tree-optimization/81354
8740 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
8741 Insert on edges rather than explicitly creating landing pads.
8742 (analyze_candidates_and_replace): Commit edge inserts.
8744 2017-08-08 Richard Biener <rguenther@suse.de>
8747 * tree-ssa-loop-niter.c: Include tree-dfa.h.
8748 (expand_simple_operations): Also look through ADDR_EXPRs with
8749 MEM_REF bases treating them as POINTER_PLUS_EXPR.
8751 2017-08-08 Richard Biener <rguenther@suse.de>
8753 PR tree-optimization/81723
8754 * tree-vect-slp.c (struct bst_traits): New hash traits.
8755 (bst_fail): New global.
8756 (vect_build_slp_tree_2): New worker, split out from ...
8757 (vect_build_slp_tree): ... this now wrapping it with using
8758 bst_fail set to cache SLP tree build fails. Properly handle
8760 (vect_analyze_slp_instance): Allocate and free bst_fail.
8762 2017-08-08 Martin Liska <mliska@suse.cz>
8765 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
8766 LABEL_DECLs that can be from a different function.
8768 2017-08-08 Bin Cheng <bin.cheng@arm.com>
8770 PR tree-optimization/81744
8771 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
8772 loop's number of iterations.
8774 2017-08-08 Martin Liska <mliska@suse.cz>
8776 * asan.c: Include header files.
8777 * attribs.c (build_decl_attribute_variant): New function moved
8779 (build_type_attribute_qual_variant): Likewise.
8780 (cmp_attrib_identifiers): Likewise.
8781 (simple_cst_list_equal): Likewise.
8782 (omp_declare_simd_clauses_equal): Likewise.
8783 (attribute_value_equal): Likewise.
8784 (comp_type_attributes): Likewise.
8785 (build_type_attribute_variant): Likewise.
8786 (lookup_ident_attribute): Likewise.
8787 (remove_attribute): Likewise.
8788 (merge_attributes): Likewise.
8789 (merge_type_attributes): Likewise.
8790 (merge_decl_attributes): Likewise.
8791 (merge_dllimport_decl_attributes): Likewise.
8792 (handle_dll_attribute): Likewise.
8793 (attribute_list_equal): Likewise.
8794 (attribute_list_contained): Likewise.
8795 * attribs.h (lookup_attribute): New function moved from tree.[ch].
8796 (lookup_attribute_by_prefix): Likewise.
8797 * bb-reorder.c: Include header files.
8798 * builtins.c: Likewise.
8799 * calls.c: Likewise.
8800 * cfgexpand.c: Likewise.
8801 * cgraph.c: Likewise.
8802 * cgraphunit.c: Likewise.
8803 * convert.c: Likewise.
8804 * dwarf2out.c: Likewise.
8805 * final.c: Likewise.
8806 * fold-const.c: Likewise.
8807 * function.c: Likewise.
8808 * gimple-expr.c: Likewise.
8809 * gimple-fold.c: Likewise.
8810 * gimple-pretty-print.c: Likewise.
8811 * gimple.c: Likewise.
8812 * gimplify.c: Likewise.
8813 * hsa-common.c: Likewise.
8814 * hsa-gen.c: Likewise.
8815 * internal-fn.c: Likewise.
8816 * ipa-chkp.c: Likewise.
8817 * ipa-cp.c: Likewise.
8818 * ipa-devirt.c: Likewise.
8819 * ipa-fnsummary.c: Likewise.
8820 * ipa-inline.c: Likewise.
8821 * ipa-visibility.c: Likewise.
8823 * lto-cgraph.c: Likewise.
8824 * omp-expand.c: Likewise.
8825 * omp-general.c: Likewise.
8826 * omp-low.c: Likewise.
8827 * omp-offload.c: Likewise.
8828 * omp-simd-clone.c: Likewise.
8829 * opts-global.c: Likewise.
8830 * passes.c: Likewise.
8831 * predict.c: Likewise.
8832 * sancov.c: Likewise.
8833 * sanopt.c: Likewise.
8834 * symtab.c: Likewise.
8835 * toplev.c: Likewise.
8836 * trans-mem.c: Likewise.
8837 * tree-chkp.c: Likewise.
8838 * tree-eh.c: Likewise.
8839 * tree-into-ssa.c: Likewise.
8840 * tree-object-size.c: Likewise.
8841 * tree-parloops.c: Likewise.
8842 * tree-profile.c: Likewise.
8843 * tree-ssa-ccp.c: Likewise.
8844 * tree-ssa-live.c: Likewise.
8845 * tree-ssa-loop.c: Likewise.
8846 * tree-ssa-sccvn.c: Likewise.
8847 * tree-ssa-structalias.c: Likewise.
8848 * tree-ssa.c: Likewise.
8849 * tree-streamer-in.c: Likewise.
8850 * tree-vectorizer.c: Likewise.
8851 * tree-vrp.c: Likewise.
8853 * ubsan.c: Likewise.
8854 * varasm.c: Likewise.
8855 * varpool.c: Likewise.
8856 * tree.c: Remove functions moved to attribs.[ch].
8858 * config/aarch64/aarch64.c: Add attrs.h header file.
8859 * config/alpha/alpha.c: Likewise.
8860 * config/arc/arc.c: Likewise.
8861 * config/arm/arm.c: Likewise.
8862 * config/avr/avr.c: Likewise.
8863 * config/bfin/bfin.c: Likewise.
8864 * config/c6x/c6x.c: Likewise.
8865 * config/cr16/cr16.c: Likewise.
8866 * config/cris/cris.c: Likewise.
8867 * config/darwin.c: Likewise.
8868 * config/epiphany/epiphany.c: Likewise.
8869 * config/fr30/fr30.c: Likewise.
8870 * config/frv/frv.c: Likewise.
8871 * config/ft32/ft32.c: Likewise.
8872 * config/h8300/h8300.c: Likewise.
8873 * config/i386/winnt.c: Likewise.
8874 * config/ia64/ia64.c: Likewise.
8875 * config/iq2000/iq2000.c: Likewise.
8876 * config/lm32/lm32.c: Likewise.
8877 * config/m32c/m32c.c: Likewise.
8878 * config/m32r/m32r.c: Likewise.
8879 * config/m68k/m68k.c: Likewise.
8880 * config/mcore/mcore.c: Likewise.
8881 * config/microblaze/microblaze.c: Likewise.
8882 * config/mips/mips.c: Likewise.
8883 * config/mmix/mmix.c: Likewise.
8884 * config/mn10300/mn10300.c: Likewise.
8885 * config/moxie/moxie.c: Likewise.
8886 * config/msp430/msp430.c: Likewise.
8887 * config/nds32/nds32-isr.c: Likewise.
8888 * config/nds32/nds32.c: Likewise.
8889 * config/nios2/nios2.c: Likewise.
8890 * config/nvptx/nvptx.c: Likewise.
8891 * config/pa/pa.c: Likewise.
8892 * config/pdp11/pdp11.c: Likewise.
8893 * config/powerpcspe/powerpcspe.c: Likewise.
8894 * config/riscv/riscv.c: Likewise.
8895 * config/rl78/rl78.c: Likewise.
8896 * config/rx/rx.c: Likewise.
8897 * config/s390/s390.c: Likewise.
8898 * config/sh/sh.c: Likewise.
8899 * config/sol2.c: Likewise.
8900 * config/sparc/sparc.c: Likewise.
8901 * config/spu/spu.c: Likewise.
8902 * config/stormy16/stormy16.c: Likewise.
8903 * config/tilegx/tilegx.c: Likewise.
8904 * config/tilepro/tilepro.c: Likewise.
8905 * config/v850/v850.c: Likewise.
8906 * config/vax/vax.c: Likewise.
8907 * config/visium/visium.c: Likewise.
8908 * config/xtensa/xtensa.c: Likewise.
8910 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8913 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
8914 constraints since the -mupper-regs-* switches have been
8916 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
8917 into a vector from a double word element that was extracted from
8918 another vector, and eliminate extra XXPERMDI instructions.
8919 (vsx_concat_<mode>_2): Likewise.
8920 (vsx_concat_<mode>_3): Likewise.
8921 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
8922 concat to allow optimizing inserts from previous extracts.
8924 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
8926 * config/i386/i386.c (ix86_stack_protect_guard): Generate
8927 memory reference to a SSP offset in TLS address space.
8928 (ix86_print_operand) <case '@'>: Remove.
8929 (ix86_print_operand_punct_valid_p): Remove '@' code.
8930 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
8932 (stack_tls_protect_set_<mode>): Remove.
8933 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
8934 (stack_tls_protect_test_<mode>): Remove.
8935 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
8937 2017-08-07 Olivier Hainque <hainque@adacore.com>
8940 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
8942 2017-08-07 Douglas Rupp <rupp@adacore.com>
8944 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
8945 variable was referenced as multidir in command.
8947 2017-08-07 Jakub Jelinek <jakub@redhat.com>
8950 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
8953 2017-08-07 Martin Liska <mliska@suse.cz>
8955 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
8956 * config/rl78/rl78.c: Add include of attribs.h.
8957 * config/sh/sh.c: Likewise.
8958 * config/v850/v850.c: Likewise.
8960 2017-08-07 Tom de Vries <tom@codesourcery.com>
8963 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
8965 2017-08-07 Martin Liska <mliska@suse.cz>
8967 * config/mips/mips.c: Include attribs.h.
8969 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
8972 * doc/invoke.texi: Document change in behvaior for -Ofast for
8975 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
8977 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
8979 (aarch64_pop_regs): Likewise.
8980 (aarch64_gen_load_pair): Likewise.
8981 (aarch64_save_callee_saves): Likewise.
8982 (aarch64_restore_callee_saves): Likewise.
8984 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8986 * config/i386/i386.c: Revert the last change.
8988 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
8991 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8993 (ix86_finalize_stack_frame_flags): This. Also clear
8994 frame_pointer_needed if -fno-omit-frame-pointer is used without
8996 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8997 with ix86_finalize_stack_frame_flags.
8998 (ix86_expand_epilogue): Likewise.
8999 (ix86_expand_split_stack_prologue): Likewise.
9001 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9004 * config/i386/i386.c (get_builtin_code_for_version): Set priority
9005 to P_AES for Westmere.
9007 2017-08-07 Jonathan Yong <10walls@gmail.com>
9009 * config/i386/mingw.opt (fset-stack-executable): Removed.
9010 * config/i386/cygming.opt (fset-stack-executable): Moved
9012 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
9014 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
9016 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
9018 2017-08-07 Marek Polacek <polacek@redhat.com>
9021 * fold-const.c (fold_indirect_ref_1): Check type_domain.
9023 2017-08-07 Martin Liska <mliska@suse.cz>
9025 * attribs.h (canonicalize_attr_name): New function.
9026 (cmp_attribs): Move from c-format.c and adjusted.
9027 (is_attribute_p): Moved from tree.h.
9028 * tree-inline.c: Add new includes.
9029 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
9030 (private_is_attribute_p): Remove.
9031 (private_lookup_attribute): Likewise.
9032 (private_lookup_attribute_by_prefix): Simplify.
9033 (remove_attribute): Use is_attribute_p.
9034 * tree.h: Remove removed declarations.
9036 2017-08-07 Jakub Jelinek <jakub@redhat.com>
9039 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
9040 instead of computing it in the function. Formatting fix.
9041 (expand_case): Don't rely on default_edge being the first edge,
9042 clear it if removing it, pass default_edge to
9043 emit_case_dispatch_table.
9044 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
9047 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
9049 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
9050 insn in the function, emit NOP after the insn.
9052 2017-08-06 Tom de Vries <tom@codesourcery.com>
9054 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
9057 2017-08-06 Tom de Vries <tom@codesourcery.com>
9059 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
9062 2017-08-04 Yury Gribov <tetra2005@gmail.com>
9064 PR tree-optimization/57371
9065 * match.pd: New pattern.
9067 2017-08-04 Marek Polacek <polacek@redhat.com>
9070 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
9071 perform the computation in offset_int.
9073 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9075 PR tree-optimization/81136
9076 * tree-vectorizer.h: Include tree-hash-traits.h.
9077 (vec_base_alignments): New typedef.
9078 (vec_info): Add a base_alignments field.
9079 (vect_record_base_alignments): Declare.
9080 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
9082 (DR_IS_CONDITIONAL_IN_STMT): New macro.
9083 (create_data_ref): Add an is_conditional_in_stmt argument.
9084 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
9085 the is_conditional_in_stmt field.
9086 (data_ref_loc): Add an is_conditional_in_stmt field.
9087 (get_references_in_stmt): Set the is_conditional_in_stmt field.
9088 (find_data_references_in_stmt): Update call to create_data_ref.
9089 (graphite_find_data_references_in_stmt): Likewise.
9090 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
9091 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
9092 (vect_record_base_alignment): New function.
9093 (vect_record_base_alignments): Likewise.
9094 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
9095 for nested statements even if we fail to compute a misalignment.
9096 Use pooled base alignments for unconditional references.
9097 (vect_find_same_alignment_drs): Compare base addresses instead
9099 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
9100 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
9102 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9104 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
9105 Add an explicit name for the enum. Use auto_vec for slp_instances
9107 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
9109 (_bb_vec_info): Add a constructor and destructor.
9110 (vinfo_for_stmt): Return NULL for uids of -1 as well.
9111 (destroy_loop_vec_info): Delete.
9112 (vect_destroy_datarefs): Likewise.
9113 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
9114 (vec_info::vec_info): New function.
9115 (vec_info::~vec_info): Likewise.
9116 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
9117 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
9118 destroy_loop_vec_info.
9119 * tree-vect-loop.c (new_loop_vec_info): Replace with...
9120 (_loop_vec_info::_loop_vec_info): ...this.
9121 (destroy_loop_vec_info): Replace with...
9122 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
9123 the stmt_vec_infos. Leave handling of vec_info information to its
9124 destructor. Remove explicit vector releases.
9125 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
9126 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
9127 * tree-vect-slp.c (new_bb_vec_info): Replace with...
9128 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
9129 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
9130 (destroy_bb_vec_info): Replace with...
9131 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
9132 information to its destructor.
9133 (vect_slp_analyze_bb_1): Use new and delete instead of
9134 new_bb_vec_info and destroy_bb_vec_info.
9135 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
9138 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9140 * tree-data-ref.h (subscript): Add access_fn field.
9141 (data_dependence_relation): Add could_be_independent_p.
9142 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
9143 (same_access_functions): Move to tree-data-ref.c.
9144 * tree-data-ref.c (ref_contains_union_access_p): New function.
9145 (access_fn_component_p): Likewise.
9146 (access_fn_components_comparable_p): Likewise.
9147 (dr_analyze_indices): Add a reference to access_fn_component_p.
9148 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
9150 (constant_access_functions): Likewise.
9151 (add_other_self_distances): Likewise.
9152 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
9153 (initialize_data_dependence_relation): Use XCNEW and remove
9154 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
9155 of access functions that have the same type. Allow the
9156 subsequence to end with different bases in some circumstances.
9157 Record the chosen access functions in SUB_ACCESS_FN.
9158 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
9159 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
9160 (subscript_dependence_tester_1): Likewise dra and drb.
9161 (build_classic_dist_vector): Update calls accordingly.
9162 (subscript_dependence_tester): Likewise.
9163 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
9164 DDR_COULD_BE_INDEPENDENT_P.
9165 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
9166 comp_alias_ddrs instead of may_alias_ddrs.
9167 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
9169 (vect_analyze_data_ref_dependence): Use it if
9170 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
9171 distance vectors if that fails.
9172 (dependence_distance_ge_vf): New function.
9173 (vect_prune_runtime_alias_test_list): Use it. Don't clear
9174 LOOP_VINFO_MAY_ALIAS_DDRS.
9176 2017-08-04 Richard Biener <rguenther@suse.de>
9179 * fold-const.c (fold_binary_loc): Properly restrict
9180 minus_var0 && minus_var1 case when associating undefined overflow
9183 2017-08-04 Tom de Vries <tom@codesourcery.com>
9185 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
9187 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9189 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9190 Don't start diagnostic messages with a capital letter.
9191 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9193 (rs6000_invalid_builtin): Likewise.
9194 (rs6000_trampoline_init): Likewise.
9196 2017-08-03 Jakub Jelinek <jakub@redhat.com>
9199 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
9200 after setting changeable df flags.
9202 2017-08-03 Richard Biener <rguenther@suse.de>
9204 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
9205 up if the use is in USE - X.
9207 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
9209 * toplev.c (dumpfile.h): New include.
9210 (internal_error_reentered): New static function. Use it...
9211 (internal_error_function): ...here to handle reentered internal_error.
9213 2017-08-03 Richard Biener <rguenther@suse.de>
9216 * fold-const.c (split_tree): Add minus_var and minus_con
9217 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
9218 here but always use minus_*.
9219 (associate_trees): Assert we never associate with MINUS_EXPR
9220 and NULL first operand. Do not recurse for PLUS_EXPR operands
9221 when associating as MINUS_EXPR either.
9222 (fold_binary_loc): Track minus_var and minus_con.
9224 2017-08-03 Tom de Vries <tom@codesourcery.com>
9227 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
9228 ACCEL_COMPILER, apply finish_options on
9229 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
9231 2017-08-03 Tom de Vries <tom@codesourcery.com>
9234 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
9235 function_entry_patch_area_size > 0.
9237 2017-08-03 Jakub Jelinek <jakub@redhat.com>
9240 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
9241 instead of 10??LU, perform unit multiplication in wide_int,
9242 don't change alloc_object_size_limit if the limit is larger
9245 PR tree-optimization/81655
9246 PR tree-optimization/81588
9247 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
9248 the case when ranges[i].low and high are 1 for unsigned type with
9252 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
9253 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
9255 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9257 * tree-vrp.h: Add include guard.
9259 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
9262 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
9263 (ud2): New insn pattern.
9264 * config/i386/i386.c (ix86_expand_epilogue):
9265 For naked functions, generate ud2 instead of trap insn.
9267 2017-08-02 Marek Polacek <polacek@redhat.com>
9270 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
9272 2017-08-02 Tom de Vries <tom@codesourcery.com>
9273 Cesar Philippidis <cesar@codesourcery.com>
9275 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
9276 Add missing edge probabilities.
9278 2017-08-02 Tamar Christina <tamar.christina@arm.com>
9280 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
9283 2017-08-02 Jakub Jelinek <jakub@redhat.com>
9286 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
9287 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
9288 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
9289 prologue_seq sequences - if any.
9291 2017-08-02 Richard Biener <rguenther@suse.de>
9293 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
9294 via vectors if supported, integer extracts via punning if supported
9295 or otherwise vector extracts.
9297 2017-08-02 Richard Biener <rguenther@suse.de>
9299 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
9301 (bitmap_insert_into_set): ... this.
9303 2017-08-02 Richard Biener <rguenther@suse.de>
9305 PR tree-optimization/81633
9307 2015-08-17 Alan Hayward <alan.hayward@arm.com>
9309 PR tree-optimization/71752
9310 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
9312 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
9314 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
9315 (machine_function::call_ms2sysv_pad_out): Remove field.
9316 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
9317 (ix86_compute_frame_layout): Likewise.
9319 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
9322 * config/i386/i386.c (ix86_set_func_type): Disallow naked
9323 attribute with interrupt attribute.
9325 2017-08-01 Andrew Pinski <apinski@cavium.com>
9327 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
9328 BIT_INSERT_EXPR's operand 1
9329 to see if the types precision matches.
9331 2017-08-01 Martin Liska <mliska@suse.cz>
9334 * builtins.c (expand_builtin_memcpy_args): Remove.
9335 (expand_builtin_memcpy): Call newly added function
9336 expand_builtin_memory_copy_args.
9337 (expand_builtin_memcpy_with_bounds): Likewise.
9338 (expand_builtin_mempcpy): Remove last argument.
9339 (expand_builtin_mempcpy_with_bounds): Likewise.
9340 (expand_builtin_memory_copy_args): New function created from
9341 expand_builtin_mempcpy_args with small modifications.
9342 (expand_builtin_mempcpy_args): Remove.
9343 (expand_builtin_stpcpy): Remove unused argument.
9344 (expand_builtin): Likewise.
9345 (expand_builtin_with_bounds): Likewise.
9347 2017-08-01 Martin Liska <mliska@suse.cz>
9350 Make mempcpy more optimal (PR middle-end/70140).
9352 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9355 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9356 __builtin_vec_cmpne verify both arguments are compatible vectors
9357 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
9358 verify arg1_type is a pointer or array type. For __builtin_vec_st,
9359 move computation of aligned to after checking the argument types.
9363 * config/rs6000/vsx.md (vextract_fp_from_shorth,
9364 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
9367 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
9368 Doug Rupp <rupp@adacore.com>
9369 Olivier Hainque <hainque@adacore.com>
9371 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
9372 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
9374 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
9375 for TARGET_OS_CPP_BUILTIN.
9376 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
9377 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
9379 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
9380 passing required abi options to the assembler for EABI configurations.
9381 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
9382 of .text.hot and .text.unlikely sections for kernel modules when
9383 using ARM style exceptions.
9384 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
9385 options. Add EXTRA_CC1_SPEC.
9386 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
9388 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
9390 (ARM_TARGET2_DWARF_FORMAT): Define.
9391 * config/arm/t-vxworks: Adjust multilib control to removal of the
9392 Diab command line options.
9394 2017-08-01 Martin Liska <mliska@suse.cz>
9396 PR gcov-profile/81561
9397 * gcov.c (unblock): Make unblocking safe as we need to preserve
9398 index correspondence of blocks and block_lists.
9400 2017-08-01 Richard Biener <rguenther@suse.de>
9402 PR tree-optimization/81181
9403 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9404 (compute_antic): ... end of iteration here.
9406 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
9408 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
9409 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
9410 (ftree-slp-vectorize): Likewise.
9411 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
9412 can no longer be set independent of flag_tree_loop_vectorize.
9413 * omp-general.c (emp_max_vf): Likewise.
9414 * opts.c (enable_fdo_optimizations): Remove references to
9415 flag_tree_vectorize, these are now implicit.
9416 (common_handle_option): Remove handling for OPT_ftree_vectorize,
9417 and leave it for the options machinery.
9419 2017-08-01 Martin Liska <mliska@suse.cz>
9422 * builtins.c (expand_builtin_memcpy_args): Remove.
9423 (expand_builtin_memcpy): Call newly added function
9424 expand_builtin_memory_copy_args.
9425 (expand_builtin_memcpy_with_bounds): Likewise.
9426 (expand_builtin_mempcpy): Remove last argument.
9427 (expand_builtin_mempcpy_with_bounds): Likewise.
9428 (expand_builtin_memory_copy_args): New function created from
9429 expand_builtin_mempcpy_args with small modifications.
9430 (expand_builtin_mempcpy_args): Remove.
9431 (expand_builtin_stpcpy): Remove unused argument.
9432 (expand_builtin): Likewise.
9433 (expand_builtin_with_bounds): Likewise.
9435 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9438 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
9439 print "ds:" only for immediates in generic address space.
9441 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9444 * config/i386/i386.c (ix86_funciton_naked): New prototype.
9445 (ix86_function_ok_for_sibcall): Return false for naked functions.
9447 2017-08-01 Richard Biener <rguenther@suse.de>
9449 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
9450 (compute_antic): Seed worklist with exit block predecessors.
9451 * cfganal.c (dfs_find_deadend): For a cycle return the source
9452 of the edge closing it.
9454 2017-08-01 Tamar Christina <tamar.christina@arm.com>
9456 * config/aarch64/aarch64.c
9457 (aarch64_can_const_movi_rtx_p): Move 0 check.
9459 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9461 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
9462 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
9464 * match.pd: Ditto in address comparison pattern.
9466 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9468 PR tree-optimization/81627
9469 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
9470 closed ssa form for store-store chain.
9472 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9474 PR tree-optimization/81620
9475 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
9476 for store-store chain.
9478 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9480 PR tree-optimization/81588
9481 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
9482 ranges[i].in_p, invert comparison code ccode. For >/>=,
9483 swap rhs1 and rhs2 and comparison code unconditionally,
9484 for </<= don't do that. Don't swap rhs1/rhs2 again if
9485 ranges[i].in_p, instead invert comparison code ccode if
9486 opcode or oe->rank is BIT_IOR_EXPR.
9489 * optabs.def (vec_extract_optab, vec_init_optab): Change from
9490 a direct optab to conversion optab.
9491 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
9492 with GET_MODE_INNER as last argument instead of optab_handler.
9493 * expmed.c (extract_bit_field_1): Likewise. Use vector from
9494 vector extraction if possible and optab is available.
9495 * expr.c (store_constructor): Use convert_optab_handler instead
9496 of optab_handler. Use vector initialization from smaller
9497 vectors if possible and optab is available.
9498 * tree-vect-stmts.c (vectorizable_load): Likewise.
9499 * doc/md.texi (vec_extract, vec_init): Document that the optabs
9501 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
9502 of vec_init from half-sized vectors with the same element mode.
9503 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
9504 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
9505 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
9506 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
9507 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
9508 after mode in gen_vec_extract* calls.
9509 (vec_extract<mode>): Renamed to ...
9510 (vec_extract<mode><ssescalarmodelower>): ... this.
9511 (vec_extract<mode><ssehalfvecmodelower>): New expander.
9512 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
9513 element mode after mode in gen_vec_init* calls.
9514 (VEC_INIT_HALF_MODE): New mode iterator.
9515 (vec_init<mode>): Renamed to ...
9516 (vec_init<mode><ssescalarmodelower>): ... this.
9517 (vec_init<mode><ssehalfvecmodelower>): New expander.
9518 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
9519 (vec_extractv2sfsf): ... this.
9520 (vec_initv2sf): Renamed to ...
9521 (vec_initv2sfsf): ... this.
9522 (vec_extractv2si): Renamed to ...
9523 (vec_extractv2sisi): ... this.
9524 (vec_initv2si): Renamed to ...
9525 (vec_initv2sisi): ... this.
9526 (vec_extractv4hi): Renamed to ...
9527 (vec_extractv4hihi): ... this.
9528 (vec_initv4hi): Renamed to ...
9529 (vec_initv4hihi): ... this.
9530 (vec_extractv8qi): Renamed to ...
9531 (vec_extractv8qiqi): ... this.
9532 (vec_initv8qi): Renamed to ...
9533 (vec_initv8qiqi): ... this.
9534 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
9535 (vec_init<mode>): Renamed to ...
9536 (vec_init<mode><VEC_base_l>): ... this.
9537 (vec_extract<mode>): Renamed to ...
9538 (vec_extract<mode><VEC_base_l>): ... this.
9539 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
9540 (vec_initv2sfsf): ... this.
9541 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
9542 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9543 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
9544 element mode after mode in gen_vec_init* calls.
9545 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
9546 (vec_init<mode><Vel>): ... this.
9547 (vec_extract<mode>): Renamed to ...
9548 (vec_extract<mode><Vel>): ... this.
9549 * config/aarch64/iterators.md (Vel): New mode attribute.
9550 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
9551 Add element mode after mode in gen_vec_extract* calls.
9552 * config/s390/vector.md (non_vec_l): New mode attribute.
9553 (vec_extract<mode>): Renamed to ...
9554 (vec_extract<mode><non_vec_l>): ... this.
9555 (vec_init<mode>): Renamed to ...
9556 (vec_init<mode><non_vec_l>): ... this.
9557 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
9558 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
9560 * config/arm/iterators.md (V_elem_l): New mode attribute.
9561 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
9562 (vec_extract<mode><V_elem_l>): ... this.
9563 (vec_extractv2di): Renamed to ...
9564 (vec_extractv2didi): ... this.
9565 (vec_init<mode>): Renamed to ...
9566 (vec_init<mode><V_elem_l>): ... this.
9567 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
9568 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
9569 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
9570 Add element mode after gen_vec_extract* calls.
9571 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
9572 (vec_init<mode><unitmode>): ... this.
9573 (vec_extract<mode>): Renamed to ...
9574 (vec_extract<mode><unitmode>): ... this.
9575 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
9576 (vec_init<mode><unitmode>): ... this.
9577 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
9578 (vec_initv2sfsf): ... this.
9579 (vec_extractv2sf): Renamed to ...
9580 (vec_extractv2sfsf): ... this.
9581 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
9582 Add element mode after gen_vec_extract* calls.
9583 * config/mips/mips.md (unitmode): New mode iterator.
9584 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
9585 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
9586 * config/spu/spu.md (inner_l): New mode attribute.
9587 (vec_init<mode>): Renamed to ...
9588 (vec_init<mode><inner_l>): ... this.
9589 (vec_extract<mode>): Renamed to ...
9590 (vec_extract<mode><inner_l>): ... this.
9591 * config/sparc/sparc.md (veltmode): New mode iterator.
9592 (vec_init<VMALL:mode>): Renamed to ...
9593 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
9594 * config/ia64/vect.md (vec_initv2si): Renamed to ...
9595 (vec_initv2sisi): ... this.
9596 (vec_initv2sf): Renamed to ...
9597 (vec_initv2sfsf): ... this.
9598 (vec_extractv2sf): Renamed to ...
9599 (vec_extractv2sfsf): ... this.
9600 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
9601 (vec_init<mode>): Renamed to ...
9602 (vec_init<mode><VEC_base_l>): ... this.
9603 (vec_extract<mode>): Renamed to ...
9604 (vec_extract<mode><VEC_base_l>): ... this.
9605 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
9606 (vec_initv2sfsf): ... this.
9607 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
9608 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9609 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
9610 gen_vec_init* calls.
9612 2017-08-01 Richard Biener <rguenther@suse.de>
9614 PR tree-optimization/81297
9615 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
9616 TREE_OVERFLOW from INTEGER_CSTs.
9618 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
9620 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
9622 2017-07-31 Carl Love <cel@us.ibm.com>
9624 * config/rs6000/rs6000-c: Add support for built-in functions
9625 vector signed char vec_xl_be (signed long long, signed char *);
9626 vector unsigned char vec_xl_be (signed long long, unsigned char *);
9627 vector signed int vec_xl_be (signed long long, signed int *);
9628 vector unsigned int vec_xl_be (signed long long, unsigned int *);
9629 vector signed long long vec_xl_be (signed long long, signed long long *);
9630 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
9631 vector signed short vec_xl_be (signed long long, signed short *);
9632 vector unsigned short vec_xl_be (signed long long, unsigned short *);
9633 vector double vec_xl_be (signed long long, double *);
9634 vector float vec_xl_be (signed long long, float *);
9635 * config/rs6000/altivec.h (vec_xl_be): Add #define.
9636 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
9637 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
9639 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
9640 (altivec_expand_builtin): Add switch statement to call
9641 altivec_expand_xl_be for each builtin.
9642 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
9643 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
9644 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
9645 __builtin_vsx_le_be_v16qi.
9646 * doc/extend.texi: Update the built-in documentation file for the
9647 new built-in functions.
9649 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9652 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
9654 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9656 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9658 * config.gcc: Add z14.
9659 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
9660 CPU model numbers for z13s and z14.
9661 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
9663 * config/s390/s390-opts.h (enum processor_type): Rename
9664 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9665 * config/s390/s390.c (processor_table): Add field for CPU name to
9666 be passed to Binutils.
9667 (s390_asm_output_machine_for_arch): Use the new field in
9668 processor_table for Binutils.
9669 (s390_expand_builtin): Replace arch12 with z14.
9670 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9671 (s390_get_sched_attrmask): Likewise.
9672 (s390_get_unit_mask): Likewise.
9673 * config/s390/s390.opt: Add z14 to processor_type enum.
9675 2017-07-31 Martin Jambor <mjambor@suse.cz>
9678 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
9679 regardless of optimization level.
9681 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
9682 Martin Liska <mliska@suse.cz>
9684 * predict.def: Remove old comment and adjust probability.
9685 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
9688 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9691 * config/i386/i386.c (ix86_function_naked): New function.
9692 (ix86_can_use_return_insn_p): Return false for naked functions.
9693 (ix86_expand_prologue): Skip prologue for naked functions.
9694 (ix86_expand_epilogue): Skip epilogue for naked functions
9695 and emit trap instruction.
9696 (ix86_warn_func_return): New function.
9697 (ix86_attribute_table): Add "naked" attribute specification.
9698 (TARGET_WARN_FUNC_RETURN): Define.
9699 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
9701 2017-07-31 Martin Liska <mliska@suse.cz>
9703 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
9704 (dump_gimple_bb_header): Always dump BB info.
9705 (pp_cfg_jump): Do not append info about BB when dumping a jump.
9707 2017-07-31 Martin Liska <mliska@suse.cz>
9710 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
9711 also with current_function_decl non-null equality.
9713 2017-07-31 Jakub Jelinek <jakub@redhat.com>
9716 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
9717 change type to the element type, instead add eltype variable and
9718 use it where we are interested in the element type.
9720 PR tree-optimization/81603
9721 * ipa-polymorphic-call.c
9722 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
9723 offset arithmetic in offset_int, bail out if the resulting bit offset
9724 doesn't fit into shwi.
9726 2017-07-31 Martin Liska <mliska@suse.cz>
9728 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
9729 (gimplify_save_expr): Fix comment.
9731 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
9734 * config/i386/i386.c (ix86_function_arg): Update arguments for
9736 (ix86_compute_frame_layout): Set the initial stack offset to
9737 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
9738 INCOMING_FRAME_SP_OFFSET.
9739 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
9740 stack before exception handler returns.
9741 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
9742 the 'ERROR_CODE' for exception handler.
9744 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
9746 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
9747 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
9748 (ASM_OUTPUT_REG_POP): Ditto.
9749 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
9750 instead of asm_fprintf to output pure string.
9752 2017-07-29 Jakub Jelinek <jakub@redhat.com>
9754 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
9755 to imported_module_or_decl hook.
9756 (debug_nothing_tree_tree_tree_bool): Remove.
9757 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
9758 * debug.c (do_nothing_debug_hooks): Use
9759 debug_nothing_tree_tree_tree_bool_bool instead of
9760 debug_nothing_tree_tree_tree_bool.
9761 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
9762 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
9763 * sdbout.c (sdb_debug_hooks): Likewise.
9764 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
9765 (gen_namespace_die): Add DW_AT_export_symbols attribute if
9767 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
9768 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
9769 attribute, don't add anything.
9771 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9773 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
9774 (fold_build2_stat_loc): Likewise.
9775 (fold_build3_stat_loc): Likewise.
9776 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
9777 (fold_build1_loc): Remove macro.
9778 (fold_build2_loc): Likewise.
9779 (fold_build3_loc): Likewise.
9781 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9783 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
9784 (gimple_build_debug_bind_source_stat): Likewise.
9785 * gimple.h (gimple_build_debug_bind): Remove macro.
9786 (gimple_build_debug_bind_source): Likewise.
9788 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9790 * bitmap.c (bitmap_alloc): Adjust.
9791 (bitmap_gc_alloc): Likewise.
9792 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
9794 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9796 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
9797 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
9798 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
9799 (bitmap_gc_alloc_stat): Likewise.
9800 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
9802 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9804 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
9805 * rtl.h (shallow_copy_rtx): Remove macro.
9807 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9809 * emit-rtl.c (gen_raw_REG): Adjust.
9810 * gengenrtl.c (gendef): Likewise.
9811 * rtl.c (rtx_alloc_stat): Remove _stat from name.
9812 * rtl.h (rtx_alloc): Remove macro.
9814 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9816 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
9817 (build_tree_list_stat): Likewise.
9818 * tree.h (build_tree_list): Remove macro.
9819 (build_tree_list_vec): Likewise.
9821 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9823 * tree.c (make_vector_stat): Remove _stat from name.
9824 (build_vector_stat): Likewise.
9825 * tree.h (make_vector_stat): Remove macro.
9826 (build_vector_stat): Likewise.
9828 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9830 * tree.h (build_var_debug_value): Remove prototype.
9832 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9834 * tree.c (tree_cons_stat): Remove _stat from name.
9835 * tree.h (tree_cons): Remove macro.
9837 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9839 * tree.c (build_vl_exp_stat): Remove _stat from name.
9840 * tree.h (build_vl_exp): Remove macro.
9842 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9844 * tree.c (build_decl_stat): Remove _stat from name.
9845 * tree.h (build_decl): Remove macro.
9847 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9849 * gimple.c (gimple_build_with_ops_stat): Adjust.
9850 (gimple_alloc_stat): Remove _stat from name.
9851 * gimple.h (gimple_alloc): Remove macro.
9853 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9855 * tree.c (make_tree_vec_stat): Remove _stat from name.
9856 (grow_tree_vec_stat): Likewise.
9857 * tree.h (make_tree_vec_stat): Adjust prototype.
9858 (grow_tree_vec_stat): Likewise.
9859 (make_tree_vec): Remove macro.
9860 (grow_tree_vec): Likewise.
9862 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9864 * fold-const.c (fold_build1_stat_loc): Adjust.
9865 (fold_build2_stat_loc): Likewise.
9866 (fold_build3_stat_loc): Likewise.
9867 * tree.c (build0_stat): Remove _stat from name.
9868 (build1_stat): Likewise.
9869 (build2_stat): Likewise.
9870 (build3_stat): Likewise.
9871 (build4_stat): Likewise.
9872 (build5_stat): Likewise.
9873 * tree.h (build1_loc): Remove macro, and rename _stat function
9875 (build2_loc): Likewise.
9876 (build3_loc): Likewise.
9877 (build4_loc): Likewise.
9878 (build5_loc): Likewise.
9880 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9882 * tree.c (make_int_cst_stat): Remove _stat from name.
9883 * tree.h (make_int_cst_stat): Adjust prototype.
9884 (make_int_cst): Remove macro.
9886 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9888 * tree.c (make_tre_binfo_stat): Remove _stat from name.
9889 * tree.h (make_tree_binfo_stat): Adjust prototype.
9890 (make_tree_binfo): Remove.
9892 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9894 * tree.c (copy_node_stat): Rename to copy_node.
9895 (build_distinct_type_copy): Adjust.
9896 * tree.h (copy_node_stat): Adjust prototype.
9897 (copy_node): Remove macro.
9899 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
9901 * tree.c (make_node_stat): rename to make_node.
9902 (build_tree_list_stat): Adjust.
9903 (build0_stat): Likewise.
9904 (build2_stat): Likewise.
9905 (build3_stat): Likewise.
9906 (build4_stat): Likewise.
9907 (build5_stat): Likewise.
9908 (build_decl_stat): Likewise.
9909 * tree.h (make_node_stat): Adjust prototype.
9910 (make_node): remove macro.
9912 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
9914 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
9915 (PPC_FEATURE2_SCV): Likewise.
9916 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
9918 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9920 * config/aarch64/aarch64.c
9921 (aarch64_internal_mov_immediate): Add new special pattern.
9922 * config/aarch64/aarch64.md (*movdi_aarch64):
9923 Add reg/32bit const mov case.
9925 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9926 Richard Sandiford <richard.sandiford@linaro.org>
9928 * config/aarch64/aarch64.md (mov<mode>): Generalize.
9929 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
9930 Add integer and movi cases.
9931 (movi-split-hf-df-sf split, fp16): New.
9932 (enabled): Added TARGET_FP_F16INST.
9933 * config/aarch64/iterators.md (GPF_HF): New.
9934 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
9936 2017-07-28 Tamar Christina <tamar.christina@arm.com>
9938 * config/aarch64/aarch64.c
9939 (aarch64_simd_container_mode): Add prototype.
9940 (aarch64_expand_mov_immediate): Add HI support.
9941 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
9942 (aarch64_can_const_movi_rtx_p): New.
9943 (aarch64_preferred_reload_class):
9944 Remove restrictions of using FP registers for certain SIMD operations.
9945 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
9946 (aarch64_valid_floating_const): Add integer move validation.
9947 (aarch64_simd_imm_scalar_p): Remove.
9948 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
9949 (aarch64_legitimate_constant_p): Expand list of supported cases.
9950 * config/aarch64/aarch64-protos.h
9951 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
9952 (aarch64_reinterpret_float_as_int): New.
9953 (aarch64_simd_imm_scalar_p): Remove.
9954 * config/aarch64/constraints.md (Uvi): New.
9955 (Dd): Split into Ds and new Dd.
9956 * config/aarch64/aarch64.md (*movsi_aarch64):
9958 (*movdi_aarch64): Add SIMD mov case.
9960 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9962 * tree-predcom.c: (struct chain): Handle store-store chain in which
9963 stores for elimination only store loop invariant values.
9964 (execute_pred_commoning_chain): Ditto.
9965 (prepare_initializers_chain_store_elim): Ditto.
9966 (prepare_finalizers): Ditto.
9967 (is_inv_store_elimination_chain): New function.
9968 (initialize_root_vars_store_elim_1): New function.
9970 2017-07-28 Bin Cheng <bin.cheng@arm.com>
9972 * tree-predcom.c: Revise general description of the pass.
9973 (enum chain_type): New enum type for store elimination.
9974 (struct chain): New field supporting store elimination.
9975 (struct component): Ditto.
9976 (dump_chain): Dump store-stores chain.
9977 (release_chain): Release resources.
9978 (split_data_refs_to_components): Compute and create component
9979 contains only stores for elimination.
9980 (get_chain_last_ref_at): New function.
9981 (make_invariant_chain): Initialization.
9982 (make_rooted_chain): Specify chain type in parameter and record it.
9983 (add_looparound_copies): Skip for store-stores chain.
9984 (determine_roots_comp): Compute type of chain and pass it to
9986 (initialize_root_vars_store_elim_2): New function.
9987 (finalize_eliminated_stores): New function.
9988 (remove_stmt): Handle store for elimination.
9989 (execute_pred_commoning_chain): Execute predictive commoning on
9991 (determine_unroll_factor): Skip unroll for store-stores chain.
9992 (prepare_initializers_chain_store_elim): New function.
9993 (prepare_initializers_chain): Hanlde store-store chain.
9994 (prepare_finalizers_chain, prepare_finalizers): New function.
9995 (tree_predictive_commoning_loop): Return integer value indicating
9996 if loop is unrolled or lcssa form is corrupted.
9997 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
9999 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10001 * tree-predcom.c (initialize_root): Delete.
10002 (execute_pred_commoning_chain): Initialize root vars and replace
10003 reference of non-combined chain directly, rather than call above
10006 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10008 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
10009 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
10011 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10013 * tree-predcom.c (struct chain): New field init_seq.
10014 (release_chain): Release init_seq.
10015 (prepare_initializers_chain): Record intialization stmts in above
10017 (insert_init_seqs): New function.
10018 (tree_predictive_commoning_loop): Call insert_init_seqs.
10020 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10022 * tree-predcom.c (determine_roots_comp): Skip trivial components.
10024 2017-07-28 Richard Biener <rguenther@suse.de>
10026 * match.pd: Remove superfluous :c.
10027 * genmatch.c (simplify::id): Add member.
10028 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
10030 (current_id): New global.
10031 (dt_node::parent): Move from ...
10032 (dt_operand::parent): ... here. Add for_id member.
10033 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
10034 (decision_tree::find_node): Relax order requirement when
10035 merging DT_TRUE nodes to ones inbetween the current simplify
10036 and the one we try to merge with. Add diagnostic whenever
10037 we need to enforce pattern order by not merging.
10038 (decision_tree::insert): Set current_id.
10039 (decision_tree::print_node): Dump parent node and for_id.
10040 (parser::last_id): Add member.
10041 (parser::push_simplify): Assign unique id.
10042 (parser::parser): Initialize last_id.
10044 2017-07-28 Martin Liska <mliska@suse.cz>
10047 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
10048 gimple_build_debug_bind.
10050 2017-07-28 Richard Biener <rguenther@suse.de>
10052 PR tree-optimization/81502
10053 * match.pd: Add pattern combining BIT_INSERT_EXPR with
10055 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
10057 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
10058 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
10059 for BIT_FIELD_REF args.
10060 * fold-const.c (make_bit_field_ref): Likewise.
10061 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
10063 2017-07-28 Jakub Jelinek <jakub@redhat.com>
10066 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
10067 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
10068 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
10069 Or it into SANITIZER_UNDEFINED.
10070 * ubsan.c: Include gimple-fold.h and varasm.h.
10071 (ubsan_expand_ptr_ifn): New function.
10072 (instrument_pointer_overflow): New function.
10073 (maybe_instrument_pointer_overflow): New function.
10074 (instrument_object_size): Formatting fix.
10075 (pass_ubsan::execute): Call instrument_pointer_overflow
10076 and maybe_instrument_pointer_overflow.
10077 * internal-fn.c (expand_UBSAN_PTR): New function.
10078 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
10079 * sanitizer.def (__ubsan_handle_pointer_overflow,
10080 __ubsan_handle_pointer_overflow_abort): New builtins.
10081 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
10082 * internal-fn.def (UBSAN_PTR): New internal function.
10083 * opts.c (sanitizer_opts): Add pointer-overflow.
10084 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
10085 * fold-const.c (build_range_check): Compute pointer range check in
10086 integral type if pointer arithmetics would be needed. Formatting
10089 2017-07-28 Martin Liska <mliska@suse.cz>
10092 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
10093 parameters that are of a variable-length.
10095 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
10097 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
10099 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
10100 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
10101 (CRT_CALL_STATIC_FUNCTION): Likewise.
10102 (ASM_DEFAULT_SPEC): New define.
10103 (ASM_SPEC32): Likewise.
10104 (ASM_SPEC64): Likewise.
10105 (ASM_SPEC_COMMON): Likewise.
10106 (ASM_SPEC): Likewise.
10107 (INVALID_64BIT): Likewise.
10108 (LINK_OS_DEFAULT_SPEC): Likewise.
10109 (LINK_OS_SPEC32): Likewise.
10110 (LINK_OS_SPEC64): Likewise.
10111 (POWERPC_LINUX): Likewise.
10112 (PTRDIFF_TYPE): Likewise.
10113 (RESTORE_FP_PREFIX): Likewise.
10114 (RESTORE_FP_SUFFIX): Likewise.
10115 (SAVE_FP_PREFIX): Likewise.
10116 (SAVE_FP_SUFFIX): Likewise.
10117 (SIZE_TYPE): Likewise.
10118 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
10119 (TARGET_64BIT): Likewise.
10120 (TARGET_64BIT): Likewise.
10121 (TARGET_AIX): Likewise.
10122 (WCHAR_TYPE_SIZE): Likewise.
10123 (WCHAR_TYPE): Undefine.
10124 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
10125 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
10126 (CPP_OS_RTEMS_SPEC): Delete.
10127 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
10128 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
10130 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
10132 2017-07-28 Jakub Jelinek <jakub@redhat.com>
10134 PR tree-optimization/81578
10135 * tree-parloops.c (build_new_reduction): Bail out if
10136 reduction_code isn't one of the standard OpenMP reductions.
10137 Move the details printing after that decision.
10139 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10141 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
10142 related to reload_in_progress.
10143 (splat_input_operand): Likewise.
10144 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
10146 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
10148 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
10149 (TARGET_INSTANTIATE_DECLS): Likewise.
10150 (legitimate_indexed_address_p): Delete reload_in_progress code.
10151 (rs6000_debug_legitimate_address_p): Likewise.
10152 (rs6000_eliminate_indexed_memrefs): Likewise.
10153 (rs6000_emit_le_vsx_store): Likewise.
10154 (rs6000_emit_move_si_sf_subreg): Likewise.
10155 (rs6000_emit_move): Likewise.
10156 (register_to_reg_type): Likewise.
10157 (rs6000_pre_atomic_barrier): Likewise.
10158 (rs6000_machopic_legitimize_pic_address): Likewise.
10159 (rs6000_allocate_stack_temp): Likewise.
10160 (rs6000_address_for_fpconvert): Likewise.
10161 (rs6000_address_for_altivec): Likewise.
10162 (rs6000_secondary_memory_needed_rtx): Delete function.
10163 (rs6000_check_sdmode): Likewise.
10164 (rs6000_alloc_sdmode_stack_slot): Likewise.
10165 (rs6000_instantiate_decls): Likewise.
10166 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
10167 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
10168 Delete reload_in_progress.
10169 (*vec_reload_and_plus_<mptrsize>): Likewise.
10170 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
10171 (vsx_div_v2di): Likewise.
10172 (vsx_udiv_v2di): Likewise.
10174 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10176 * config/rs6000/rs6000.opt (mlra): Replace with stub.
10177 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
10178 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
10179 (rs6000_debug_reg_global): Delete print of LRA status.
10180 (rs6000_option_override_internal): Delete dead LRA related code.
10181 (rs6000_lra_p): Delete function.
10182 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
10184 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
10186 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
10187 * config/riscv/rtems.h: New file.
10189 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10190 Sudakshina Das <sudi.das@arm.com>
10192 * config/aarch64/aarch64.md
10193 (define_split for and<mode>3nr_compare): Move
10194 non aarch64_logical_operand to a register.
10195 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
10196 register immediate operand to a register.
10197 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
10199 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10201 PR middle-end/81564
10202 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
10204 2017-07-27 Richard Biener <rguenther@suse.de>
10206 PR tree-optimization/81573
10207 PR tree-optimization/81494
10208 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
10209 multi defuse cycle case.
10211 2017-07-27 Richard Biener <rguenther@suse.de>
10213 PR tree-optimization/81571
10214 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
10217 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
10219 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
10220 earlier and only if MASK_FPU is set. Adjust formatting.
10222 2017-07-27 Martin Liska <mliska@suse.cz>
10224 * opt-functions.awk: Add validation of value of Init.
10225 * optc-gen.awk: Pass new argument.
10227 2017-07-27 Martin Liska <mliska@suse.cz>
10229 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
10230 Fix wrong condition.
10232 2017-07-27 Martin Liska <mliska@suse.cz>
10234 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
10235 BBs and edges seen by autoFDO.
10237 2017-07-27 Richard Biener <rguenther@suse.de>
10239 PR tree-optimization/81502
10240 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
10241 with incompatible but same sized type.
10242 (execute_update_addresses_taken): Likewise.
10244 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
10246 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
10247 flag_tree_loop_vectorize rather than flag_tree_vectorize.
10249 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10252 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
10253 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
10254 Change s_operand to memory_operand.
10256 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
10258 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
10259 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
10260 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
10261 Emit instructions rather than returning an expression. Handle TFmode
10262 and KFmode by casting to TImode.
10263 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
10264 (rs6000_emit_le_vsx_store): Likewise.
10265 * config/rs6000/vsx.md (VSX_TI): New iterator.
10266 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
10267 (*vsx_le_undo_permute_<mode>): Likewise.
10268 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
10269 emit the split sequence.
10270 (*vsx_le_perm_store_<mode>): Likewise.
10272 2017-07-27 Jakub Jelinek <jakub@redhat.com>
10274 PR tree-optimization/81555
10275 PR tree-optimization/81556
10276 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
10277 if true, force CHANGED for the recursive invocation.
10278 (reassociate_bb): Remember original length of ops array, pass
10279 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
10281 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
10282 attributes for noipa attribute. For naked attribute use
10283 lookup_attribute first before lookup_attribute_spec.
10284 * final.c (rest_of_handle_final): Disable IPA RA for functions with
10286 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
10287 for functions with noipa attribute.
10288 (cgraph_externally_visible_p): Return true for functions with noipa
10290 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
10291 for functions with noipa attribute.
10292 * doc/extend.texi: Document noipa function attribute.
10293 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
10294 also for functions with noipa attribute.
10295 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
10297 2017-07-26 Andrew Pinski <apinski@cavium.com>
10299 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
10300 vec_unalign_load_cost and vec_unalign_store_cost.
10302 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10304 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
10305 -mvsx-small-integer option.
10306 (ISA_3_0_MASKS_IEEE): Likewise.
10307 (OTHER_VSX_VECTOR_MASKS): Likewise.
10308 (POWERPC_MASKS): Likewise.
10309 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
10310 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
10311 code, only testing for DImode being allowed in non-VSX floating
10313 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
10314 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
10316 (rs6000_option_override_internal): Delete -mvsx-small-integer.
10317 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
10318 TARGET_P8_VECTOR test.
10319 (rs6000_secondary_reload_simple_move): Likewise.
10320 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
10321 since TARGET_P9_VECTOR was already tested.
10322 (rs6000_opt_masks): Remove -mvsx-small-integer.
10323 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
10324 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10326 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
10327 test for TARGET_VEXTRACTUB was used, and that uses
10329 (p9 extract splitter): Likewise.
10330 (vsx_extract_<mode>_di_p9): Likewise.
10331 (vsx_extract_<mode>_store_p9): Likewise.
10332 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
10333 for TARGET_P9_VECTOR was used. Delete code that is now dead with
10334 the elimination of TARGET_VSX_SMALL_INTEGER.
10335 (vsx_extract_<mode>_p8): Likewise.
10336 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
10337 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
10338 (vsx_set_<mode>_p9): Likewise.
10339 (vsx_set_v4sf_p9): Likewise.
10340 (vsx_set_v4sf_p9_zero): Likewise.
10341 (vsx_insert_extract_v4sf_p9): Likewise.
10342 (vsx_insert_extract_v4sf_p9_2): Likewise.
10343 * config/rs6000/rs6000.md (sign extend splitter): Change
10344 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
10345 (floatsi<mode>2_lfiwax_mem): Likewise.
10346 (floatunssi<mode>2_lfiwzx_mem): Likewise.
10347 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
10348 since a test for TARGET_P9_VECTOR was used.
10349 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10350 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
10351 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10352 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10353 TARGET_P8_VECTOR test.
10354 (fix_trunc<mode>si2_stfiwx): Likewise.
10355 (fix_trunc<mode>si2_internal): Likewise.
10356 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
10357 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10359 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10360 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10361 TARGET_P8_VECTOR test.
10362 (fixuns_trunc<mode>si2_stfiwx): Likewise.
10363 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
10364 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10366 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10367 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
10368 since a test for TARGET_P9_VECTOR was used.
10369 (splitter for loading small constants): Likewise.
10371 2017-07-26 Andrew Pinski <apinski@cavium.com>
10373 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
10376 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
10379 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
10380 (fp_valid_at): Likewise.
10382 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10384 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
10385 (qdf24xx_addrcost_table): Likewise.
10386 (cortexa57_tunings): Update to use generic_branch_cost.
10387 (cortexa72_tunings): Likewise.
10388 (cortexa73_tunings): Likewise.
10389 (qdf24xx_tunings): Likewise.
10391 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10393 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
10394 (thunderx2t99_branch_cost): Likewise.
10395 (cortexa35_tunings): Update to use generic_branch_cost.
10396 (cortexa53_tunings): Likewise.
10397 (cortexa57_tunings): Likewise.
10398 (cortexa72_tunings): Likewise.
10399 (cortexa73_tunings): Likewise.
10400 (thunderx2t99_tunings): Likewise.
10402 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10404 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
10405 (sparc_option_override): Honour MASK_FSMULD.
10406 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
10407 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
10408 * config/sparc/sparc.opt (mfsmuld): New option.
10409 * doc/invoke.texi (mfsmuld): Document option.
10411 2017-07-26 Marek Polacek <polacek@redhat.com>
10413 PR middle-end/70992
10414 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
10416 2017-07-26 Richard Biener <rguenther@suse.de>
10418 * gimple-match-head.c (do_valueize): Return OP if valueize
10420 (get_def): New helper to get at the def stmt of a SSA name
10421 if valueize allows.
10422 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
10423 do_valueize to get at the def stmt.
10424 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
10426 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
10428 PR middle-end/46932
10429 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
10431 2017-07-26 Martin Liska <mliska@suse.cz>
10434 * function.c (expand_function_start): Make expansion of
10435 nonlocal_goto_save_area after parm_birth_insn.
10437 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10439 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
10440 from all CPU target flags enable members.
10442 2017-07-26 Richard Biener <rguenther@suse.de>
10444 * genmatch.c (dt_simplify::gen): Make iterator vars const.
10445 (decision_tree::gen): Make 'type' const.
10446 (write_predicate): Likewise.
10448 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10450 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
10451 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
10452 (rs6000_option_override_internal): Likewise.
10453 (rs6000_expand_vector_set): Likewise.
10454 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
10455 (TARGET_UPPER_REGS_SF): Likewise.
10456 (TARGET_UPPER_REGS_DI): Likewise.
10457 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
10458 (TARGET_DIRECT_MOVE_64BIT): Likewise.
10459 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
10460 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10461 (Splitters for DI constants in Altivec registers): Likewise.
10462 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
10463 (vsx_set_v4sf_p9): Likewise.
10464 (vsx_set_v4sf_p9_zero): Likewise.
10465 (vsx_insert_extract_v4sf_p9): Likewise.
10466 (vsx_insert_extract_v4sf_p9_2): Likewise.
10468 2017-07-25 Carl Love <cel@us.ibm.com>
10470 * doc/extend.texi: Update the built-in documentation file for the
10471 existing built-in functions
10472 vector signed char vec_cnttz (vector signed char);
10473 vector unsigned char vec_cnttz (vector unsigned char);
10474 vector signed short vec_cnttz (vector signed short);
10475 vector unsigned short vec_cnttz (vector unsigned short);
10476 vector signed int vec_cnttz (vector signed int);
10477 vector unsigned int vec_cnttz (vector unsigned int);
10478 vector signed long long vec_cnttz (vector signed long long);
10479 vector unsigned long long vec_cnttz (vector unsigned long long);
10481 2017-07-25 Andrew Pinski <apinski@cavium.com>
10483 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
10484 accesses where the use is for the first operand of a BIT_INSERT.
10486 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10489 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10490 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10492 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10494 * config/i386/gstabs.h: Delete.
10495 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
10497 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
10499 * config/i386/i386.c (ix86_decompose_address): Do not check for
10500 register RTX when looking at index_reg or base_reg.
10501 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
10503 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
10505 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
10506 to update EH info here.
10508 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10510 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
10512 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10514 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
10516 2017-07-25 Torsten Duwe <duwe@suse.de>
10518 * common.opt: Introduce -fpatchable-function-entry
10519 command line option, and its variables function_entry_patch_area_size
10520 and function_entry_patch_area_start.
10521 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
10522 including a two-value parser.
10523 * target.def (print_patchable_function_entry): New target hook.
10524 * targhooks.h (default_print_patchable_function_entry): New function.
10525 * targhooks.c (default_print_patchable_function_entry): Likewise.
10526 * toplev.c (process_options): Switch off IPA-RA if
10527 patchable function entries are being generated.
10528 * varasm.c (assemble_start_function): Look at the
10529 patchable-function-entry command line switch and current
10530 function attributes and maybe generate NOP instructions by
10531 calling the print_patchable_function_entry hook.
10532 * doc/extend.texi: Document patchable_function_entry attribute.
10533 * doc/invoke.texi: Document -fpatchable_function_entry
10534 command line option.
10535 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
10537 * doc/tm.texi: Re-generate.
10539 2017-07-25 Jakub Jelinek <jakub@redhat.com>
10542 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
10543 TARGET_AVX512DQ rather than TARGET_AVX512BW.
10545 2017-07-25 Tamar Christina <tamar.christina@arm.com>
10547 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
10549 2017-07-25 Richard Biener <rguenther@suse.de>
10551 PR tree-optimization/81455
10552 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
10553 not walk in cycles when looking for guards.
10555 2017-07-25 Richard Biener <rguenther@suse.de>
10557 PR tree-optimization/81529
10558 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
10559 when optimizing backedge uses.
10561 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
10563 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
10565 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
10566 to dl_section_ref. On AIX, append an expression to subtract
10567 the size of the section length to dl_section_ref.
10569 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
10571 * configure.ac: If any of the config.* scripts fail, exit 1.
10572 * configure: Regenerate.
10574 2017-07-25 Richard Biener <rguenther@suse.de>
10576 PR middle-end/81546
10577 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
10578 of immediate uses, be more verbose on errors.
10580 2017-07-25 Richard Biener <rguenther@suse.de>
10582 PR tree-optimization/81510
10583 * tree-vect-loop.c (vect_is_simple_reduction): When the
10584 reduction stmt is not inside the loop bail out.
10586 2017-07-25 Richard Biener <rguenther@suse.de>
10588 PR tree-optimization/81303
10589 * tree-vect-loop-manip.c (vect_loop_versioning): Build
10590 profitability check against LOOP_VINFO_NITERSM1.
10592 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10594 * domwalk.c (cmp_bb_postorder): Simplify.
10595 (sort_bbs_postorder): New function. Use it...
10596 (dom_walker::walk): ...here to optimize common cases.
10598 2017-07-25 Martin Liska <mliska@suse.cz>
10601 * ipa-visibility.c (function_and_variable_visibility): Make the
10602 redirection just on target that supports aliasing.
10603 Fix GNU coding style.
10605 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
10608 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
10610 * config/arm/rtems.h: Likewise.
10611 * config/bfin/rtems.h: Likewise.
10612 * config/i386/rtemself.h: Likewise.
10613 * config/lm32/rtems.h: Likewise.
10614 * config/m32c/rtems.h: Likewise.
10615 * config/m68k/rtemself.h: Likewise.
10616 * config/microblaze/rtems.h: Likewise.
10617 * config/mips/rtems.h: Likewise.
10618 * config/moxie/rtems.h: Likewise.
10619 * config/nios2/rtems.h: Likewise.
10620 * config/powerpcspe/rtems.h: Likewise.
10621 * config/rs6000/rtems.h: Likewise.
10622 * config/rtems.h: Likewise.
10623 * config/sh/rtems.h: Likewise.
10624 * config/sh/rtemself.h: Likewise.
10625 * config/sparc/rtemself.h: Likewise.
10627 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10630 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
10631 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
10632 * tree-ssa-structalias.c (alias_get_name): Same.
10634 2017-07-25 Bin Cheng <bin.cheng@arm.com>
10637 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
10638 instructions if no du chain is found.
10640 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10642 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
10644 2017-07-25 Richard Biener <rguenther@suse.de>
10646 PR middle-end/81505
10647 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
10650 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10652 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
10653 upper-regs options.
10654 (ISA_2_7_MASKS_SERVER): Likewise.
10655 (ISA_3_0_MASKS_IEEE): Likewise.
10656 (OTHER_P8_VECTOR_MASKS): Likewise.
10657 (OTHER_VSX_VECTOR_MASKS): Likewise.
10658 (POWERPC_MASKS): Likewise.
10659 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
10660 duplicate list of options.
10661 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
10662 explicit -mupper-regs options.
10663 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
10664 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
10665 alias for -mupper-regs-df.
10666 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
10667 (rs6000_init_hard_regno_mode_ok): Likewise.
10668 (rs6000_option_override_internal): Likewise.
10669 (rs6000_opt_masks): Likewise.
10670 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
10671 options in terms of whether -mvsx or -mpower8-vector was used.
10672 (TARGET_UPPER_REGS_DI): Likewise.
10673 (TARGET_UPPER_REGS_SF): Likewise.
10674 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
10675 -mupper-regs-* options.
10677 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10679 * passes.c (emergency_dump_function): Print some empty lines and a
10680 header before the RTL dump.
10682 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10684 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
10686 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
10689 * config/aarch64/aarch64.c (aarch64_classify_symbol):
10690 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
10692 2017-07-24 Carl Love <cel@us.ibm.com>
10694 * config/rs6000/rs6000-c.c: Add support for built-in functions
10695 vector float vec_extract_fp32_from_shorth (vector unsigned short);
10696 vector float vec_extract_fp32_from_shortl (vector unsigned short);
10697 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
10698 vec_extract_fp_from_shortl): Add defines for the two builtins.
10699 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
10700 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
10702 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
10703 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
10704 * doc/extend.texi: Update the built-in documentation file for the
10705 new built-in function.
10707 2017-07-24 Jakub Jelinek <jakub@redhat.com>
10710 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
10712 * doc/generic.texi: Likewise.
10713 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10714 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10716 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
10718 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
10719 (aarch64_mls_elt_merge<mode>): Likewise.
10721 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
10723 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
10724 having __cxa_atexit.
10726 2017-07-23 Michael Collison <michael.collison@arm.com>
10728 * config/arm/arm.c (arm_option_override): Deprecate
10729 use of -mstructure-size-boundary.
10730 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
10731 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
10733 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10736 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
10737 Reduce cost estimate for direct moves.
10739 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
10742 * config/i386/i386.c (ix86_option_override_internal): Disable
10743 BMI, BMI2 and TBM instructions for -m16.
10745 2017-07-21 Carl Love <cel@us.ibm.com>
10747 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10748 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10749 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10750 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10751 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10752 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10753 VMULOSW): New enum "unspec" values.
10754 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10755 altivec_vmulosw): New patterns.
10756 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10757 VMULOSW): Add definitions.
10759 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
10761 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
10762 (qdf24xx): Likewise.
10763 * config/aarch64/aarch64-options-extensions.def (rdma); New.
10764 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
10765 (AARCH64_FL_V8_1): Renumber.
10766 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
10767 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
10768 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
10769 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
10770 rdma to feature modifiers list.
10772 2017-07-21 Yury Gribov <tetra2005@gmail.com>
10774 PR middle-end/56727
10775 * ipa-visibility (function_and_variable_visibility): Convert
10776 recursive PLT call to direct call if appropriate.
10778 2017-07-21 Andrew Pinski <apinski@cavium.com>
10780 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
10781 operand 1 to see if the types precision matches.
10782 * fold-const.c (operand_equal_p): Likewise.
10784 2017-07-21 Richard Biener <rguenther@suse.de>
10786 PR tree-optimization/81303
10787 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
10788 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
10789 (vect_peeling_hash_get_lowest_cost): Adjust.
10790 (vect_enhance_data_refs_alignment): Likewise. Use
10791 vect_get_peeling_costs_all_drs to compute the penalty for no
10792 peeling to match up costs.
10794 2017-07-21 Richard Biener <rguenther@suse.de>
10796 PR tree-optimization/81500
10797 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
10798 we didn't identify a reduction path.
10800 2017-07-21 Tom de Vries <tom@codesourcery.com>
10801 Cesar Philippidis <cesar@codesourcery.com>
10803 PR gcov-profile/81442
10804 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
10807 2017-07-21 Tom de Vries <tom@codesourcery.com>
10810 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
10812 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
10813 nvptx_override_options_after_change.
10815 2017-07-21 Ulrich Drepper <drepper@redhat.com>
10817 * dwarf2out.c (output_file_names): Avoid double testing for
10818 dwarf_version >= 5.
10820 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
10822 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
10824 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10826 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
10828 (try_crossjump_to_edge): Do not punt on partitioned functions.
10830 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
10832 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
10833 Put all BBs reachable only via paths crossing cold region to cold
10835 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
10837 2016-07-21 Richard Biener <rguenther@suse.de>
10839 PR tree-optimization/81303
10840 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
10841 into account prologue and epilogue iterations when raising
10842 min_profitable_iters to sth at least covering one vector iteration.
10844 2017-07-21 Tamar Christina <tamar.christina@arm.com>
10846 * config/arm/arm.c (arm_test_cpu_arch_dat):
10849 2017-07-20 Nathan Sidwell <nathan@acm.org>
10851 Remove TYPE_METHODS.
10852 * tree.h (TYPE_METHODS): Delete.
10853 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
10854 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
10855 (dbxout_type_methods): Scan TYPE_FIELDS.
10856 (dbxout_type): Don't check TYPE_METHODS here.
10857 * function.c (use_register_for_decl): Always ignore register for
10858 class types when not optimizing.
10859 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
10860 * tree.c (free_lang_data_in_type): Stitch out member functions and
10861 templates from TYPE_FIELDS.
10862 (build_distinct_type_copy, verify_type_variant,
10863 verify_type): Member fns are on TYPE_FIELDS.
10864 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
10865 * tree-pretty-print.c (dump_generic_node): Likewise.
10867 2017-07-20 Jakub Jelinek <jakub@redhat.com>
10870 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
10871 V2TImode and V4TImode.
10872 (ix86_expand_vector_extract): Likewise.
10873 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
10874 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
10875 (ssescalarmode): Handle V4TImode and V2TImode.
10876 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
10877 (*vec_extractv2ti, *vec_extractv4ti): New insns.
10878 (VEXTRACTI128_MODE): New mode iterator.
10879 (splitter for *vec_extractv?ti first element): New.
10880 (VEC_INIT_MODE): New mode iterator.
10881 (vec_init<mode>): Consolidate 3 expanders into one using
10882 VEC_INIT_MODE mode iterator.
10884 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10886 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
10887 non_spilled_static_chain_regno_p.
10889 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
10891 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
10893 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
10895 * bb-reorder.c (connect_traces): Allow copying of blocks within
10898 2017-07-20 Richard Biener <rguenther@suse.de>
10900 * gimple.h (gimple_phi_result): Add gphi * overload.
10901 (gimple_phi_result_ptr): Likewise.
10902 (gimple_phi_arg): Likewise. Adjust index assert to only
10903 allow actual argument accesses rather than all slots available
10905 (gimple_phi_arg_def): Add gphi * overload.
10906 * tree-phinodes.c (make_phi_node): Initialize only actual
10908 (resize_phi_node): Clear memory not covered by old node,
10909 do not initialize excess argument slots.
10910 (reserve_phi_args_for_new_edge): Initialize new argument slot
10913 2017-07-20 Bin Cheng <bin.cheng@arm.com>
10915 PR tree-optimization/81388
10917 2016-07-21 Bin Cheng <bin.cheng@arm.com>
10919 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
10920 by removing computation of may_be_zero.
10922 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
10923 Tom de Vries <tom@codesourcery.com>
10925 PR middle-end/81030
10926 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
10927 when gimple level profile disagrees with what RTL expander did.
10929 2017-07-20 Richard Biener <rguenther@suse.de>
10931 PR tree-optimization/61171
10932 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
10933 (vect_analyze_stmt): Add slp instance parameter.
10934 (vectorizable_reduction): Likewise.
10935 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
10936 (vect_is_simple_reduction): Deal with chains not detected
10937 as SLP reduction chain, specifically not properly associated
10938 chains containing a mix of plus/minus.
10939 (get_reduction_op): Remove.
10940 (get_initial_defs_for_reduction): Simplify, pass in whether
10941 this is a reduction chain, pass in the SLP node for the PHIs.
10942 (vect_create_epilog_for_reduction): Get the SLP instance as
10944 (vectorizable_reduction): Get the SLP instance as arg.
10945 During analysis remember the SLP node with the PHIs in the
10946 instance. Simplify getting at the vectorized reduction PHIs.
10947 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
10948 through SLP instance.
10949 (vect_slp_analyze_operations): Likewise.
10950 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
10951 (vect_transform_stmt): Likewise.
10953 2017-07-20 Tom de Vries <tom@codesourcery.com>
10955 PR tree-optimization/81489
10956 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
10957 read of phi arg location to before loop that modifies phi.
10959 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10961 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
10964 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10966 PR middle-end/81331
10967 * except.c (execute): Fix ordering issue.
10969 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10971 PR rtl-optimization/81423
10972 * combine.c (make_compound_operation_int): Don't try to optimize
10973 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
10975 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
10977 PR rtl-optimization/81423
10978 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
10979 with a constant that is -1 in the truncated to mode.
10981 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10983 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
10984 (determine_unlikely_bbs): ... here.
10985 * predict.h (propagate_unlikely_bbs_forward): Declare.
10986 * cfgexpand.c (pass_expand::execute): Use it.
10987 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
10989 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
10990 propagate_unlikely_bbs_forward.
10992 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
10994 PR middle-end/81331
10995 * except.c (maybe_add_nop_after_section_switch): New function.
10998 2017-07-19 Tom de Vries <tom@codesourcery.com>
11000 * gimple.h (gimple_phi_set_arg): Make assert more strict.
11002 2017-07-19 Tom de Vries <tom@codesourcery.com>
11004 * gimple.h (gimple_phi_arg): Make assert more strict.
11006 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
11008 * config.gcc (powerpc*-*-*): Add mmintrin.h.
11009 * config/rs6000/mmintrin.h: New file.
11010 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
11012 2017-07-19 Jakub Jelinek <jakub@redhat.com>
11014 PR tree-optimization/81346
11015 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
11017 2017-07-19 Tom de Vries <tom@codesourcery.com>
11019 * config/nvptx/nvptx.md (VECIM): Add V2DI.
11021 2017-07-19 Tom de Vries <tom@codesourcery.com>
11023 * config/nvptx/nvptx-modes.def: Add V2DImode.
11024 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
11025 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
11026 (nvptx_output_mov_insn): Handle lack of mov.b128.
11027 (nvptx_print_operand): Handle 'H' and 'L' codes.
11028 (nvptx_vector_mode_supported): Allow V2DImode.
11029 (nvptx_preferred_simd_mode): New function.
11030 (nvptx_data_alignment): New function.
11031 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
11032 nvptx_preferred_simd_mode.
11033 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
11035 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
11037 2017-07-19 Tom de Vries <tom@codesourcery.com>
11039 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
11040 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
11041 (nvptx_vector_mode_supported): New function. Allow V2SImode.
11042 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
11043 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
11044 (mov<VECIM>_insn): New define_insn.
11045 (define_expand "mov<VECIM>): New define_expand.
11047 2017-07-19 Tom de Vries <tom@codesourcery.com>
11049 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
11051 2017-07-19 Jakub Jelinek <jakub@redhat.com>
11053 PR tree-optimization/81346
11054 * fold-const.h (fold_div_compare, range_check_type): Declare.
11055 * fold-const.c (range_check_type): New function.
11056 (build_range_check): Use range_check_type.
11057 (fold_div_compare): No longer static, rewritten into
11058 a match.pd helper function.
11059 (fold_comparison): Don't call fold_div_compare here.
11060 * match.pd (X / C1 op C2): New optimization using fold_div_compare
11061 as helper function.
11063 2017-07-19 Nathan Sidwell <nathan@acm.org>
11065 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
11066 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
11067 * tree.c (find_decls_types_r, verify_type): Use
11068 TYPE_{MIN,MAX}_VALUE_RAW.
11069 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
11070 (hash_tree): Likewise.
11071 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
11073 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
11076 2017-07-18 Tom de Vries <tom@codesourcery.com>
11078 PR middle-end/81464
11079 * omp-expand.c (expand_omp_for_static_chunk): Handle
11080 equal-argument loop exit phi.
11082 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
11085 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
11086 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
11087 operand 2 predicate.
11088 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
11089 operand 2 predicate.
11090 (ror,rol -> rorx splitters): Use const_int_operand as
11091 operand 2 predicate.
11093 2017-06-18 Richard Biener <rguenther@suse.de>
11095 PR tree-optimization/81410
11096 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
11097 the gap in the ! slp_perm SLP case after each group.
11099 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11101 PR middle-end/81463
11102 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
11105 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11107 PR middle-end/81462
11108 * predict.c (set_even_probabilities): Cleanup; do not affect
11109 probabilities that are already known.
11110 (combine_predictions_for_bb): Call even when count is set.
11112 2017-07-18 Nathan Sidwell <nathan@acm.org>
11114 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
11117 2017-07-18 Bin Cheng <bin.cheng@arm.com>
11120 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
11121 optimization for loop niter analysis.
11123 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
11126 * config/avr/avr.c (avr_optimize_casesi): Don't use
11127 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
11129 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11131 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
11132 body_cost_vec from _vect_peel_extended_info.
11133 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
11134 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
11137 2017-07-18 Bin Cheng <bin.cheng@arm.com>
11139 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
11141 2017-07-18 Richard Biener <rguenther@suse.de>
11143 PR tree-optimization/80620
11144 PR tree-optimization/81403
11145 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
11146 info when re-using a VN table entry.
11148 2017-07-18 Richard Biener <rguenther@suse.de>
11150 PR tree-optimization/81418
11151 * tree-vect-loop.c (vectorizable_reduction): Properly compute
11152 vectype_in. Verify that with lane-reducing reduction operations
11153 we have a single def-use cycle.
11155 2017-07-17 Carl Love <cel@us.ibm.com>
11157 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
11159 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11160 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11161 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11162 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11163 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11164 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11165 VMULOSW): New enum "unspec" values.
11166 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11167 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11168 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11169 altivec_vmulosw): New patterns.
11170 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11171 VMULOSW): Add definitions.
11172 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
11174 * config/alpha/alpha.c: Include predict.h.
11176 2017-07-17 Yury Gribov <tetra2005@gmail.com>
11178 * tree-vrp.c (compare_assert_loc): Fix comparison function
11179 to return predictable results.
11181 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11183 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
11185 (subdi3): Likewise.
11186 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
11187 * doc/invoke.texi (mexpand-adddi): Update text.
11189 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11191 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
11192 that also clobbers the CC register. The old expand code is moved
11194 (*arc_clzsi2): ... here.
11195 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
11196 the CC register. The old expand code is moved to ...
11197 (arc_ctzsi2): ... here.
11199 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11201 * config/arc/arc.opt (mindexed-loads): Use initial value
11202 TARGET_INDEXED_LOADS_DEFAULT.
11203 (mauto-modify-reg): Use initial value
11204 TARGET_AUTO_MODIFY_REG_DEFAULT.
11205 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11206 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11207 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11208 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11210 2017-07-17 Martin Liska <mliska@suse.cz>
11213 * opts.c (finish_options): Do not allow -fgnu-tm
11214 w/ -fsanitize={kernel-,}address. Say sorry.
11216 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11219 * tree-loop-distribution.c (classify_partition): Only assert on
11220 numer of iterations.
11221 (merge_dep_scc_partitions): Delete prameter. Update function call.
11222 (distribute_loop): Remove code handling loop with unknown niters.
11223 (pass_loop_distribution::execute): Skip loop with unknown niters.
11225 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11228 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
11229 function sort_partitions_by_post_order.
11231 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11233 PR tree-optimization/81374
11234 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
11235 the max index of basic blocks, rather than number of basic blocks.
11237 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11239 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
11241 (arc_legitimate_pic_operand_p): Likewise.
11242 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
11244 (arc_needs_pcl_p): Likewise.
11245 (arc_legitimate_pc_offset_p): Likewise.
11246 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
11247 function is also used in constrains.md.
11248 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
11249 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
11250 PLUS. Only return true/false in known cases, otherwise assert.
11251 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
11252 is already called in arc_legitimate_constant_p.
11253 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
11255 (LEGITIMATE_PIC_OPERAND_P): Use
11256 arc_raw_symbolic_reference_mentioned_p function.
11257 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
11262 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11263 Andrew Burgess <andrew.burgess@embecosm.com>
11265 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
11266 (arc_return_address_register): New function.
11267 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
11268 (arc_handle_fndecl_attribute): Add naked attribute.
11269 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11270 (TARGET_WARN_FUNC_RETURN): Likewise.
11271 (arc_allocate_stack_slots_for_args): New function.
11272 (arc_warn_func_return): Likewise.
11273 (machine_function): Change type fn_type.
11274 (arc_compute_function_type): Consider new naked function type,
11275 change function return type.
11276 (arc_must_save_register): Adapt to handle new
11277 arc_compute_function_type's return type.
11278 (arc_expand_prologue): Likewise.
11279 (arc_expand_epilogue): Likewise.
11280 (arc_return_address_regs): Delete.
11281 (arc_return_address_register): New function.
11282 (arc_epilogue_uses): Use above function.
11283 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
11284 (arc_function_type): Change encoding, add naked type.
11285 (ARC_INTERRUPT_P): Change to handle the new encoding.
11286 (ARC_FAST_INTERRUPT_P): Likewise.
11287 (ARC_NORMAL_P): Define.
11288 (ARC_NAKED_P): Likewise.
11289 (arc_compute_function_type): Delete prototype.
11290 * config/arc/arc.md (in_ret_delay_slot): Use
11291 arc_return_address_register function.
11292 (simple_return): Likewise.
11293 (p_return_i): Likewise.
11295 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11297 PR tree-optimization/81428
11298 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
11299 can't be built for those types.
11301 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11303 Remove stuff dead since r239246.
11305 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
11306 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
11307 (avr_inform_devices): Remove dead stuff.
11309 2017-07-17 Tamar Christina <tamar.christina@arm.com>
11311 * config/arm/arm_neon.h: Fix softp typo.
11313 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11315 PR tree-optimization/81365
11316 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
11317 aggregate moves onto bb predecessor edges, make sure there are no
11318 loads that could alias the lhs in between the start of bb and the
11321 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11324 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
11325 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
11326 [LSHIFTRT, outer_code = TRUNCATE]: Same.
11328 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11330 PR tree-optimization/81396
11331 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
11332 (init_symbolic_number): Initialize it to 1.
11333 (perform_symbolic_merge): Add n_ops from both operands into the new
11335 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
11336 without base_addr as useless if they need more than one operation.
11337 (bswap_replace): Handle !bswap case for NULL base_addr.
11339 2017-07-17 Tom de Vries <tom@codesourcery.com>
11342 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
11345 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
11347 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
11348 conditional builtin define __FIX_LEON3FT_B2BST.
11350 2017-07-17 Daniel Cederman <cederman@gaisler.com>
11352 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
11353 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
11356 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11358 PR rtl-optimization/81424
11359 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
11360 to remove potential trapping from operands if -fnon-call-exceptions.
11362 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11364 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
11365 profile_proability for scalling.
11366 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
11368 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11370 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
11372 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11374 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
11375 fixpoint arithmetics.
11377 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11379 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
11380 fixpoint arithmetics.
11382 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11384 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
11385 fixpoint arithmetics.
11387 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11389 * profile-count.h (profile_probability::from_reg_br_prob_note,
11390 profile_probability::to_reg_br_prob_note): New functions.
11391 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
11392 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
11393 * predict.c (probability_reliable_p): Update.
11394 (edge_probability_reliable_p): Update.
11395 (br_prob_note_reliable_p): Update.
11396 (invert_br_probabilities): Update.
11397 (add_reg_br_prob_note): New function.
11398 (combine_predictions_for_insn): Update.
11399 * asan.c (asan_clear_shadow): Update.
11400 * cfgbuild.c (compute_outgoing_frequencies): Update.
11401 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
11402 (update_br_prob_note): Update.
11403 (rtl_verify_edges): Update.
11404 (purge_dead_edges): Update.
11405 (fixup_reorder_chain): Update.
11406 * emit-rtl.c (try_split): Update.
11407 * ifcvt.c (cond_exec_process_insns): Update.
11408 (cond_exec_process_if_block): Update.
11409 (dead_or_predicable): Update.
11410 * internal-fn.c (expand_addsub_overflow): Update.
11411 (expand_neg_overflow): Update.
11412 (expand_mul_overflow): Update.
11413 * loop-doloop.c (doloop_modify): Update.
11414 * loop-unroll.c (compare_and_jump_seq): Update.
11415 * optabs.c (emit_cmp_and_jump_insn_1): Update.
11416 * predict.h: Update.
11417 * reorg.c (mostly_true_jump): Update.
11419 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
11420 * config/alpha/alpha.c (emit_unlikely_jump): Update.
11421 * config/arc/arc.c: (emit_unlikely_jump): Update.
11422 * config/arm/arm.c: (emit_unlikely_jump): Update.
11423 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
11424 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
11425 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
11426 (ix86_print_operand): Update.
11427 (ix86_split_fp_branch): Update.
11428 (predict_jump): Update.
11429 * config/ia64/ia64.c (ia64_print_operand): Update.
11430 * config/mmix/mmix.c (mmix_print_operand): Update.
11431 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
11432 (rs6000_expand_split_stack_prologue): Update.
11433 * config/rs6000/rs6000.c: Update.
11434 * config/s390/s390.c (s390_expand_vec_strlen): Update.
11435 (s390_expand_vec_movstr): Update.
11436 (s390_expand_cs_tdsi): Update.
11437 (s390_expand_split_stack_prologue): Update.
11438 * config/sh/sh.c (sh_print_operand): Update.
11439 (expand_cbranchsi4): Update.
11440 (expand_cbranchdi4): Update.
11441 * config/sparc/sparc.c (output_v9branch): Update.
11442 * config/spu/spu.c (get_branch_target): Update.
11443 (ea_load_store_inline): Update.
11444 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
11445 * config/tilepro/tilepro.c: Update.
11447 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11449 * gimplify.c (mostly_copy_tree_r): Revert latest change.
11450 (gimplify_save_expr): Likewise.
11452 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11454 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
11456 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11458 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
11460 * timevar.def (TV_IPA_FNSUMMARY): Define.
11462 2017-07-16 Daniel Cederman <cederman@gaisler.com>
11464 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
11465 to back store errata sensitive sequence from being generated.
11466 (sqrtdf2_fix): Likewise.
11468 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11470 * tree-ssa-threadupdate.c (compute_path_counts,
11471 update_joiner_offpath_counts): Use profile_probability.
11473 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
11476 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11478 * config/arm/arm-c.c (arm_cpu_builtins): Define
11479 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11481 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
11483 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11484 array entries to represent __ieee128 versions of the
11485 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
11486 scalar_extract_sig, and scalar_insert_exp built-in functions.
11487 (altivec_resolve_overloaded_builtin): Add special case handling
11488 for the __builtin_scalar_insert_exp function, as represented by
11489 the P9V_BUILTIN_VEC_VSIEDP constant.
11490 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
11491 exponent support for __ieee128 argument.
11492 (VSESQP): Add scalar extract signature support for __ieee128
11494 (VSTDCNQP): Add scalar test negative support for __ieee128
11496 (VSIEQP): Add scalar insert exponent support for __int128 argument
11497 with __ieee128 result.
11498 (VSIEQPF): Add scalar insert exponent support for __ieee128
11499 argument with __ieee128 result.
11500 (VSTDCQP): Add scalar test data class support for __ieee128
11502 (VSTDCNQP): Add overload support for scalar test negative with
11503 __ieee128 argument.
11504 (VSTDCQP): Add overload support for scalar test data class
11505 __ieee128 argument.
11506 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
11507 UNSPEC_VSX_SXSIGDP.
11508 (UNSPEC_VSX_SIEXPQP): New constant.
11509 (xsxexpqp): New insn for VSX scalar extract exponent quad
11511 (xsxsigqp): New insn for VSX scalar extract significand quad
11513 (xsiexpqpf): New insn for VSX scalar insert exponent quad
11514 precision with floating point argument.
11515 (xststdcqp): New expand for VSX scalar test data class quad
11517 (xststdcnegqp): New expand for VSX scalar test negative quad
11519 (xststdcqp): New insn to match expansions for VSX scalar test data
11520 class quad precision and VSX scalar test negative quad precision.
11521 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
11522 special case operand checking to enforce that second operand of
11523 VSX scalar test data class with quad precision argument is a 7-bit
11525 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11526 prototypes and descriptions of __ieee128 versions of
11527 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
11528 scalar_test_data_class, and scalar_test_neg built-in functions.
11530 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11532 PR tree-optimization/81162
11533 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
11534 replace a negate with an add.
11536 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
11538 * doc/invoke.texi (arm/-mcpu): Document +crypto.
11540 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11542 * config/arm/arm-c.c (arm_cpu_builtins): Define
11543 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11545 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11547 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
11548 (armv8-r): Set ARM Cortex-R52 as default CPU.
11549 * config/arm/arm-tables.opt: Regenerate.
11550 * config/arm/arm-tune.md: Regenerate.
11551 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11553 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
11554 extension for -mcpu=cortex-r52.
11556 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11558 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
11559 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
11560 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
11561 (fp-armv8): Define it as FP_ARMv8 only.
11562 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
11563 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
11565 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
11566 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
11567 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
11568 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
11569 than TARGET_FPU_ARMV8.
11570 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
11571 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
11572 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
11574 * config/arm/neon.md (neon_vrint): Likewise.
11575 (neon_vcvt): Likewise.
11576 (neon_<fmaxmin_op><mode>): Likewise.
11577 (<fmaxmin><mode>3): Likewise.
11578 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
11579 * config/arm/predicates.md (arm_cond_move_operator): Check against
11580 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
11582 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
11584 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
11585 to top of function.
11587 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11589 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
11590 loop in comment with memset.
11592 2017-07-14 Martin Liska <mliska@suse.cz>
11594 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
11595 * dwarf2out.c (is_java): Remove the function.
11596 (output_pubname): Remove usage of the function.
11597 (lower_bound_default): Remove usage of DW_LANG_Java.
11598 (gen_compile_unit_die): Likewise.
11599 * gcc.c: Remove compiler defaults for .java and .zip files.
11600 * gimple-expr.c (remove_suffix): Change as there's no longer
11601 extension than 4-letter one.
11602 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
11603 (gimplify_save_expr): Likewise.
11604 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
11605 as it's possible even for other languages than Java.
11606 * langhooks.h (struct lang_hooks): Remove Java from a comment.
11607 * lto-opts.c (lto_write_options): Remove reference to Java.
11608 * opts.c (strip_off_ending): Update file extension handling.
11609 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
11610 * tree-eh.c (lower_resx): Likewise.
11611 * tree.c (free_lang_data_in_type): Remove dead code.
11612 (find_decls_types_r): Likewise.
11613 (build_common_builtin_nodes): Remove Java from a comment.
11614 (verify_type): Remove dead code.
11615 * varasm.c (assemble_external): Remove Java from a comment.
11617 2017-07-14 Martin Liska <mliska@suse.cz>
11619 * opts.c (finish_options): Add quotes.
11620 (common_handle_option): Likewise.
11622 2017-07-14 Martin Liska <mliska@suse.cz>
11624 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
11625 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
11626 Remove N_SO_PASCAL.
11627 * dwarf2out.c (lower_bound_default): Do not handle
11629 (gen_compile_unit_die): Likewise.
11630 * gcc.c: Remove default extension binding for GNU Pascal.
11631 * stmt.c: Remove Pascal language from a comment.
11632 * xcoffout.c: Likewise.
11634 2017-07-13 David Malcolm <dmalcolm@redhat.com>
11637 * diagnostic-show-locus.c (fixit_cmp): New function.
11638 (layout::layout): Sort m_fixit_hints.
11639 (column_range::column_range): Assert that the values are valid.
11640 (struct char_span): New struct.
11641 (correction::overwrite): New method.
11642 (struct source_line): New struct.
11643 (line_corrections::add_hint): Add assertions. Reimplement memcpy
11644 calls in terms of classes source_line and char_span, and
11645 correction::overwrite.
11646 (selftest::test_overlapped_fixit_printing_2): New function.
11647 (selftest::diagnostic_show_locus_c_tests): Call it.
11649 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11651 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
11652 early if there is no lhs.
11654 2017-07-13 Martin Liska <mliska@suse.cz>
11656 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
11657 (gen_reference_type_die): Likewise.
11658 * stor-layout.c: Remove Pascal-related comment.
11660 2017-07-13 Martin Liska <mliska@suse.cz>
11662 * opts.c (finish_options): Add quotes to error messages.
11663 (parse_sanitizer_options): Likewise.
11665 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
11667 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
11669 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
11671 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
11673 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
11675 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
11677 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
11679 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11682 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
11683 provides the hardware capability bits, define the macro
11684 __BUILTIN_CPU_SUPPORTS__.
11685 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
11686 if GLIBC does not provide the hardware capability bits. Add a
11687 gcc_unreachable call if the built-in cpu function is neither
11688 __builtin_cpu_is nor __builtin_cpu_supports.
11689 (rs6000_get_function_versions_dispatcher): Change the warning
11690 that an old GLIBC is used which does not export the capability
11691 bits to be an error.
11692 * doc/extend.texi (target_clones attribute): Document the
11693 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
11694 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
11695 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
11696 the macros defined by GCC if the newer GLIBC is available.
11698 2017-07-12 Jeff Law <law@redhat.com>
11700 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
11701 remaining includes slightly.
11702 * config/riscv/riscv-builtins.c: Include profile-count.h.
11704 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11707 * config/avr/avr.c (avr_set_current_function): In diagnostic
11708 messages: Quote keywords and (parts of) identifiers.
11709 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
11712 2017-07-12 Carl Love <cel@us.ibm.com>
11714 * config/rs6000/rs6000-c.c: Add support for built-in functions
11715 vector bool char vec_revb (vector bool char);
11716 vector bool short vec_revb (vector short char);
11717 vector bool int vec_revb (vector bool int);
11718 vector bool long long vec_revb (vector bool long long);
11719 * doc/extend.texi: Update the built-in documentation file for the
11720 new built-in functions.
11722 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11724 * config/s390/s390.md: Remove movcc splitter.
11726 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11728 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
11729 load/store on condition.
11731 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
11734 * config/avr/avr.c (avr_encode_section_info)
11735 [progmem && !TREE_READONLY]: Error if progmem object needs
11738 2017-07-11 Michael Collison <michael.collison@arm.com>
11740 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
11743 2017-07-11 Carl Love <cel@us.ibm.com>
11745 * config/rs6000/rs6000-c.c: Add support for builtins
11746 vector unsigned int vec_parity_lsbb (vector signed int);
11747 vector unsigned int vec_parity_lsbb (vector unsigned int);
11748 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
11749 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
11750 vector unsigned long long vec_parity_lsbb (vector signed long long);
11751 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
11752 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
11753 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
11754 * doc/extend.texi: Update the built-in documentation file for the
11755 new built-in functions.
11757 2017-07-11 David Malcolm <dmalcolm@redhat.com>
11759 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
11760 (layout::m_primary_loc): New field.
11761 (layout::layout): Initialize new field. Move location filtering
11762 logic from here to...
11763 (layout::maybe_add_location_range): ...this new method. Add
11764 support for filtering to just the lines already specified by other
11766 (layout::will_show_line_p): New method.
11767 (gcc_rich_location::add_location_if_nearby): New method.
11768 (selftest::test_add_location_if_nearby): New test function.
11769 (selftest::diagnostic_show_locus_c_tests): Call it.
11770 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
11773 2017-07-11 Tom de Vries <tom@codesourcery.com>
11775 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
11776 (bb_first_real_insn): New function.
11777 (nvptx_single): Add extra initialization of broadcasted condition
11780 2017-07-11 Nathan Sidwell <nathan@acm.org>
11782 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
11784 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
11786 * doc/extend.texi (AVR Function Attributes): Remove weblink to
11787 Binutils doc as TEXI will mess them up.
11788 * doc/invoke.texi (AVR Options): Same here.
11790 2017-07-11 Daniel Cederman <cederman@gaisler.com>
11792 * config/sparc/sparc.opt (mfix-ut700): New option.
11793 (mfix-gr712rc): Likewise.
11794 (sparc_fix_b2bst): New variable.
11795 * doc/invoke.texi (SPARC options): Document them.
11796 (ARM options): Fix warnings.
11797 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
11798 instructions to prevent sequences that can trigger the store-store
11799 errata for certain LEON3FT processors.
11800 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
11801 (sparc_option_override): Set sparc_fix_b2bst appropriately.
11802 * config/sparc/sparc.md (fix_b2bst): New attribute.
11803 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
11805 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
11808 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
11810 (*rsqrtsf2_sse): Ditto.
11812 (div<mode>3): Macroize insn from divdf3 and divsf3
11813 using MODEF mode iterator.
11815 2017-07-10 Martin Sebor <msebor@redhat.com>
11817 PR tree-optimization/80397
11818 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
11819 instead of testing for equality to INTEGER_TYPE.
11821 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
11823 * config.gcc: Remove uclibc from arc target spec.
11825 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
11827 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
11829 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11832 * lto-wrapper.c (remove_option): New function.
11833 (merge_and_complain): Merge PIC/PIE options more realistically.
11835 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11837 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
11841 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
11842 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
11843 * config.in: Regenerate.
11844 * configure: Regenerate.
11845 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
11846 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
11847 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
11848 (TARGET_GASISR_PROLOGUES): ...target mask.
11849 * common/config/avr/avr-common.c
11850 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
11851 Set -mgas-isr-prologues.
11852 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
11853 INSERT_PASS_BEFORE for it.
11854 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
11855 * config/avr/avr.c (avr_option_override)
11856 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
11857 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
11858 (avr_attribute_table) <no_gccisr>: Add new function attribute.
11859 (avr_set_current_function) <is_no_gccisr>: Init machine field.
11860 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
11862 (make_avr_pass_pre_proep): New function.
11863 (emit_push_sfr) <treg>: Add argument to function and use it
11864 instead of TMP_REG.
11865 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
11866 and set machine->gasisr.yes.
11867 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
11868 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
11869 __gcc_isr.n_pushed to .L__stack_usage.
11870 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
11871 (avr_asm_final_postscan_insn): ...this new static function.
11872 * config/avr/avr.h (machine_function)
11873 <is_no_gccisr, use_L__stack_usage>: New fields.
11874 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
11875 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
11876 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
11877 (gasisr, *gasisr): New expander and insn.
11878 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11879 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
11880 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
11882 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
11884 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
11887 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
11889 Move jump-tables out of .text again.
11892 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
11893 (ASM_OUTPUT_ADDR_VEC): New function.
11894 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
11895 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
11896 INSN_ADDRESSes as asm comment.
11897 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
11898 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
11899 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
11900 * config/avr/avr.md (*tablejump): Adjust comment.
11901 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
11902 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
11904 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
11905 (avr_output_addr_vec): New proto.
11906 (avr_log_t) <insn_addresses>: New field.
11908 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
11911 * config/i386/i386.c (ix86_function_arg_advance): Set
11912 outgoing_args_on_stack to true if there are outgoing arguments
11914 (ix86_function_arg): Likewise.
11915 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
11916 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
11917 * config/i386/i386.h (machine_function): Add
11918 outgoing_args_on_stack.
11920 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
11922 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11923 supporting pthreds.
11924 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
11926 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
11928 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
11929 (REAL_H): Remove $(MACHMODE_H).
11930 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
11932 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
11933 $(MACHMODE_H) and double-int.h.
11934 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
11936 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
11939 2017-07-07 Andrew Pinski <apinski@cavium.com>
11941 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
11942 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
11944 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
11946 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
11947 Add warning if GCC was not configured to link against a GLIBC that
11948 exports the hardware capability bits.
11949 (make_resolver_func): Make resolver function private and not a
11950 COMDAT function. Create the name with clone_function_name instead
11951 of make_unique_name.
11954 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
11955 correct operand in doing the split.
11957 2017-07-07 Carl Love <cel@us.ibm.com>
11959 * config/rs6000/rs6000-c: Add support for built-in function
11960 vector unsigned short vec_pack_to_short_fp32 (vector float,
11962 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
11963 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
11964 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
11965 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
11966 (convert_4f32_8i16): Add define_expand.
11967 * doc/extend.texi: Update the built-in documentation file for the
11968 new built-in function.
11970 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11972 * config/sparc/m8.md: New file.
11973 * config/sparc/sparc.md: Include m8.md.
11975 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
11977 * config/sparc/sparc.opt: New option -mvis4b.
11978 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
11979 (sparc_option_override): Handle VIS4B.
11980 (enum sparc_builtins): Define
11981 SPARC_BUILTIN_DICTUNPACK{8,16,32},
11982 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
11983 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
11984 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
11985 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
11986 (check_constant_argument): New function.
11987 (sparc_vis_init_builtins): Define builtins
11988 __builtin_vis_dictunpack{8,16,32},
11989 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
11990 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
11991 __builtin_vis_fpcmpde{8,16,32}shl and
11992 __builtin_vis_fpcmpur{8,16,32}shl.
11993 (sparc_expand_builtin): Check that the constant operands to
11994 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
11995 constant and in range.
11996 * config/sparc/sparc-c.c (sparc_target_macros): Handle
11998 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
11999 (SPARC_IMM5_P): Likewise.
12000 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
12001 (enabled): Handle vis4b.
12002 (UNSPEC_DICTUNPACK): New unspec.
12003 (UNSPEC_FPCMPSHL): Likewise.
12004 (UNSPEC_FPUCMPSHL): Likewise.
12005 (UNSPEC_FPCMPDESHL): Likewise.
12006 (UNSPEC_FPCMPURSHL): Likewise.
12007 (cpu_feature): New CPU feature `vis4b'.
12008 (dictunpack{8,16,32}): New insns.
12009 (FPCSMODE): New mode iterator.
12010 (fpcscond): New code iterator.
12011 (fpcsucond): Likewise.
12012 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
12013 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
12014 (fpcmpde{8,16,32}{si,di}shl): Likewise.
12015 (fpcmpur{8,16,32}{si,di}shl): Likewise.
12016 * config/sparc/constraints.md: Define constraints `q' for unsigned
12017 2-bit integer constants and `t' for unsigned 5-bit integer
12019 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
12021 (imm5_operand_dictunpack16): Likewise.
12022 (imm5_operand_dictunpack32): Likewise.
12023 (imm2_operand): Likewise.
12024 * doc/invoke.texi (SPARC Options): Document -mvis4b.
12025 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
12026 ditunpack* and fpcmp*shl builtins.
12028 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12030 * config.gcc: Handle m8 in --with-{cpu,tune} options.
12031 * config.in: Add HAVE_AS_SPARC6 define.
12032 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
12034 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
12036 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
12037 (CPP_CPU_SPEC): Handle m8.
12038 (ASM_CPU_SPEC): Likewise.
12039 * config/sparc/sparc-opts.h (enum processor_type): Add
12041 * config/sparc/sparc.c (m8_costs): New struct.
12042 (sparc_option_override): Handle TARGET_CPU_m8.
12043 (sparc32_initialize_trampoline): Likewise.
12044 (sparc64_initialize_trampoline): Likewise.
12045 (sparc_issue_rate): Likewise.
12046 (sparc_register_move_cost): Likewise.
12047 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
12048 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
12049 (ASM_CPU64_DEFAULT_SPEC): Likewise.
12050 (CPP_CPU_SPEC): Handle M8.
12051 (ASM_CPU_SPEC): Likewise.
12052 (AS_M8_FLAG): Define.
12053 * config/sparc/sparc.md: Add m8 to the cpu attribute.
12054 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
12055 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
12057 * configure: Regenerate.
12058 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
12061 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12063 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
12065 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
12066 ("*movdi_insn_sp32"): Do not set v3pipe.
12067 ("*movsi_insn"): Likewise.
12068 ("*movdi_insn_sp64"): Likewise.
12069 ("*movsf_insn"): Likewise.
12070 ("*movdf_insn_sp32"): Likewise.
12071 ("*movdf_insn_sp64"): Likewise.
12072 ("*zero_extendsidi2_insn_sp64"): Likewise.
12073 ("*sign_extendsidi2_insn"): Likewise.
12074 ("*mov<VM32:mode>_insn"): Likewise.
12075 ("*mov<VM64:mode>_insn_sp64"): Likewise.
12076 ("*mov<VM64:mode>_insn_sp32"): Likewise.
12077 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
12078 ("<vlop:code><VL:mode>3"): Likewise.
12079 ("*not_<vlop:code><VL:mode>3"): Likewise.
12080 ("*nand<VL:mode>_vis"): Likewise.
12081 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
12082 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
12083 ("one_cmpl<VL:mode>2"): Likewise.
12084 ("faligndata<VM64:mode>_vis"): Likewise.
12085 ("alignaddrsi_vis"): Likewise.
12086 ("alignaddrdi_vis"): Likweise.
12087 ("alignaddrlsi_vis"): Likewise.
12088 ("alignaddrldi_vis"): Likewise.
12089 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
12090 ("bmaskdi_vis"): Likewise.
12091 ("bmasksi_vis"): Likewise.
12092 ("bshuffle<VM64:mode>_vis"): Likewise.
12093 ("cmask8<P:mode>_vis"): Likewise.
12094 ("cmask16<P:mode>_vis"): Likewise.
12095 ("cmask32<P:mode>_vis"): Likewise.
12096 ("pdistn<P:mode>_vis"): Likewise.
12097 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
12099 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12101 * config/sparc/sparc.md ("subtype"): New insn attribute.
12102 ("*wrgsr_sp64"): Set insn subtype.
12103 ("*rdgsr_sp64"): Likewise.
12104 ("alignaddrsi_vis"): Likewise.
12105 ("alignaddrdi_vis"): Likewise.
12106 ("alignaddrlsi_vis"): Likewise.
12107 ("alignaddrldi_vis"): Likewise.
12108 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
12109 ("fexpand_vis"): Likewise.
12110 ("fpmerge_vis"): Likewise.
12111 ("faligndata<VM64:mode>_vis"): Likewise.
12112 ("bshuffle<VM64:mode>_vis"): Likewise.
12113 ("cmask8<P:mode>_vis"): Likewise.
12114 ("cmask16<P:mode>_vis"): Likewise.
12115 ("cmask32<P:mode>_vis"): Likewise.
12116 ("fchksm16_vis"): Likewise.
12117 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
12118 ("fmean16_vis"): Likewise.
12119 ("fp<plusminus_insn>64_vis"): Likewise.
12120 ("<plusminus_insn>v8qi3"): Likewise.
12121 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
12122 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
12123 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
12124 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
12125 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
12126 ("*movqi_insn"): Likewise.
12127 ("*movhi_insn"): Likewise.
12128 ("*movsi_insn"): Likewise.
12129 ("movsi_pic_gotdata_op"): Likewise.
12130 ("*movdi_insn_sp32"): Likewise.
12131 ("*movdi_insn_sp64"): Likewise.
12132 ("movdi_pic_gotdata_op"): Likewise.
12133 ("*movsf_insn"): Likewise.
12134 ("*movdf_insn_sp32"): Likewise.
12135 ("*movdf_insn_sp64"): Likewise.
12136 ("*zero_extendhisi2_insn"): Likewise.
12137 ("*zero_extendqihi2_insn"): Likewise.
12138 ("*zero_extendqisi2_insn"): Likewise.
12139 ("*zero_extendqidi2_insn"): Likewise.
12140 ("*zero_extendhidi2_insn"): Likewise.
12141 ("*zero_extendsidi2_insn_sp64"): Likewise.
12142 ("ldfsr"): Likewise.
12143 ("prefetch_64"): Likewise.
12144 ("prefetch_32"): Likewise.
12145 ("tie_ld32"): Likewise.
12146 ("tie_ld64"): Likewise.
12147 ("*tldo_ldub_sp32"): Likewise.
12148 ("*tldo_ldub1_sp32"): Likewise.
12149 ("*tldo_ldub2_sp32"): Likewise.
12150 ("*tldo_ldub_sp64"): Likewise.
12151 ("*tldo_ldub1_sp64"): Likewise.
12152 ("*tldo_ldub2_sp64"): Likewise.
12153 ("*tldo_ldub3_sp64"): Likewise.
12154 ("*tldo_lduh_sp32"): Likewise.
12155 ("*tldo_lduh1_sp32"): Likewise.
12156 ("*tldo_lduh_sp64"): Likewise.
12157 ("*tldo_lduh1_sp64"): Likewise.
12158 ("*tldo_lduh2_sp64"): Likewise.
12159 ("*tldo_lduw_sp32"): Likewise.
12160 ("*tldo_lduw_sp64"): Likewise.
12161 ("*tldo_lduw1_sp64"): Likewise.
12162 ("*tldo_ldx_sp64"): Likewise.
12163 ("*mov<VM32:mode>_insn"): Likewise.
12164 ("*mov<VM64:mode>_insn_sp64"): Likewise.
12165 ("*mov<VM64:mode>_insn_sp32"): Likewise.
12167 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12169 * config/sparc/sparc.md ("type"): New insn type viscmp.
12170 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
12172 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
12173 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
12174 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
12175 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
12177 ("n7_vis_logical_11cycle"): Likewise.
12178 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
12179 * config/sparc/niagara2.md ("niag3_vis": Likewise.
12180 * config/sparc/niagara.md ("niag_vis"): Likewise.
12181 * config/sparc/ultra3.md ("us3_fga"): Likewise.
12182 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
12184 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12186 * config/sparc/sparc.md: New instruction type `bmask'.
12187 (bmaskdi_vis): Use the `bmask' type.
12188 (bmasksi_vis): Likewise.
12189 * config/sparc/ultra3.md (us3_array): Likewise.
12190 * config/sparc/niagara7.md (n7_array): Likewise.
12191 * config/sparc/niagara4.md (n4_array): Likewise.
12192 * config/sparc/niagara2.md (niag2_vis): Likewise.
12193 (niag3_vis): Likewise.
12194 * config/sparc/niagara.md (niag_vis): Likewise.
12196 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12198 * ipa-comdats.c: Remove optimize check from gate.
12199 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
12200 for functions not optimized.
12201 (ipa_fn_summary_read): Skip optimize check.
12202 (ipa_fn_summary_write): Likewise.
12203 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
12205 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
12207 (can_inline_edge_p): Check flag_pcc_struct_return for match.
12208 (check_callers): Give up on caller which is not optimized.
12209 (inline_small_functions): Likewise.
12210 (ipa_inline): Do not give up when not optimizing.
12211 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
12212 away unoptimizes cdtors.
12213 (whole_program_function_and_variable_visibility): Do
12214 ipa_discover_readonly_nonaddressable_vars in LTO mode.
12215 * ipa.c (process_references): Do not check optimize.
12216 (symbol_table::remove_unreachable_nodes): Update optimize check.
12217 (set_writeonly_bit): Update optimize check.
12218 (pass_ipa_cdtor_merge::gate): Do not check optimize.
12219 (pass_ipa_single_use::gate): Remove.
12221 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12223 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
12224 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
12225 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
12226 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
12227 permute_load, permute_store, adjust_extract, adjust_splat,
12228 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
12229 replace_swap_with_copy, dump_swap_insn_table,
12230 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
12231 recombine_lvx_pattern, recombine_stvx_pattern,
12232 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
12233 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
12234 to file rs6000-p8swap.c.
12235 * config/rs6000/rs6000-p8swap.c: New file.
12236 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
12237 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
12238 and rs6000*-*-* targets.
12240 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12242 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
12244 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12246 * lto-wrapper.c (merge_and_complain): Do not merge
12247 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
12248 fsigned_zeros, ftrapping_math, fwrapv.
12249 (append_compiler_options): Do not track these options.
12250 (append_linker_options): Likewie
12252 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12254 * cgraphunit.c (cgraph_node::finalize_function): When
12255 !flag_toplevel_reorde set no_reorder flag.
12256 (varpool_node::finalize_decl): Likewise.
12257 (symbol_table::compile): Drop no toplevel reorder path.
12259 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12261 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
12262 edges; zero probability is not better than uninitialized.
12264 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
12266 * asan.h (asan_sanitize_allocas_p): Declare.
12267 * asan.c (asan_sanitize_allocas_p): New function.
12268 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
12269 (handle_builtin_alloca): Likewise.
12270 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
12271 if !asan_sanitize_allocas_p.
12272 * params.def (asan-instrument-allocas): Add new option.
12273 * params.h (ASAN_PROTECT_ALLOCAS): Define.
12274 * opts.c (common_handle_option): Disable allocas sanitization for
12277 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
12279 * asan.c: Include gimple-fold.h.
12280 (get_last_alloca_addr): New function.
12281 (handle_builtin_stackrestore): Likewise.
12282 (handle_builtin_alloca): Likewise.
12283 (asan_emit_allocas_unpoison): Likewise.
12284 (get_mem_refs_of_builtin_call): Add new parameter, remove const
12285 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
12286 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
12287 (instrument_builtin_call): Pass gimple iterator to
12288 get_mem_refs_of_builtin_call.
12289 (last_alloca_addr): New global.
12290 * asan.h (asan_emit_allocas_unpoison): Declare.
12291 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
12292 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
12293 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
12294 if function calls alloca.
12295 * gimple-fold.c (replace_call_with_value): Remove static keyword.
12296 * gimple-fold.h (replace_call_with_value): Declare.
12297 * internal-fn.c: Include asan.h.
12298 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
12299 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
12301 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12303 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
12304 (C_SELFTEST_FLAGS): New.
12305 (CPP_SELFTEST_FLAGS): New.
12306 (SELFTEST_DEPS): New, from deps of s-selftest.
12307 (C_SELFTEST_DEPS): New, from deps of s-selftest.
12308 (CPP_SELFTEST_DEPS): New.
12309 (selftest): Add dependency on s-selftest-c++.
12310 (s-selftest): Rename to...
12311 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
12312 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
12313 than SELFTEST_FLAGS.
12314 (selftest-gdb): Rename to...
12315 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
12317 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
12318 (selftest-valgrind): Rename to...
12319 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
12321 (selftest-valgrind): Reintroduce as an alias for
12322 selftest-c-valgrind.
12323 (s-selftest-c++): New.
12324 (selftest-c++-gdb): New.
12325 (selftest-c++-valgrind): New.
12327 2017-07-06 Olivier Hainque <hainque@adacore.com>
12329 * gcc.c (process_command): When deciding if undefined variables
12330 should be ignored when processing specs, accept "gcc -v" as well.
12332 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12334 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
12335 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
12337 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12339 * config/arm/arm-cpus.in (armv8-r): Add new entry.
12340 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
12341 * config/arm/arm-tables.opt: Regenerate.
12342 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
12344 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
12346 2017-07-06 Carl Love <cel@us.ibm.com>
12348 * ChangeLog: Clean up from mid air collision
12350 2017-07-06 Carl Love <cel@us.ibm.com>
12352 * config/rs6000/rs6000-c.c: Add support for built-in functions
12353 vector signed int vec_subc (vector signed int, vector signed int);
12354 vector signed __int128 vec_subc (vector signed __int128,
12355 vector signed __int128);
12356 vector unsigned __int128 vec_subc (vector unsigned __int128,
12357 vector unsigned __int128);
12358 vector signed int vec_sube (vector signed int, vector signed int,
12359 vector signed int);
12360 vector unsigned int vec_sube (vector unsigned int,
12361 vector unsigned int,
12362 vector unsigned int);
12363 vector signed __int128 vec_sube (vector signed __int128,
12364 vector signed __int128,
12365 vector signed__int128);
12366 vector unsigned __int128 vec_sube (vector unsigned __int128,
12367 vector unsigned __int128,
12368 vector unsigned __int128);
12369 vector signed int vec_subec (vector signed int, vector signed int,
12370 vector signed int);
12371 vector unsigned int vec_subec (vector unsigned int,
12372 vector unsigned int,
12373 vector unsigned int);
12374 vector signed __int128 vec_subec (vector signed __int128,
12375 vector signed __int128,
12376 vector signed__int128);
12377 vector unsigned __int128 vec_subec (vector unsigned __int128,
12378 vector unsigned __int128,
12379 vector unsigned __int128);
12380 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
12381 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
12382 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
12383 BU_ALTIVEC_OVERLOAD_X definitions.
12384 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
12385 * doc/extend.texi: Update the built-in documentation file for the new
12386 built-in functions.
12388 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12391 * diagnostic-show-locus.c (layout::layout): Use start and finish
12392 spelling location for the start and finish of each range.
12393 * genmatch.c (linemap_client_expand_location_to_spelling_point):
12394 Add unused aspect param.
12395 * input.c (expand_location_1): Add "aspect" param, and use it
12396 to access the correct part of the location.
12397 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
12399 (expand_location_to_spelling_point): Likewise.
12400 (linemap_client_expand_location_to_spelling_point): Add "aspect"
12401 param, and pass it to expand_location_1.
12403 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
12405 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
12406 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
12407 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
12408 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
12409 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
12410 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
12411 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
12412 _mm_maskz_getmant_ss): New intrinsics.
12413 (__builtin_ia32_getexpss128_mask): Changed to ...
12414 __builtin_ia32_getexpss128_round ... this.
12415 (__builtin_ia32_getexpsd128_mask): Changed to ...
12416 __builtin_ia32_getexpsd128_round ... this.
12417 * config/i386/i386-builtin-types.def
12418 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
12419 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
12420 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
12421 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
12422 __builtin_ia32_getmantss_mask_round): New builtins.
12423 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
12424 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
12425 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
12426 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
12427 * config/i386/sse.md
12428 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
12429 avx512f_sgetexp<mode><mask_scalar_name>
12430 <round_saeonly_scalar_name> ... this.
12431 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
12432 %0, %1, %2<round_saeonly_op3>}): Changed to ...
12433 vgetexp<ssescalarmodesuffix>
12434 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12435 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
12436 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
12437 avx512f_vgetmant<mode><mask_scalar_name>
12438 <round_saeonly_scalar_name> ... this.
12439 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
12440 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
12441 vgetmant<ssescalarmodesuffix>
12442 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
12443 %0<mask_scalar_operand4>, %1, %2
12444 <round_saeonly_scalar_mask_op4>, %3} ... this.
12445 * config/i386/subst.md (mask_scalar_operand4,
12446 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
12447 round_saeonly_scalar_nimm_predicate): New subst attributes.
12449 2017-07-06 Julia Koval <julia.koval@intel.com>
12451 * config/i386/i386.c (ix86_erase_embedded_rounding):
12452 Remove code for old rounding pattern.
12454 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
12456 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
12458 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
12460 * doc/sourcebuild.texi (Test Directives, Variants of
12461 dg-require-support): Add documentation for dg-require-stack-check.
12463 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
12465 * config/i386/subst.md (mask_scalar, round_scalar,
12466 round_saeonly_scalar): New meta-templates.
12467 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
12468 round_scalar_mask_operand3, round_scalar_mask_op3,
12469 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
12470 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
12471 round_saeonly_scalar_constraint,
12472 round_saeonly_scalar_prefix): New subst attribute.
12473 * config/i386/sse.md
12474 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
12475 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
12476 <round_scalar_name> ... this.
12477 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
12478 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
12479 <round_scalar_name> ... this.
12480 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
12481 <sse>_vm<code><mode>3<mask_scalar_name>
12482 <round_saeonly_scalar_name> ... this.
12483 (v<plusminus_mnemonic><ssescalarmodesuffix>
12484 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12485 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12486 v<plusminus_mnemonic><ssescalarmodesuffix>
12487 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12488 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12489 (v<multdiv_mnemonic><ssescalarmodesuffix>
12490 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12491 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12492 v<multdiv_mnemonic><ssescalarmodesuffix>
12493 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12494 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12495 (v<maxmin_float><ssescalarmodesuffix>
12496 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
12497 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
12498 v<maxmin_float><ssescalarmodesuffix>
12499 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12500 %0<mask_scalar_operand3>, %1, %<iptr>2
12501 <round_saeonly_scalar_mask_op3>} ... this.
12503 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
12505 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
12506 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
12508 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
12509 Alan Hayward <alan.hayward@arm.com>
12510 David Sherwood <david.sherwood@arm.com>
12512 * combine.c (simplify_if_then_else): Remove "enum" before
12514 * compare-elim.c (can_eliminate_compare): Likewise.
12515 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12517 (aarch64_lookup_simd_builtin_type): Likewise.
12518 (aarch64_simd_builtin_type): Likewise.
12519 (aarch64_init_simd_builtin_types): Likewise.
12520 (aarch64_simd_expand_args): Likewise.
12521 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
12523 (aarch64_reverse_mask): Likewise.
12524 (aarch64_simd_emit_reg_reg_move): Likewise.
12525 (aarch64_gen_adjusted_ldpstp): Likewise.
12526 (aarch64_ccmp_mode_to_code): Likewise.
12527 (aarch64_operands_ok_for_ldpstp): Likewise.
12528 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12529 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
12531 (aarch64_min_divisions_for_recip_mul): Likewise.
12532 (aarch64_reassociation_width): Likewise.
12533 (aarch64_get_condition_code_1): Likewise.
12534 (aarch64_simd_emit_reg_reg_move): Likewise.
12535 (aarch64_simd_attr_length_rglist): Likewise.
12536 (aarch64_reverse_mask): Likewise.
12537 (aarch64_operands_ok_for_ldpstp): Likewise.
12538 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12539 (aarch64_gen_adjusted_ldpstp): Likewise.
12540 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
12542 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
12543 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12544 (arm_lookup_simd_builtin_type): Likewise.
12545 (arm_simd_builtin_type): Likewise.
12546 (arm_init_simd_builtin_types): Likewise.
12547 (arm_expand_builtin_args): Likewise.
12548 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
12549 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
12550 (ft32_setup_incoming_varargs): Likewise.
12551 (ft32_function_arg): Likewise.
12552 (ft32_function_arg_advance): Likewise.
12553 (ft32_pass_by_reference): Likewise.
12554 (ft32_arg_partial_bytes): Likewise.
12555 (ft32_valid_pointer_mode): Likewise.
12556 (ft32_addr_space_pointer_mode): Likewise.
12557 (ft32_addr_space_legitimate_address_p): Likewise.
12558 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
12560 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
12561 (ix86_emit_outlined_ms2sysv_restore): Likewise.
12562 (iamcu_alignment): Likewise.
12563 (canonicalize_vector_int_perm): Likewise.
12564 (ix86_noce_conversion_profitable_p): Likewise.
12565 (ix86_mpx_bound_mode): Likewise.
12566 (ix86_operands_ok_for_move_multiple): Likewise.
12567 * config/microblaze/microblaze-protos.h
12568 (microblaze_expand_conditional_branch_reg): Likewise.
12569 * config/microblaze/microblaze.c
12570 (microblaze_expand_conditional_branch_reg): Likewise.
12571 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12573 (rs6000_reassociation_width): Likewise.
12574 (rs6000_invalid_binary_op): Likewise.
12575 (fusion_p9_p): Likewise.
12576 (emit_fusion_p9_load): Likewise.
12577 (emit_fusion_p9_store): Likewise.
12578 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
12580 (riscv_hard_regno_mode_ok_p): Likewise.
12581 (riscv_address_insns): Likewise.
12582 (riscv_split_symbol): Likewise.
12583 (riscv_legitimize_move): Likewise.
12584 (riscv_function_value): Likewise.
12585 (riscv_hard_regno_nregs): Likewise.
12586 (riscv_expand_builtin): Likewise.
12587 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
12588 (riscv_build_integer): Likewise.
12589 (riscv_split_integer): Likewise.
12590 (riscv_legitimate_constant_p): Likewise.
12591 (riscv_cannot_force_const_mem): Likewise.
12592 (riscv_regno_mode_ok_for_base_p): Likewise.
12593 (riscv_valid_base_register_p): Likewise.
12594 (riscv_valid_offset_p): Likewise.
12595 (riscv_valid_lo_sum_p): Likewise.
12596 (riscv_classify_address): Likewise.
12597 (riscv_legitimate_address_p): Likewise.
12598 (riscv_address_insns): Likewise.
12599 (riscv_load_store_insns): Likewise.
12600 (riscv_force_binary): Likewise.
12601 (riscv_split_symbol): Likewise.
12602 (riscv_force_address): Likewise.
12603 (riscv_legitimize_address): Likewise.
12604 (riscv_move_integer): Likewise.
12605 (riscv_legitimize_const_move): Likewise.
12606 (riscv_legitimize_move): Likewise.
12607 (riscv_address_cost): Likewise.
12608 (riscv_subword): Likewise.
12609 (riscv_output_move): Likewise.
12610 (riscv_canonicalize_int_order_test): Likewise.
12611 (riscv_emit_int_order_test): Likewise.
12612 (riscv_function_arg_boundary): Likewise.
12613 (riscv_pass_mode_in_fpr_p): Likewise.
12614 (riscv_pass_fpr_single): Likewise.
12615 (riscv_pass_fpr_pair): Likewise.
12616 (riscv_get_arg_info): Likewise.
12617 (riscv_function_arg): Likewise.
12618 (riscv_function_arg_advance): Likewise.
12619 (riscv_arg_partial_bytes): Likewise.
12620 (riscv_function_value): Likewise.
12621 (riscv_pass_by_reference): Likewise.
12622 (riscv_setup_incoming_varargs): Likewise.
12623 (riscv_print_operand): Likewise.
12624 (riscv_elf_select_rtx_section): Likewise.
12625 (riscv_save_restore_reg): Likewise.
12626 (riscv_for_each_saved_reg): Likewise.
12627 (riscv_register_move_cost): Likewise.
12628 (riscv_hard_regno_mode_ok_p): Likewise.
12629 (riscv_hard_regno_nregs): Likewise.
12630 (riscv_class_max_nregs): Likewise.
12631 (riscv_memory_move_cost): Likewise.
12632 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
12633 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
12634 (rl78_addr_space_address_mode): Likewise.
12635 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12637 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12638 (rs6000_reassociation_width): Likewise.
12639 (rs6000_invalid_binary_op): Likewise.
12640 (fusion_p9_p): Likewise.
12641 (emit_fusion_p9_load): Likewise.
12642 (emit_fusion_p9_store): Likewise.
12643 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
12644 (ok_for_simple_move_operands): Likewise.
12645 (ok_for_simple_move_strict_operands): Likewise.
12646 (ok_for_simple_arith_logic_operands): Likewise.
12647 (visium_legitimize_reload_address): Likewise.
12648 (visium_select_cc_mode): Likewise.
12649 (output_cbranch): Likewise.
12650 (visium_split_double_move): Likewise.
12651 (visium_expand_copysign): Likewise.
12652 (visium_expand_int_cstore): Likewise.
12653 (visium_expand_fp_cstore): Likewise.
12654 * config/visium/visium.c (visium_pass_by_reference): Likewise.
12655 (visium_function_arg): Likewise.
12656 (visium_function_arg_advance): Likewise.
12657 (visium_libcall_value): Likewise.
12658 (visium_setup_incoming_varargs): Likewise.
12659 (visium_legitimate_constant_p): Likewise.
12660 (visium_legitimate_address_p): Likewise.
12661 (visium_legitimize_address): Likewise.
12662 (visium_secondary_reload): Likewise.
12663 (visium_register_move_cost): Likewise.
12664 (visium_memory_move_cost): Likewise.
12665 (prepare_move_operands): Likewise.
12666 (ok_for_simple_move_operands): Likewise.
12667 (ok_for_simple_move_strict_operands): Likewise.
12668 (ok_for_simple_arith_logic_operands): Likewise.
12669 (visium_function_value_1): Likewise.
12670 (rtx_ok_for_offset_p): Likewise.
12671 (visium_legitimize_reload_address): Likewise.
12672 (visium_split_double_move): Likewise.
12673 (visium_expand_copysign): Likewise.
12674 (visium_expand_int_cstore): Likewise.
12675 (visium_expand_fp_cstore): Likewise.
12676 (visium_split_cstore): Likewise.
12677 (visium_select_cc_mode): Likewise.
12678 (visium_split_cbranch): Likewise.
12679 (output_cbranch): Likewise.
12680 (visium_print_operand_address): Likewise.
12681 * expmed.c (flip_storage_order): Likewise.
12682 * expmed.h (emit_cstore): Likewise.
12683 (flip_storage_order): Likewise.
12684 * genrecog.c (validate_pattern): Likewise.
12685 * hsa-gen.c (gen_hsa_addr): Likewise.
12686 * internal-fn.c (expand_arith_overflow): Likewise.
12687 * ira-color.c (allocno_copy_cost_saving): Likewise.
12688 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12689 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
12690 (process_invariant_for_inheritance): Likewise.
12691 * lra-eliminations.c (move_plus_up): Likewise.
12692 * omp-low.c (lower_oacc_reductions): Likewise.
12693 * simplify-rtx.c (simplify_subreg): Likewise.
12694 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
12695 (TARGET_CHKP_BOUND_MODE): Likewise..
12696 * targhooks.c (default_chkp_bound_mode): Likewise.
12697 (default_setup_incoming_vararg_bounds): Likewise.
12698 * targhooks.h (default_chkp_bound_mode): Likewise.
12699 (default_setup_incoming_vararg_bounds): Likewise.
12700 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
12701 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
12702 (have_whole_vector_shift): Likewise.
12703 * tree-vect-stmts.c (vectorizable_load): Likewise.
12704 * doc/tm.texi: Regenerate.
12706 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12708 Graceful degrade if Binutils PR21472 is not available.
12711 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
12712 .rodata in flash test fails.
12713 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
12714 * confgure: Regenerate.
12715 * config.in: Regenerate.
12716 * config/avr/avr.c (avr_asm_named_section)
12717 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
12718 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
12719 (avr_asm_init_sections): Same.
12721 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
12723 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
12724 (fma<VH:mode>4_intrinsic): Likewise.
12725 (*fmsub<VCVTF:mode>4): Likewise.
12726 (*fmsub<VH:mode>4_intrinsic): Likewise.
12728 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
12731 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
12732 Don't depend on "optimize > 0".
12733 (out_movhi_r_mr, out_movqi_mr_r): Same.
12734 (out_movhi_mr_r, out_movqi_r_mr): Same.
12735 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
12736 io_address_operand on "optimize > 0".
12738 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12740 * tree-loop-distribution.c: Add general explanantion on the pass.
12741 (generate_loops_for_partition): Mark distributed loop.
12742 (pg_add_dependence_edges): New parameter. Handle alias data
12743 dependence specially and record it in the parameter if asked.
12744 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
12745 (init_partition_graph_vertices, add_partition_graph_edge): New.
12746 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
12747 (free_partition_graph_vdata, build_partition_graph): New.
12748 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
12749 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
12750 (data_ref_segment_size, latch_dominated_by_data_ref): New.
12751 (compute_alias_check_pairs, version_loop_by_alias_check): New.
12752 (version_for_distribution_p, finalize_partitions): New.
12753 (distribute_loop): Handle alias data dependence specially. Factor
12754 out loop fusion code as functions and call these functions.
12756 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12758 * tree-loop-distribution.c (classify_partition): New parameter and
12759 better handle reduction statement.
12760 (rdg_build_partitions): Revise comment.
12761 (distribute_loop): Compute statements in all partitions and pass it
12762 to classify_partition.
12764 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12766 * tree-loop-distribution.c (enum partition_type): New.
12767 (struct partition): New field type.
12768 (partition_merge_into): Add parameter. Update partition type.
12769 (data_dep_in_cycle_p, update_type_for_merge): New functions.
12770 (build_rdg_partition_for_vertex): Compute partition type.
12771 (rdg_build_partitions): Dump partition type.
12772 (distribute_loop): Update calls to partition_merge_into.
12774 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12776 * tree-loop-distribution.c (struct ddr_hasher): New.
12777 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
12779 (classify_partition): Call get_data_dependence.
12780 (pg_add_dependence_edges): Ditto.
12781 (distribute_loop): Release data dependence hash table.
12783 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12785 * tree-loop-distribution.c (ref_base_address): Delete.
12786 (similar_memory_accesses): Rename ...
12787 (share_memory_accesses): ... to this. Check if partitions access
12788 the same memory reference.
12789 (distribute_loop): Call share_memory_accesses.
12791 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12793 * tree-loop-distribution.c (struct partition): New field recording
12794 its data reference.
12795 (partition_alloc, partition_free): Init and release data refs.
12796 (partition_merge_into): Merge data refs.
12797 (build_rdg_partition_for_vertex): Collect data refs for partition.
12798 (pg_add_dependence_edges): Change parameters from vector to bitmap.
12800 (distribute_loop): Remve data refs from vertice data of partition
12803 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12805 * tree-loop-distribution.c (params.h): Include header file.
12806 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
12807 (datarefs_vec): New global var.
12808 (create_rdg_vertices): Use datarefs_vec directly.
12809 (free_rdg): Don't free data references.
12810 (build_rdg): Update use. Don't free data references.
12811 (distribute_loop): Compute global variable for data references.
12812 Bail out if there are too many data references.
12814 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12816 * tree-loop-distribution.c (loop_nest): New global var.
12817 (build_rdg): Use loop directly, rather than loop nest.
12818 (pg_add_dependence_edges): Remove loop nest parameter. Use global
12820 (distribute_loop): Compute global variable loop nest. Update use.
12822 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12824 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
12825 (partition_merge_into): New parameter. Dump reason for fusion.
12826 (distribute_loop): Update use of partition_merge_into.
12828 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12830 * tree-loop-distribution.c (bb_top_order_index): New.
12831 (bb_top_order_index_size, bb_top_order_cmp): New.
12832 (stmts_from_loop): Use topological order.
12833 (pass_loop_distribution::execute): Compute and release topological
12834 order for basic blocks.
12836 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12838 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
12841 2017-07-05 Bin Cheng <bin.cheng@arm.com>
12843 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
12844 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
12845 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
12846 * internal-fn.def (LOOP_DIST_ALIAS): New.
12847 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
12848 (fold_loop_internal_call): ... this.
12849 (vect_loop_dist_alias_call): New function.
12850 (set_uid_loop_bbs): Call fold_loop_internal_call.
12851 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
12854 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12857 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
12858 Require dead FLAGS_REG at the beginning of a peephole.
12860 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
12863 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
12864 arguments in the call to __builtin_ia32_sbb_u32.
12865 (_subborrow_u64): Swap _X and _Y arguments in the call to
12866 __builtin_ia32_sbb_u64.
12868 2017-07-04 Jakub Jelinek <jakub@redhat.com>
12871 * tree-vrp.c (compare_assert_loc): Turn into a function template
12872 with stable template parameter. Only test if a->e is NULL,
12873 !a->e == !b->e has been verified already. Use e == NULL or
12874 e != NULL instead of e or ! e tests. If stable is true, don't use
12875 iterative_hash_expr, on the other side allow a or b or both NULL
12876 and sort the NULLs last.
12877 (process_assert_insertions): Sort using compare_assert_loc<false>
12878 instead of compare_assert_loc, later sort using
12879 compare_assert_loc<true> before calling process_assert_insertions_for
12880 in a loop. Use break instead of continue once seen NULL pointer.
12882 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
12884 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
12885 Cortex-R7 and Cortex-R8 processors.
12887 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12889 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
12890 uninitialized while src is not.
12892 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
12894 * common/config/arm/arm-common.c: Adjust include path for
12896 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
12897 (arm-cpu.h): Create in build directory. Adjust dependency rules.
12898 (arm-cpu-data.h): Likewise.
12899 (arm-cpu-cdata.h): Likewise.
12900 * config/arm/arm-cpu.h: Delete.
12901 * config/arm/arm-cpu-cdata.h: Delete.
12902 * config/arm/arm-cpu-data.h: Delete.
12904 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
12906 * config/arm/arm-cpus.in (cortex-a55): New.
12907 (cortex-a75): Likewise.
12908 (cortex-a75.cortex-a55): Likewise.
12909 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
12911 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
12912 * config/arm/arm-cpu-cdata.h: Regenerate.
12913 * config/arm/arm-cpu-data.h: Regenerate.
12914 * config/arm/arm-cpu.h: Regenerate.
12915 * config/arm/arm-tables.opt: Regenerate.
12916 * config/arm/arm-tune.md: Regenerate.
12918 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12920 * haifa-sched.c (sched_create_recovery_edges): Update profile.
12922 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
12924 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
12927 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
12929 PR tree-optimization/81292
12930 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
12931 full_string_p, also call adjust_related_strinfos if the adjustment
12932 is simple, otherwise invalidate related strinfos.
12934 2017-07-04 Martin Liska <mliska@suse.cz>
12937 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
12938 newly created variable as DECL_IGNORED_P.
12940 2017-07-04 Martin Liska <mliska@suse.cz>
12943 * ipa-inline.c (inline_small_functions):
12944 Use xstrdup_for_dump.
12946 2017-07-04 Tom de Vries <tom@codesourcery.com>
12948 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
12950 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12953 * config/darwin.c (darwin_function_switched_text_sections):
12956 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
12958 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
12960 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
12962 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
12964 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
12966 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
12967 min_profitable_iters, and th as inclusive lower bounds.
12968 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
12969 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
12970 for min_profitable_iters and min_profitable_estimate.
12971 (vect_transform_loop): Treat th as an inclusive lower bound.
12972 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
12974 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
12977 * config/darwin.c (darwin_function_switched_text_sections):
12978 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
12979 in two pieces, and suppress the use of buf.
12981 2017-07-03 Nathan Sidwell <nathan@acm.org>
12983 * hash-table.h (hash_table_mod1): Fix indentation.
12985 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
12987 PR middle-end/81290
12988 * predict.c (force_edge_cold): Be more careful about propagation
12990 * profile-count.h (profile_probability::guessed,
12991 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
12993 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
12995 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
12997 * doc/invoke.texi (rcpc architecture extension): Document it.
12999 2017-07-03 Richard Biener <rguenther@suse.de>
13001 PR tree-optimization/60510
13002 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
13003 the scalar reduction PHI and use it.
13004 (vectorizable_reduction): Properly guard the single_defuse_cycle
13005 path for non-SLP reduction chains where we cannot use it.
13006 Rework reduc_def/index and vector type deduction. Rework
13007 vector operand gathering during reduction op code-gen.
13008 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
13009 chains dissolve the chain and leave it to non-SLP reduction
13012 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13014 * tree-data-ref.h (dr_alignment): Declare.
13015 * tree-data-ref.c (dr_alignment): New function.
13016 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
13017 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
13019 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
13021 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13023 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
13024 and base_misalignment fields.
13025 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
13026 * tree-data-ref.c: Include builtins.h.
13027 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
13028 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
13029 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
13030 * tree-vect-data-refs.c: Include tree-cfg.h.
13031 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
13032 fields instead of calculating an alignment here.
13033 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
13034 innermost_loop_behavior fields.
13036 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13038 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
13040 (DR_STEP_ALIGNMENT): New macro.
13041 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
13042 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
13043 (create_data_ref): Print it.
13044 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
13045 to tell whether the step preserves vector (mis)alignment.
13046 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
13047 Move the check for an integer step and generalise to all INTEGER_CST.
13048 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
13049 Print the outer step alignment.
13051 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13053 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
13054 with offset_alignment.
13055 (DR_ALIGNED_TO): Delete.
13056 (DR_OFFSET_ALIGNMENT): New macro.
13057 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
13058 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
13059 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
13060 (create_data_ref): Likewise.
13061 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
13062 (vect_analyze_data_refs): Likewise.
13063 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
13064 creating dummy innermost behavior.
13066 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13068 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
13069 with a "innermost_loop_behavior *" and refeence tree.
13070 * tree-data-ref.c (dr_analyze_innermost): Likewise.
13071 (create_data_ref): Update call accordingly.
13072 * tree-predcom.c (find_looparound_phi): Likewise.
13074 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13076 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
13077 fields with dr_wrt_vec_loop.
13078 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
13079 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
13080 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
13081 (vect_dr_behavior): New function.
13082 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
13083 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
13084 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
13085 track whether the step preserves the misalignment.
13086 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
13087 Use vect_dr_behavior.
13088 (vect_setup_realignment): Update call accordingly.
13089 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
13090 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
13091 call to vect_create_addr_base_for_vector_ref.
13092 (vect_create_cond_for_align_checks): Likewise.
13093 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
13094 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
13095 (vect_recog_mask_conversion_pattern): Likewise.
13096 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
13097 (new_stmt_vec_info): Remove redundant zeroing.
13099 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
13101 * common/config/arm/arm-common.c (arm_be8_option): New function.
13102 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
13103 (ISA_ARMv6): Add isa_bit_be8.
13104 * config/arm/arm.h (arm_be8_option): Add prototype.
13105 (BE8_SPEC_FUNCTION): New define.
13106 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
13107 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
13108 (mlittle-endian): Similarly.
13109 (mbe8, mbe32): New options.
13110 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
13111 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
13113 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13115 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
13117 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13119 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
13120 (cleanup_tree_cfg_bb): Use it.
13121 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
13123 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
13125 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13128 * loop-doloop.c (add_test): Update profile.
13130 2017-07-03 Martin Liska <mliska@suse.cz>
13133 * sanopt.c (rewrite_usage_of_param): New function.
13134 (sanitize_rewrite_addressable_params): Likewise.
13135 (pass_sanopt::execute): Call rewrite_usage_of_param.
13137 2017-07-03 Richard Biener <rguenther@suse.de>
13139 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
13140 back to using VIEW_CONVERT_EXPR.
13142 2017-07-03 Martin Liska <mliska@suse.cz>
13145 * doc/extend.texi: Document when a resolver function is
13146 generated for target_clones.
13148 2017-07-03 Martin Liska <mliska@suse.cz>
13150 * asan.c (asan_emit_stack_protection): Unpoison just red zones
13151 and shadow memory of auto variables which are subject of
13152 use-after-scope sanitization.
13153 (asan_expand_mark_ifn): Add do set only when is_poison.
13155 2016-07-03 Richard Biener <rguenther@suse.de>
13157 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
13159 (vect_force_simple_reduction): Record reduction def -> phi mapping.
13160 (vectorizable_reduction): Perform reduction PHI creation when
13161 visiting a reduction PHI and adjust and simplify code generation
13162 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
13163 (vect_transform_loop): Visit reduction PHIs.
13164 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
13165 defs into the SLP tree.
13166 (vect_build_slp_tree): Reduction defs terminate the recursion.
13167 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
13169 (vect_get_vec_defs_for_stmt_copy): Export.
13170 (vect_get_vec_defs): Likewise.
13171 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
13173 (vect_get_vec_defs_for_stmt_copy): Declare.
13174 (vect_get_vec_defs): Likewise.
13176 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13178 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
13179 parameter with a "loop" parameter and use it instead of the
13180 loop containing DR_STMT. Don't check simple_iv when doing
13181 BB analysis. Describe the two analysis modes in the comment.
13183 2017-07-03 Tom de Vries <tom@codesourcery.com>
13185 PR tree-optimization/69468
13186 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
13187 (find_same_succ_bb): Handle ignore_edge_flags.
13189 2017-07-03 Tom de Vries <tom@codesourcery.com>
13191 PR tree-optimization/81192
13192 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
13194 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
13196 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
13198 2017-07-03 Tom de Vries <tom@codesourcery.com>
13200 PR tree-optimization/81192
13201 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
13202 BB_SAME_SUCC (bb) == NULL.
13204 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13206 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
13209 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13211 * dumpfile.c: Include profile-count.h
13212 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
13214 (insert_cond_bb): Update profile.
13215 * tree-cfg.h (insert_cond_bb): Update prototype.
13216 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
13217 * tree-dump.c: Do not include tree-cfg.
13219 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13221 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
13223 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13225 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
13228 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13230 * tree-complex.c (expand_complex_div_wide): update profile.
13232 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13233 Alan Hayward <alan.hayward@arm.com>
13234 David Sherwood <david.sherwood@arm.com>
13236 * Makefile.in (MACHMODE_H): Remove insn-modes.h
13237 (CORETYPES_H): New define.
13238 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
13239 (insn-modes-inline.h, s-modes-inline-h): New rules.
13240 (generated_files): Add insn-modes-inline.h.
13241 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
13242 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
13243 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
13244 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
13245 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
13246 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
13247 (build/gencodes.o, build/genconditions.o): Likewise.
13248 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
13249 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
13250 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
13251 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
13252 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
13253 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
13254 * coretypes.h: Include everything up to real.h for generators.
13255 Include insn-modes.h first. Include wide-int-print.h after
13256 wide-int.h. Include insn-modes-inline.h and then machmode.h.
13257 * machmode.h: Don't include insn-modes.h here.
13258 * function-tests.c: Remove includes of signop.h, machmode.h,
13259 double-int.h and wide-int.h.
13261 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
13263 * optc-save-gen.awk: Likewise.
13264 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
13265 * godump.c: Remove include of wide-int-print.h.
13266 * pretty-print.h: Likewise.
13267 * wide-int-print.cc: Likewise.
13268 * wide-int.cc: Likewise.
13269 * hash-map-tests.c: Remove include of signop.h.
13270 * hash-set-tests.c: Likewise.
13271 * rtl-tests.c: Likewise.
13272 * mkconfig.sh: Remove include of machmode.h.
13273 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
13275 (emit_insn_modes_inline_h): ...this new function. Emit the code
13276 into an insn-modes-inline.h header file, adding appropriate
13277 include guards and end comments.
13278 (emit_insn_modes_c_header): Remove include of machmode.h.
13279 (emit_min_insn_modes_c_header): Include coretypes.h rather than
13281 (main): Handle -i flag and call emit_insn_modes_inline_h when
13284 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13286 * tree-ssa-strlen.c (strinfo): Rename the length field to
13287 nonzero_chars. Add a full_string_p field.
13288 (compare_nonzero_chars, zero_length_string_p): New functions.
13289 (get_addr_stridx): Add an offset_out parameter.
13290 Use compare_nonzero_chars.
13291 (get_stridx): Update accordingly. Use compare_nonzero_chars.
13292 (new_strinfo): Update after above changes to strinfo.
13293 (set_endptr_and_length): Set full_string_p.
13294 (get_string_length): Update after above changes to strinfo.
13295 (unshare_strinfo): Update call to new_strinfo.
13296 (maybe_invalidate): Likewise.
13297 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
13298 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
13299 as a uhwi instead of an shwi. Update after above changes to
13300 strinfo and new_strinfo.
13301 (zero_length_string): Assert that chainsi contains full strings.
13302 Use zero_length_string_p. Update call to new_strinfo.
13303 (adjust_related_strinfos): Update after above changes to strinfo.
13304 Copy full_string_p from origsi.
13305 (adjust_last_stmt): Use zero_length_string_p.
13306 (handle_builtin_strlen): Update after above changes to strinfo and
13307 new_strinfo. Install the lhs as the string length if the previous
13308 entry didn't describe a full string.
13309 (handle_builtin_strchr): Update after above changes to strinfo
13311 (handle_builtin_strcpy): Likewise.
13312 (handle_builtin_strcat): Likewise.
13313 (handle_builtin_malloc): Likewise.
13314 (handle_pointer_plus): Likewise.
13315 (handle_builtin_memcpy): Likewise. Track nonzero characters
13316 that aren't necessarily followed by a nul terminator.
13317 (handle_char_store): Likewise.
13319 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13321 PR tree-optimization/80769
13322 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
13323 for malloc and calloc. Document the new invariant that all related
13324 strinfos have delayed lengths or none do.
13325 (verify_related_strinfos): Move earlier in file.
13326 (set_endptr_and_length): New function, split out from...
13327 (get_string_length): ...here. Also set the lengths of related
13329 (zero_length_string): Assert that chainsi has known (rather than
13331 (adjust_related_strinfos): Likewise.
13333 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13335 PR tree-optimization/81136
13336 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
13337 assert that two references with the same misalignment have the same
13338 compile-time misalignment if those compile-time misalignments
13341 2017-07-01 Andi Kleen <ak@linux.intel.com>
13343 * print-tree.c (print_node): Print all attributes.
13345 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13347 * cfg.c (scale_bbs_frequencies): New function.
13348 * cfg.h (scale_bbs_frequencies): Declare it.
13349 * cfgloopanal.c (single_likely_exit): Cleanup.
13350 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
13352 (scale_loop_profile): Likewise.
13353 (loop_version): Likewise.
13354 (create_empty_loop_on_edge): Update.
13355 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
13356 scale_loop_frequencies, scale_loop_profile, loopify,
13357 loop_version): Update prototypes.
13358 * modulo-sched.c (sms_schedule): Update.
13359 * predict.c (unlikely_executed_edge_p): Also check probability.
13360 (probably_never_executed_edge_p): Fix typo.
13361 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13362 * tree-parloops.c (gen_parallel_loop): Update.
13363 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
13364 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13365 * tree-ssa-loop-split.c (split_loop): Update.
13366 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13367 * tree-vect-loop-manip.c (vect_do_peeling): Update.
13368 (vect_loop_versioning): Update.
13369 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13371 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13373 * trans-mem.c (split_bb_make_tm_edge): Update profile.
13375 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13377 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
13378 to keep profile consistent.
13380 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13382 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
13383 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
13384 * profile-count.h (max_safe_multiplier): Make unsigned.
13385 (profile_count::guessed_zero): New.
13387 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13389 * bb-reorder.c (fix_up_crossing_landing_pad,
13390 fix_crossing_conditional_branches): Use make_single_succ_edge
13391 to keep profile consistent.
13393 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13395 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
13398 2017-07-01 Jakub Jelinek <jakub@redhat.com>
13401 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
13402 the right scopes, make sure cond_jump isn't preserved between multiple
13403 iterations. Search for fallthru edge whenever there are 3+ edges and
13404 use find_fallthru_edge for it.
13406 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13408 Patch by Alexander Monakov <amonakov@ispras.ru>
13409 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
13410 probabilities consistently.
13412 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13414 * pa.c (pa_expand_compare_and_swap_loop): Update call of
13415 emit_cmp_and_jump_insns.
13417 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13420 * tree-inline.c (expand_call_inline): Combine profile statuses.
13422 2017-06-30 Andrew Pinski <apinski@cavium.com>
13424 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
13425 fold_stmt returned true.
13427 2017-06-30 Nathan Sidwell <nathan@acm.org>
13429 * ggc.h (empty_string): Delete.
13430 * cfgexpand.c (expand_asm_stmt): Use plain "".
13431 * optabs.c (expand_asm_memory_barrier): Likewise.
13432 * stringpool.c (empty_string): Delete.
13433 (digit_vector, digit_string): Delete.
13434 (ggc_alloc_string): Use plain "", don't optimize single digit
13435 strings. Use ggc_alloc_atomic.
13437 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
13439 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
13440 comparison set and one other set, use the cost of the non-comparison
13443 2017-06-30 Nathan Sidwell <nathan@acm.org>
13445 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
13448 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
13450 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
13451 loops. Remove now unneeded calls to gimple_switch_set_label() that
13452 just set removed labels to NULL_TREE.
13454 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
13456 * tree-ssanames.c (set_range_info_raw): Abstract from ...
13457 (set_range_info): ...here. Only call set_range_info_raw if domain
13459 (set_nonzero_bits): Call set_range_info_raw.
13460 * tree-ssanames.h (set_range_info_raw): New.
13462 2017-06-30 Jakub Jelinek <jakub@redhat.com>
13465 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
13466 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
13467 of nonimmediate_operand and <store_mask_constraint> instead of m
13468 for the input operand. For V8FI iterator, always split if input
13469 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
13470 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
13471 <store_mask_predicate> instead of register_operand and
13472 <store_mask_constraint> instead of v for the input operand. Make
13473 sure both operands aren't MEMs for if not <mask_applied>.
13475 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
13477 * lto-wrapper.c (copy_file) Close both file descriptors before
13480 2017-06-30 Martin Liska <mliska@suse.cz>
13483 * multiple_target.c (create_dispatcher_calls): Make ifunc
13484 also for function that don't have calls or are not referenced.
13486 2017-06-30 Richard Biener <rguenther@suse.de>
13488 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
13489 analyze the first scalar stmt. Move vector type computation
13490 for the BB case here from ...
13491 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
13492 live operation processing in the SLP case properly.
13494 2017-06-30 Richard Biener <rguenther@suse.de>
13496 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
13498 2017-06-30 Martin Liska <mliska@suse.cz>
13501 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
13502 before BUILT_IN_UNWIND_RESUME when ASAN is used.
13504 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
13506 * doc/invoke.texi (AArch64): Add missing options and remove redundant
13509 2017-06-30 Richard Biener <rguenther@suse.de>
13511 PR tree-optimization/81249
13512 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
13513 condition reduction result to original scalar type.
13515 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13517 * profile-count.h (enum profile_quality): Fix typos and whitespace
13520 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13522 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
13523 type for branch probabilities.
13525 2017-06-29 Julian Brown <julian@codesourcery.com>
13526 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13528 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
13529 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
13530 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
13531 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
13533 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13535 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
13536 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
13537 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
13538 CC usage from generic code to here.
13539 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
13540 CC usage into the target macros.
13542 2017-06-29 Maya Rashish <coypu@sdf.org>
13544 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
13547 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13549 * arm/arm-builtins.c: Include profile-count.h
13550 * except.c (sjlj_emit_function_enter): Use
13551 profile_probability::unlikely.
13553 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13555 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
13556 and tocrel_offset be pointer args rather than implicitly using
13558 (legitimate_constant_pool_address_p, rs6000_emit_move,
13559 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
13560 tocrel_offset and use in toc_relative_expr_p call.
13561 (print_operand, print_operand_address): Use static tocrel_base_oac
13562 and tocrel_offset_oac.
13563 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
13566 2017-06-29 Maya Rashish <coypu@sdf.org>
13568 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
13570 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
13572 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
13573 objects, take into account only the alignment of 'op0' and 'mode1' if
13576 2017-06-29 Steve Ellcey <sellcey@cavium.com>
13578 * ccmp.c (ccmp_tree_comparison_p): New function.
13579 (ccmp_candidate_p): Update to use above function.
13580 (get_compare_parts): New function.
13581 (expand_ccmp_next): Update to use new functions.
13582 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
13584 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
13585 take mode as argument.
13586 * ccmp.h (expand_ccmp_expr): Add mode as argument.
13587 * expr.c (expand_expr_real_1): Pass mode as argument.
13589 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
13591 * combine.c (combine_instructions): Print insns to dump_file, together
13594 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13596 * asan.c (asan_emit_stack_protection): Update.
13597 (create_cond_insert_point): Update.
13598 * auto-profile.c (afdo_propagate_circuit): Update.
13599 * basic-block.h (struct edge_def): Turn probability to
13600 profile_probability.
13601 (EDGE_FREQUENCY): Update.
13602 * bb-reorder.c (find_traces_1_round): Update.
13603 (better_edge_p): Update.
13604 (sanitize_hot_paths): Update.
13605 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
13606 (make_single_succ_edge): Update.
13607 (check_bb_profile): Update.
13608 (dump_edge_info): Update.
13609 (update_bb_profile_for_threading): Update.
13610 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
13611 probabilitycount to 0.
13612 * cfgbuild.c (compute_outgoing_frequencies): Update.
13613 * cfgcleanup.c (try_forward_edges): Update.
13614 (outgoing_edges_match): Update.
13615 (try_crossjump_to_edge): Update.
13616 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
13617 (expand_gimple_tailcall): Update.
13618 (construct_init_block): Use make_single_succ_edge.
13619 (construct_exit_block): Use make_single_succ_edge.
13620 * cfghooks.c (verify_flow_info): Update.
13621 (redirect_edge_succ_nodup): Update.
13622 (split_edge): Update.
13623 (account_profile_record): Update.
13624 * cfgloopanal.c (single_likely_exit): Update.
13625 * cfgloopmanip.c (scale_loop_profile): Update.
13626 (set_zero_probability): Remove.
13627 (duplicate_loop_to_header_edge): Update.
13628 * cfgloopmanip.h (loop_version): Update prototype.
13629 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
13630 (force_nonfallthru_and_redirect): Update.
13631 (update_br_prob_note): Update.
13632 (rtl_verify_edges): Update.
13633 (purge_dead_edges): Update.
13634 (rtl_lv_add_condition_to_bb): Update.
13635 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
13636 * cgraphunit.c (init_lowered_empty_function): Update.
13637 (cgraph_node::expand_thunk): Update.
13638 * cilk-common.c: Include profile-count.h
13639 * dojump.c (inv): Remove.
13640 (jumpifnot): Update.
13641 (jumpifnot_1): Update.
13642 (do_jump_1): Update.
13644 (do_jump_by_parts_greater_rtx): Update.
13645 (do_compare_rtx_and_jump): Update.
13646 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
13647 do_jump_1. do_compare_rtx_and_jump): Update prototype.
13648 * dwarf2cfi.c: Include profile-count.h
13649 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
13650 (sjlj_emit_dispatch_table): Likewise.
13651 * explow.c: Include profile-count.h
13652 * expmed.c (emit_store_flag_force): Update.
13653 (do_cmp_and_jump): Update.
13654 * expr.c (compare_by_pieces_d::generate): Update.
13655 (compare_by_pieces_d::finish_mode): Update.
13656 (emit_block_move_via_loop): Update.
13657 (store_expr_with_bounds): Update.
13658 (store_constructor): Update.
13659 (expand_expr_real_2): Update.
13660 (expand_expr_real_1): Update.
13661 * expr.h (try_casesi, try_tablejump): Update prototypes.
13662 * gimple-pretty-print.c (dump_probability): Update.
13663 (dump_profile): New.
13664 (dump_gimple_label): Update.
13665 (dump_gimple_bb_header): Update.
13666 * graph.c (draw_cfg_node_succ_edges): Update.
13667 * hsa-gen.c (convert_switch_statements): Update.
13668 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
13669 (find_if_case_1): Update.
13670 (find_if_case_2): Update.
13671 * internal-fn.c (expand_arith_overflow_result_store): Update.
13672 (expand_addsub_overflow): Update.
13673 (expand_neg_overflow): Update.
13674 (expand_mul_overflow): Update.
13675 (expand_vector_ubsan_overflow): Update.
13676 * ipa-cp.c (good_cloning_opportunity_p): Update.
13677 * ipa-split.c (split_function): Use make_single_succ_edge.
13678 * ipa-utils.c (ipa_merge_profiles): Update.
13679 * loop-doloop.c (add_test): Update.
13680 (doloop_modify): Update.
13681 * loop-unroll.c (compare_and_jump_seq): Update.
13682 (unroll_loop_runtime_iterations): Update.
13683 * lra-constraints.c (lra_inheritance): Update.
13684 * lto-streamer-in.c (input_cfg): Update.
13685 * lto-streamer-out.c (output_cfg): Update.
13686 * mcf.c (adjust_cfg_counts): Update.
13687 * modulo-sched.c (sms_schedule): Update.
13688 * omp-expand.c (expand_omp_for_init_counts): Update.
13689 (extract_omp_for_update_vars): Update.
13690 (expand_omp_ordered_sink): Update.
13691 (expand_omp_for_ordered_loops): Update.
13692 (expand_omp_for_generic): Update.
13693 (expand_omp_for_static_nochunk): Update.
13694 (expand_omp_for_static_chunk): Update.
13695 (expand_cilk_for): Update.
13696 (expand_omp_simd): Update.
13697 (expand_omp_taskloop_for_outer): Update.
13698 (expand_omp_taskloop_for_inner): Update.
13699 * omp-simd-clone.c (simd_clone_adjust): Update.
13700 * optabs.c (expand_doubleword_shift): Update.
13701 (expand_abs): Update.
13702 (emit_cmp_and_jump_insn_1): Update.
13703 (expand_compare_and_swap_loop): Update.
13704 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
13705 * predict.c (predictable_edge_p): Update.
13706 (edge_probability_reliable_p): Update.
13707 (set_even_probabilities): Update.
13708 (combine_predictions_for_insn): Update.
13709 (combine_predictions_for_bb): Update.
13710 (propagate_freq): Update.
13711 (estimate_bb_frequencies): Update.
13712 (force_edge_cold): Update.
13713 * profile-count.c (profile_count::dump): Add missing space into dump.
13714 (profile_count::debug): Add newline.
13715 (profile_count::differs_from_p): Explicitly convert to unsigned.
13716 (profile_count::stream_in): Update.
13717 (profile_probability::dump): New member function.
13718 (profile_probability::debug): New member function.
13719 (profile_probability::differs_from_p): New member function.
13720 (profile_probability::differs_lot_from_p): New member function.
13721 (profile_probability::stream_in): New member function.
13722 (profile_probability::stream_out): New member function.
13723 * profile-count.h (profile_count_quality): Rename to ...
13724 (profile_quality): ... this one.
13725 (profile_probability): New.
13726 (profile_count): Update.
13727 * profile.c (compute_branch_probabilities): Update.
13728 * recog.c (peep2_attempt): Update.
13729 * sched-ebb.c (schedule_ebbs): Update.
13730 * sched-rgn.c (find_single_block_region): Update.
13731 (compute_dom_prob_ps): Update.
13732 (schedule_region): Update.
13733 * sel-sched-ir.c (compute_succs_info): Update.
13734 * stmt.c (struct case_node): Update.
13735 (do_jump_if_equal): Update.
13736 (get_outgoing_edge_probs): Update.
13737 (conditional_probability): Update.
13738 (emit_case_dispatch_table): Update.
13739 (expand_case): Update.
13740 (expand_sjlj_dispatch_table): Update.
13741 (emit_case_nodes): Update.
13742 * targhooks.c: Update.
13743 * tracer.c (better_p): Update.
13744 (find_best_successor): Update.
13745 * trans-mem.c (expand_transaction): Update.
13746 * tree-call-cdce.c: Update.
13747 * tree-cfg.c (gimple_split_edge): Upate.
13748 (move_sese_region_to_fn): Upate.
13749 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
13750 * tree-eh.c (lower_resx): Upate.
13751 (cleanup_empty_eh_move_lp): Upate.
13752 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13753 * tree-inline.c (copy_edges_for_bb): Update.
13754 (copy_cfg_body): Update.
13755 * tree-parloops.c (gen_parallel_loop): Update.
13756 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
13757 (gimple_gen_time_profiler): Update.
13758 * tree-ssa-dce.c (remove_dead_stmt): Update.
13759 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
13760 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
13761 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
13762 (unloop_loops): Update.
13763 (try_peel_loop): Update.
13764 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13765 * tree-ssa-loop-split.c (connect_loops): Update.
13766 (split_loop): Update.
13767 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13768 (hoist_guard): Update.
13769 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
13770 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
13771 (value_replacement): Update.
13772 * tree-ssa-reassoc.c (branch_fixup): Update.
13773 * tree-ssa-tail-merge.c (replace_block_by): Update.
13774 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
13775 (create_edge_and_update_destination_phis): Update.
13776 (compute_path_counts): Update.
13777 (recompute_probabilities): Update.
13778 (update_joiner_offpath_counts): Update.
13779 (freqs_to_counts_path): Update.
13780 (duplicate_thread_path): Update.
13781 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
13782 (struct switch_conv_info): Update.
13783 (gen_inbound_check): Update.
13784 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
13785 (vect_do_peeling): Update.
13786 (vect_loop_versioning): Update.
13787 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13788 (optimize_mask_stores): Update.
13789 * ubsan.c (ubsan_expand_null_ifn): Update.
13790 * value-prof.c (gimple_divmod_fixed_value): Update.
13791 (gimple_divmod_fixed_value_transform): Update.
13792 (gimple_mod_pow2): Update.
13793 (gimple_mod_pow2_value_transform): Update.
13794 (gimple_mod_subtract): Update.
13795 (gimple_mod_subtract_transform): Update.
13796 (gimple_ic): Update.
13797 (gimple_stringop_fixed_value): Update.
13798 (gimple_stringops_transform): Update.
13799 * value-prof.h: Update.
13801 2017-06-29 Carl Love <cel@us.ibm.com>
13803 * config/rs6000/rs6000-c.c: Add support for built-in functions
13804 vector signed int vec_signed (vector float);
13805 vector signed long long vec_signed (vector double);
13806 vector signed int vec_signed2 (vector double, vector double);
13807 vector signed int vec_signede (vector double);
13808 vector signed int vec_signedo (vector double);
13809 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
13810 instruction generator.
13811 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13812 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
13813 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
13815 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
13816 vunsignede_v2df): Add define_expands.
13817 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
13818 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
13819 VEC_UNSIGNEDO): Add definitions.
13820 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
13821 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
13822 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
13823 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
13824 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
13825 * config/rs6000/altivec.h (vec_signed, vec_signed2,
13826 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
13827 vec_unsignede, vec_unsignedo): Add builtin defines.
13828 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
13830 * doc/extend.texi: Update the built-in documentation file for the
13831 new built-in functions.
13833 2017-06-29 Richard Biener <rguenther@suse.de>
13835 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
13836 reduction chains to LOOP_VINFO_REDUCTIONS.
13837 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
13838 SLP reductions after processing reduction chains.
13840 2017-06-29 Nathan Sidwell <nathan@acm.org>
13842 * builtins.c (fold_builtin_FUNCTION): Use
13843 lang_hooks.decl_printable_name.
13845 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
13847 PR middle-end/81194
13848 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
13849 with only one label.
13850 * stmt.c (expand_case): Assert NCASES is greater than one.
13852 2017-06-29 Richard Biener <rguenther@suse.de>
13854 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
13856 (group_case_labels): Likewise.
13857 (find_taken_edge): Push sanity checking on val to workers...
13858 (find_taken_edge_cond_expr): ... here
13859 (find_taken_edge_switch_expr): ... and here, handle cases
13860 with just a default label.
13861 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
13862 (group_case_labels): Likewise.
13863 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
13864 group_case_labels does anything cleanup the CFG again.
13866 2017-06-29 Bin Cheng <bin.cheng@arm.com>
13868 PR tree-optimization/81196
13869 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
13870 exit condition comparing two IVs.
13872 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
13874 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
13875 profile to the dummy entry at the end of the list of architectures.
13876 * config/arm/arm-cpu-cdata.h: Regenerated.
13878 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13879 Michael Collison <michael.collison@arm.com>
13882 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
13884 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
13885 (*aarch64_reg_<mode>3_minus_mask): New pattern.
13886 (*aarch64_<optab>_reg_di3_mask2): New pattern.
13887 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
13888 of shift when the shift amount is masked with constant equal to
13889 the size of the mode.
13890 * config/aarch64/predicates.md (subreg_lowpart_operator): New
13893 2017-06-29 Martin Liska <mliska@suse.cz>
13895 * config/i386/i386.opt: Change range from [1,5] to [0,5].
13897 2017-06-29 Yury Gribov <tetra2005@gmail.com>
13900 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
13903 (edge_growth_cache_entry::edge_growth_cache_entry): New
13905 (reset_edge_growth_cache): Update to use constructor.
13907 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13909 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
13910 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13911 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
13913 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
13915 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
13916 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
13918 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
13920 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
13921 (*-linux-uclibc*): Add t-uclibc tmake_file.
13922 * config/t-musl: New.
13923 * config/t-uclibc: New.
13925 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
13927 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
13929 (gen_comm_data): Emit architectural setting of arch_prof.
13930 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
13932 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
13933 (armv8-m.base, armv8-m.main): Likewise.
13934 * arm-protos.h (arm_build_target): Add profile field.
13935 (arch_option): Likewise.
13936 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
13938 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
13939 arm_active_target.profile.
13941 2017-06-28 Richard Biener <rguenther@suse.de>
13943 PR middle-end/81227
13944 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
13945 TYPE_OVERFLOW_WRAPS.
13946 * match.pd (negate_expr_p): Likewise.
13947 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
13948 fold_build2, not fold_binary.
13950 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13952 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
13953 Convert memory address to Pmode.
13954 (aarch64_print_operand): Assert MEM operands are always Pmode.
13956 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
13959 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
13960 Remove redundant if.
13961 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
13962 * config/arm/aarch-common-protos.h
13963 (aarch_forward_to_shift_is_not_shifted_re): Remove.
13964 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
13966 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
13969 * multiple_target.c (create_dispatcher_calls): Set the default
13970 clone to be static, not public.
13972 2017-06-28 Richard Biener <rguenther@suse.de>
13974 * tree-vect-loop.c (vectorizable_reduction): Move special
13975 cond reduction IV var creation ...
13976 (vect_create_epilog_for_reduction): ... here. Remove induction_index
13977 parameter. Use STMT_VINFO_VECTYPE.
13978 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
13981 2017-06-28 Martin Liska <mliska@suse.cz>
13984 * ipa-visibility.c (non_local_p): Handle visibility.
13986 2017-06-28 Martin Liska <mliska@suse.cz>
13989 * common.opt: Add IntegerRange to various options.
13990 * opt-functions.awk (integer_range_info): New function.
13991 * optc-gen.awk: Add integer_range_info to cl_options struct.
13992 * opts-common.c (decode_cmdline_option): Handle
13993 CL_ERR_INT_RANGE_ARG.
13994 (cmdline_handle_error): Likewise.
13995 * opts.c (print_filtered_help): Show valid interval in
13996 when --help is provided.
13997 * opts.h (struct cl_option): Add range_min and range_max fields.
13998 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
14000 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
14002 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
14003 (x * C EQ/NE y * C): New transformation.
14005 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
14007 * genmultilib (combination_space): Accept '+' in option names.
14009 2017-06-28 Martin Liska <mliska@suse.cz>
14012 * asan.c (instrument_derefs): Bail out inner references
14013 that are hard register variables.
14015 2017-06-28 Jakub Jelinek <jakub@redhat.com>
14018 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
14019 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
14021 2017-06-28 Richard Biener <rguenther@suse.de>
14023 * tree-vectorizer.h (vect_get_vec_defs): Remove.
14024 (vect_get_slp_defs): Adjust.
14025 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
14027 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
14029 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
14030 get_initial_defs_for_reduction instead of vect_get_vec_defs.
14031 (vectorizable_reduction): Adjust.
14032 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
14034 (vect_get_slp_defs): Likewise.
14035 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
14036 (vectorizable_bswap): Adjust.
14037 (vectorizable_call): Likewise.
14038 (vectorizable_conversion): Likewise.
14039 (vectorizable_assignment): Likewise.
14040 (vectorizable_shift): Likewise.
14041 (vectorizable_operation): Likewise.
14042 (vectorizable_store): Likewise.
14043 (vectorizable_condition): Likewise.
14044 (vectorizable_comparison): Likewise.
14046 2017-06-28 Michael Collison <michael.collison@arm.com>
14049 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
14051 (arm_gen_movmemqi): Removed unused variable 'i'.
14052 Convert 'for' loop into 'while' loop.
14053 (arm_expand_prologue): Remove last unnecessary set of insn.
14054 (thumb_pop): Remove unused variable 'pushed_words'.
14055 (thumb_exit): Remove last unnecessary set of regs_to_pop.
14057 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14059 * config/s390/predicates.md: Use s390_rel_address_ok_p.
14060 * config/s390/s390-protos.h: Add prototype of
14061 s390_rel_address_ok_p.
14062 * config/s390/s390.c (s390_got_symbol): New function.
14063 (s390_rel_address_ok_p): New function.
14064 (legitimize_pic_address): Use s390_rel_address_ok_p.
14065 (s390_load_got): Use s390_got_symbol.
14066 (s390_option_override): Issue error if
14067 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
14068 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
14070 * config/s390/s390.opt: New option mpic-data-is-text-relative.
14072 2017-06-27 Andrew Pinski <apinski@cavium.com>
14074 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
14075 (X * copysign (1.0, X)): New pattern.
14076 (X * copysign (1.0, -X)): New pattern.
14077 (copysign (-1.0, CST)): New pattern.
14079 2017-06-27 Joseph Myers <joseph@codesourcery.com>
14081 * genmultilib (combination_space): Remove variable.
14082 Validate reuse rules against regular expression for any sequence
14083 of multilib options in any order.
14085 2017-06-27 Michael Collison <michael.collison@arm.com>
14087 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
14088 call aarch64_split_simd_combine.
14089 * (aarch64_combine_internal<mode>): Delete pattern.
14090 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
14091 Allow register and subreg operands.
14093 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14095 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
14096 specific need, just fallback on defaults.
14097 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
14099 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14100 Olivier Hainque <hainque@adacore.com>
14102 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
14104 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
14105 targets. Pick a default if no particular attempt applied.
14106 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
14109 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14111 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
14112 (x86_64-wrs-vxworks7): Likewise.
14114 2017-06-27 Marek Polacek <polacek@redhat.com>
14117 * ubsan.c (instrument_null): Check get_base_address's result for null.
14119 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
14121 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
14123 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
14125 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
14126 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
14127 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
14128 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
14129 New function types.
14130 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
14131 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
14132 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
14133 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
14134 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
14135 BUILT_IN_FEUPDATEENV): New builtins.
14136 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
14137 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
14138 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
14139 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
14141 (builtin_structptr_types): Adjust size.
14142 * tree.c (builtin_structptr_types): Add four entries.
14144 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14145 Olivier Hainque <hainque@adacore.com>
14147 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
14148 (TLS_SYM): New local macro, forcing reference to __tls__ on
14149 link command lines for VxWorks 7 RTPs, triggering initialization
14151 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
14152 OS features TLS support, true for RTPs on VxWorks 7.
14153 * config/vxworks.c (vxworks_override_options): Setup emutls
14156 2017-06-27 Jakub Jelinek <jakub@redhat.com>
14158 * predict.c (test_prediction_value_range): Use -1U instead of -1
14159 to avoid narrowing conversion warning.
14160 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
14161 to avoid narrowing conversion warning.
14162 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
14164 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
14166 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14168 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
14169 64bit configurations.
14170 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
14171 (SIZE_TYPE): Likewise.
14172 * config/vxworks.c (vxworks_emutls_var_fields): Use
14173 long_unsigned_type_node instead of unsigned_type_node as the offset
14174 field type, which is "pointer" mode in emutls.c.
14176 2017-06-27 Jakub Jelinek <jakub@redhat.com>
14179 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
14181 PR middle-end/81207
14182 * gimple-fold.c (replace_call_with_call_and_fold): Handle
14183 gimple_vuse copying separately from gimple_vdef copying.
14185 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14187 * value-prof.c (free_hist): Remove call to memset and the enclosing if
14190 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14191 Olivier Hainque <hainque@adacore.com>
14193 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
14194 for all vxworks7 targets.
14195 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
14196 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
14197 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
14198 variations for VX6/VX7 and 32/64bits later on in ...
14199 (VXWORKS_LIB_SPEC): Leverage new macros.
14200 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
14201 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
14203 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14205 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
14206 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
14208 2017-06-26 Carl Love <cel@us.ibm.com>
14210 * config/rs6000/rs6000-c.c: Add support for built-in functions
14211 vector bool char vec_reve (vector bool char);
14212 vector signed char vec_reve (vector signed char);
14213 vector unsigned char vec_reve (vector unsigned char);
14214 vector bool int vec_reve (vector bool int);
14215 vector signed int vec_reve (vector signed int);
14216 vector unsigned int vec_reve (vector unsigned int);
14217 vector bool long long vec_reve (vector bool long long);
14218 vector signed long long vec_reve (vector signed long long);
14219 vector unsigned long long vec_reve (vector unsigned long long);
14220 vector bool short vec_reve (vector bool short);
14221 vector signed short vec_reve (vector signed short);
14222 vector double vec_reve (vector double);
14223 vector float vec_reve (vector float);
14224 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
14225 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
14226 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
14227 (altivec_vreve): New pattern.
14228 * config/rs6000/altivec.h (vec_reve): New define.
14229 * doc/extend.texi (vec_rev): Update the built-in documentation file
14230 for the new built-in functions.
14232 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14234 PR tree-optimization/71815
14235 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
14237 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
14239 (slsr_process_phi): Likewise.
14240 (replace_uncond_cands_and_profitable_phis): Don't replace a
14241 multiply candidate with a stride of 1 (copy or cast).
14242 (phi_incr_cost): Call uses_consumed_by_stmt rather than
14244 (lowest_cost_path): Likewise.
14245 (total_savings): Likewise.
14247 2017-06-26 Richard Biener <rguenther@suse.de>
14250 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
14251 Use def_builtin_pure for all gather builtins.
14253 2017-06-26 Richard Biener <rguenther@suse.de>
14255 PR tree-optimization/81203
14256 * tree-tailcall.c (find_tail_calls): Do not move stmts into
14257 non-dominating BBs.
14259 2017-06-26 Marek Polacek <polacek@redhat.com>
14262 * doc/invoke.texi: Document -Wmultistatement-macros.
14264 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
14266 * doc/sourcebuild.texi (ARM-specific attributes): Document new
14267 arm_neon_ok_no_float_abi effective target.
14269 2017-06-26 Richard Biener <rguenther@suse.de>
14271 PR tree-optimization/80928
14272 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
14273 (copy_bbs): Set BB_DUPLICATED flag early.
14274 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
14276 (execute_on_shrinking_pred): Likewise.
14277 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
14278 BB_DUPLICATED blocks.
14279 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
14280 iterate over all PHIs considering removal of *gsi.
14282 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
14284 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
14287 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14289 * config/rs6000/rs6000-string.c: (expand_block_clear,
14290 do_load_for_compare, select_block_compare_mode,
14291 compute_current_alignment, expand_block_compare,
14292 expand_strncmp_align_check, expand_strn_compare,
14293 expand_block_move, rs6000_output_load_multiple)
14294 Move functions related to string/block move/compare
14295 to a separate file.
14296 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
14297 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
14298 for this function which is now used in two files.
14299 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
14300 * config.gcc: Add rs6000-string.o to extra_objs for
14301 targets powerpc*-*-* and rs6000*-*-*.
14303 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
14306 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
14307 32-bit, since indexed is not valid for DImode.
14308 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
14309 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
14310 (define_peephole2 for Altivec d-form load): Add 32-bit support.
14311 (define_peephole2 for Altivec d-form store): Likewise.
14314 * multiple_target.c (create_dispatcher_calls): Only create the
14315 dispatcher call if the function is the default clone of a
14316 versioned function.
14318 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
14320 PR middle-end/80902
14321 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
14322 a call, force the call to not be a tail call.
14324 2017-06-23 Jeff Law <law@redhat.com>
14326 * doc/contrib.texi: Add entry for Steven Pemberton's work on
14329 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
14331 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
14332 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
14333 handling for early expansion of vector shifts (sl,sr,sra,rl).
14334 (builtin_function_type): Add vector shift right instructions
14335 to the unsigned argument list.
14337 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
14339 rtl-optimizatoin/79286
14340 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
14341 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
14342 trap. PIC register plus a const unspec without offset can never trap.
14344 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
14346 * tree.h (builtin_structptr_type): New type.
14347 (builtin_structptr_types): Declare new array.
14348 * tree.c (builtin_structptr_types): New array.
14349 (free_lang_data, build_common_tree_nodes): Use it.
14351 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
14354 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
14357 2017-06-22 Matt Turner <mattst88@gmail.com>
14359 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
14360 Lake models to skylake case. Assume skylake for unknown
14361 models with clflushopt.
14363 2017-06-22 Jeff Law <law@redhat.com>
14365 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
14366 frame sizes that do not satisfy aarch64_uimm12_shift.
14368 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
14370 * profile-count.h (apply_probability,
14371 apply_scale, probability_in): Fix checks for zero.
14373 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14375 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
14376 * doc/cppdiropts.texi (-I @var{dir}): Document it.
14378 2016-06-22 Richard Biener <rguenther@suse.de>
14380 * tree-vect-loop.c (vect_model_reduction_cost): Handle
14381 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
14382 REDUC_MAX_EXPR support.
14383 (vectorizable_reduction): Likewise.
14384 (vect_create_epilog_for_reduction): Likewise.
14386 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
14388 * match.pd (A / (1 << B) -> A >> B): New.
14389 * generic-match-head.c: Include optabs-tree.h.
14390 * gimple-match-head.c: Likewise.
14391 * optabs-tree.h (target_supports_op_p): New.
14392 * optabs-tree.c (target_supports_op_p): New.
14394 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14396 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
14397 $gcc_cv_ld --help output.
14398 (gcc_cv_ld_demangle): Likewise.
14399 (gcc_cv_ld_eh_frame_hdr): Likewise.
14400 (gcc_cv_ld_pie): Likewise.
14401 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
14402 (gcc_cv_ld_buildid): Likewise.
14403 (gcc_cv_ld_sysroot): Likewise.
14404 (ld_bndplt_support): Likewise.
14405 (ld_pushpopstate_support): Likewise.
14406 * configure: Regenerate.
14407 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
14409 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14412 * config/i386/sse.md (round<mode>2): Renumber match_dup and
14413 operands indexes to avoid gap between operands and match_dups.
14415 2017-06-21 Andrew Pinski <apinski@cavium.com>
14417 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
14418 Increment Arith_shift and Arith_shift_reg by 1.
14419 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
14421 * config/aarch64/aarch64.c (thunderx_tunings): Enable
14422 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
14423 (aarch64_strip_extend): Add new argument and test for it.
14424 (aarch64_cheap_mult_shift_p): New function.
14425 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
14426 add a cost if it is true.
14427 Update calls to aarch64_strip_extend.
14428 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
14430 2017-06-21 Andrew Pinski <apinski@cavium.com>
14432 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
14434 (thunderxt88): Likewise.
14435 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
14436 (thunderx_prefetch_tune): New variable.
14437 (thunderx2t99_prefetch_tune): Update for the correct values.
14438 (thunderxt88_tunings): New variable.
14439 (thunderx_tunings): Use thunderx_prefetch_tune instead of
14440 generic_prefetch_tune.
14441 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
14443 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14445 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
14446 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
14447 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
14448 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
14449 (aarch64_atomic_cas<mode>, GPI): Likewise.
14451 2017-06-21 Martin Liska <mliska@suse.cz>
14453 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
14454 statements on cold and hot labels.
14455 * predict.c (tree_estimate_probability_bb): Remove the
14456 prediction from this place.
14458 2017-06-21 Martin Liska <mliska@suse.cz>
14460 PR tree-optimization/79489
14461 * gimplify.c (maybe_add_early_return_predict_stmt): New
14463 (gimplify_return_expr): Call the function.
14464 * predict.c (tree_estimate_probability_bb): Remove handling
14466 * predict.def: Update comment about early return predictor.
14467 * gimple-predict.h (is_gimple_predict): New function.
14468 * predict.def: Change default value of early return to 66.
14469 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
14471 * passes.def: Put pass_strip_predict_hints to the beginning of
14474 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
14476 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
14477 FUNCTION_DECL declarations.
14478 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
14480 (dwaf2out_decl): Likewise.
14481 * godump.c (go_early_global_decl): Skip call to the real debug hook
14482 for FUNCTION_DECL declarations.
14483 * passes.c (rest_of_decl_compilation): Skip call to the
14484 early_global_decl debug hook for FUNCTION_DECL declarations, unless
14485 -fdump-go-spec is passed.
14487 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14489 * config/i386/i386.c (struct builtin_isa): New field pure_p.
14490 Reorder for compactness.
14491 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
14492 (def_builtin_pure, def_builtin_pure2): New functions.
14493 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
14495 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14497 * match.pd (nop_convert): New predicate.
14498 ((A +- CST1) +- CST2): Allow some NOP conversions.
14500 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14503 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
14504 with ctors/dtors if GOVD_SHARED is set.
14506 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14508 * config/aarch64/aarch64.md (movti_aarch64):
14509 Emit mov rather than orr.
14510 (movtf_aarch64): Likewise.
14511 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
14512 Emit mov rather than orr.
14514 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14516 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
14517 Swap alternatives, make integer dup more expensive.
14519 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14521 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
14522 Return true for non-tls symbols.
14524 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
14526 * config/aarch64/aarch64-cores.def (cortex-a55): New.
14527 (cortex-a75): Likewise.
14528 (cortex-a75.cortex-a55): Likewise.
14529 * config/aarch64/aarch64-tune.md: Regenerate.
14530 * doc/invoke.texi (-mtune): Document new values for -mtune.
14532 2017-06-21 Tom de Vries <tom@codesourcery.com>
14534 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
14535 stack_size feature.
14536 (Effective-Target Keywords, Other attributes): Suggest using
14537 dg-add-options stack_size feature to get stack limit in stack_size
14538 effective target documentation.
14540 2017-06-21 Julian Brown <julian@codesourcery.com>
14541 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14543 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
14544 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
14545 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
14547 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
14548 attribute type list for neon_multiply.
14549 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
14550 attribute type list for neon_multiply.
14551 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
14552 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
14553 attribute type list for neon_multiply.
14554 * config/arm/types.md (crypto_pmull): Add.
14555 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
14556 attribute type list.
14558 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
14560 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
14563 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14565 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
14566 to make sure not to dereference a NULL cost_classes_ptr pointer.
14568 2017-06-20 Carl Love <cel@us.ibm.com>
14570 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14571 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14572 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14573 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14574 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14575 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14576 VMULOSW): New enum "unspec" values.
14577 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14578 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
14579 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14580 altivec_vmulosw): New patterns.
14581 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14582 VMULOSW): Add definitions.
14584 2017-06-20 Julia Koval <julia.koval@intel.com>
14586 * config/i386/i386.c: Fix rounding expand for new pattern.
14587 * config/i386/subst.md: Fix pattern (parallel -> unspec).
14589 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14591 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
14592 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
14594 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14596 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
14599 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14601 * config/aarch64/aarch64-cores.def: Rearrange to sort by
14602 architecture, then by implementer ID.
14603 * config/aarch64/aarch64-tune.md: Regenerate.
14605 2017-06-20 Richard Biener <rguenther@suse.de>
14607 PR middle-end/81097
14608 * fold-const.c (split_tree): Fold to type before negating.
14610 2017-06-20 David Malcolm <dmalcolm@redhat.com>
14612 * diagnostic-show-locus.c
14613 (selftest::test_fixit_deletion_affecting_newline): New function.
14614 (selftest::diagnostic_show_locus_c_tests): Call it.
14616 2017-06-20 Andreas Schwab <schwab@suse.de>
14619 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
14622 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14624 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
14625 __ARM_FEATURE_COPROC according to support.
14627 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14629 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14630 Rewritten to avoid overflow for > 32-bit pointers.
14633 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
14634 by removing enum keyword.
14635 (ubsan_type_descriptor): Likewise. Formatting fix.
14638 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
14639 splitter): Require TARGET_SSE2 in the condition.
14641 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
14644 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
14645 for doing vector set of SFmode on ISA 3.0.
14646 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
14647 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
14649 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
14650 SFmode value into a V4SF variable that was extracted from another
14651 V4SF variable without converting the element to double precision
14652 and back to single precision vector format.
14653 (vsx_insert_extract_v4sf_p9_2): Likewise.
14655 2017-06-19 Jakub Jelinek <jakub@redhat.com>
14657 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
14658 in UWHI to avoid undefined overflow.
14661 * ubsan.h (enum ubsan_encode_value_phase): New.
14662 (ubsan_encode_value): Change second argument to
14663 enum ubsan_encode_value_phase with default value of
14664 UBSAN_ENCODE_VALUE_GENERIC.
14665 * ubsan.c (ubsan_encode_value): Change second argument to
14666 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
14667 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
14668 create_tmp_var_raw instead of create_tmp_var and use a
14670 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
14671 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
14672 ubsan_encode_value callers.
14675 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
14676 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
14677 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
14679 2017-06-19 Richard Biener <rguenther@suse.de>
14681 PR middle-end/81118
14682 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
14683 estimates if we changed anything.
14685 2017-06-19 Richard Biener <rguenther@suse.de>
14687 PR tree-optimization/80887
14688 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
14689 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
14690 simplified lookups, then reset mprts_hook.
14691 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
14693 (try_to_simplify): Likewise.
14695 2017-06-19 Martin Liska <mliska@suse.cz>
14698 * gimplify.c (gimplify_switch_expr):
14699 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
14701 2017-06-19 Martin Liska <mliska@suse.cz>
14703 * doc/install.texi: Document that PGO runs in 4 stages.
14705 2017-06-19 Martin Liska <mliska@suse.cz>
14708 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
14709 to dispatcher function name.
14710 * multiple_target.c (replace_function_decl): New function.
14711 (create_dispatcher_calls): Redirect both edges and references.
14713 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
14715 * profile-count.c (profile_count::dump): Dump quality.
14716 (profile_count::differs_from_p): Update for unsigned val.
14717 * profile-count.h (profile_count_quality): New enum.
14718 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
14720 2017-06-19 Richard Biener <rguenther@suse.de>
14722 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
14723 struct function as arg.
14724 (estimate_numbers_of_iterations): Export overload with loop arg.
14725 (free_numbers_of_iterations_estimates_loop): Use an overload of
14726 free_numbers_of_iterations_estimates instead.
14727 * tree-cfg.c (remove_bb): Adjust.
14728 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
14729 * tree-parloops.c (gen_parallel_loop): Likewise.
14730 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
14732 (tree_unroll_loops_completely): Likewise.
14733 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
14734 Use an overload instead and export.
14735 (estimated_loop_iterations): Adjust.
14736 (max_loop_iterations): Likewise.
14737 (likely_max_loop_iterations): Likewise.
14738 (estimate_numbers_of_iterations): Take struct function as arg
14740 (loop_exits_before_overflow): Adjust.
14741 (free_numbers_of_iterations_estimates_loop): Use an overload.
14742 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
14743 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
14745 2017-06-19 Richard Biener <rguenther@suse.de>
14748 * ipa-prop.c (find_constructor_constant_at_offset): Handle
14749 RANGE_EXPR conservatively.
14751 2017-06-16 Carl Love <cel@us.ibm.com>
14753 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14754 definitions for vec_float, vec_float2, vec_floato,
14755 vec_floate built-ins.
14756 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
14757 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
14759 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
14760 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
14761 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
14762 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
14763 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
14764 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
14765 vec_floato): Add builtin defines.
14766 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
14767 Update the built-in documentation file for the new built-in
14770 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14772 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
14773 (mthumb): Mark as the negative of -marm.
14775 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14777 * doc/invoke.texi (ARM Options, -mcpu): Document supported
14779 (ARM Options, -mtune): Document that this accepts the same
14780 extension options as -mcpu.
14781 (ARM Options, -mfpu): Document addition of -mfpu=auto.
14783 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14785 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
14786 permitted extensions.
14788 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14790 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
14791 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
14792 (armv8-m.main): Add option +nodsp.
14793 * config/arm/arm-cpu-cdata.h: Regenerated.
14795 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14797 * config/arm/t-fuchsia: New file.
14798 * config.gcc (arm*-*-fuchsia*): Use it.
14800 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14802 * config/arm/t-symbian: Rewrite for new option infrastructure.
14804 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14806 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
14807 (MULTILIB_REQUIRED): Likewise.
14809 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14811 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
14812 (MULTILIB_RESUE): Likewise.
14813 (MULTILIB_MATCHES): Likewise.
14814 (MULTLIB_REQUIRED): Likewise.
14816 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14818 * config/arm/t-rtems: Rewrite for new option framework.
14820 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14822 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
14823 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
14824 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
14825 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
14826 * config/arm/t-multilib: ... here.
14827 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
14828 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
14829 armv7-a and armv8*-a when A-profile libraries have not been built.
14830 * config/arm/t-rmprofile: Rewrite.
14832 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14834 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
14835 with a backslash. Remove the backslash after substituting unescaped
14837 * doc/fragments.texi (MULTILIB_REUSE): Document it.
14839 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14841 * config.gcc: (arm*-*-*): When building a-profile libraries, force
14842 the driver to pass through the default setting of -mfloat-abi.
14843 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
14845 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
14846 (all_feat_combs): New rule.
14847 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
14849 * config/arm/t-aprofile: Rewrite.
14851 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14853 * config/arm/arm.h (FPUTYPE_AUTO): Define.
14854 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
14855 fpu is not specified by the user/command-line.
14856 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
14857 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
14858 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
14859 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
14860 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
14861 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
14863 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14865 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
14866 * config/arm/t-arm-elf: Rewritten.
14868 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14870 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
14871 have some floating-point instructions.
14872 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
14873 (TARGET_MAYBE_HARD_FLOAT): New macro.
14874 * config/arm/arm-builtins.c (arm_init_builtins): Use
14875 TARGET_MAYBE_HARD_FLOAT.
14876 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
14878 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14880 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
14881 (configargs.h): Include it.
14882 (arm_print_hint_for_fpu_option): New function.
14883 (arm_parse_fpu_option): New function.
14884 (candidate_extension): New class.
14885 (arm_canon_for_multilib): New function.
14886 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
14887 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
14888 (ARCH_CANONICAL_SPECS): New macro.
14889 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
14891 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14893 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
14894 are set after handling multilib fragments. Set target_cpu_default2
14897 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14899 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
14901 (arm*-*-*): Set target_cpu_default2 to a quoted string.
14902 * config/arm/parsecpu.awk (check_cpu): Validate any extension
14904 (check_arch): Likewise.
14905 * config/arm/arm.c (arm_configure_build_target): Handle
14906 TARGET_CPU_DEFAULT being a string constant. Scan any feature
14907 options in the default.
14909 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14911 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
14912 when an option is an alias of another.
14913 * config/arm/parsecpu.awk (optalias): New parser token.
14914 (gen_comm_data): Mark non-alias options as such. Emit entries
14915 for extension aliases.
14916 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
14917 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
14918 (armv6kz, armv6zk, armv6t2): Likewise.
14919 (armv7): Make vfpv3-d16 an alias.
14920 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
14922 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
14923 Sort in canonical order.
14924 (armv8-a): Sort in canonical order.
14925 (armv8.1-a, armv8.2-a): Likewise.
14926 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
14928 (cortex-a9): Sort in canonical order.
14929 * config/arm/arm.c (selftests.h): Include it.
14930 (arm_test_cpu_arch_data): New function.
14931 (arm_run_self_tests): New function.
14932 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
14933 (targetm): Move declaration to the end of the file.
14934 * arm-cpu-cdata.h: Regenerated.
14936 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14938 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
14939 call to target_mode_check describing the type of option passed.
14940 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
14941 (arm_target_thumb_only): Use arm_parse_arch_option_name or
14942 arm_parse_cpu_option_name to match parameters against list of
14944 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
14945 arm_arch_core_flags data structure.
14946 * config/arm/arm-cpu_cdata.h: Regenerated.
14948 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14950 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
14952 (arm_print_hint_for_cpu_option): Likewise.
14953 (arm_print_hint_for_arch_option): Likewise.
14954 (arm_parse_cpu_option_name): Likewise.
14955 (arm_parse_arch_option_name): Likewise.
14956 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
14957 of entries in the all_fpus list.
14958 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
14959 (arm_parse_cpu_option_name): Declare.
14960 (arm_parse_arch_option_name): Declare.
14961 (arm_parse_option_features): Declare.
14962 (arm_intialize_isa): Declare.
14963 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
14965 (gen_comm_data): ... here. Make definitions non-static.
14966 * config/arm/arm-cpu-data.h: Regenerated.
14967 * config/arm/arm-cpu-cdata.h: Regenerated.
14969 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14971 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
14972 (cpu_arch_extension): New structure.
14973 (cpu_arch_option, arch_option, cpu_option): New structures.
14974 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
14975 architecture types.
14976 (gen_data): Generate new format data tables.
14977 * config/arm/arm.c (cpu_tune): New structure.
14978 (cpu_option, processors): Delete.
14979 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
14980 (arm_print_hint_for_cpu_option): ... this and ...
14981 (arm_print_hint_for_arch_option): ... this.
14982 (arm_parse_arch_cpu_name): Delete. Replace with ...
14983 (arm_parse_cpu_option_name): ... this and ...
14984 (arm_parse_arch_option_name): ... this.
14985 (arm_unrecognized_feature): Change type of target parameter to
14987 (arm_parse_arch_cpu_features): Delete. Replace with ...
14988 (arm_parse_option_features): ... this.
14989 (arm_configure_build_target): Rework to use new configuration data
14991 (arm_print_tune_info): Rework for new configuration data tables.
14992 * config/arm/arm-cpu-data.h: Regenerated.
14993 * config/arm/arm-cpu.h: Regenerated.
14995 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
14997 * Makefile.in (OBJS): Move sbitmap.o from here ...
14998 (OBJS-libcommon): ... to here.
15000 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15002 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
15003 (ISA_ALL_CRYPTO): New macro.
15004 (ISA_ALL_SIMD): New macro
15005 (ISA_ALL_FP): New macro.
15006 * config/arm/arm.c (fpu_bitlist): Update initializer.
15007 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
15009 (arm9e): Add fpu. Add option for nofp
15010 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
15011 (arm926ej-s, arm1026ej-s): Likewise.
15012 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
15013 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
15014 neon-fp16, neon-vfpv4, nofp and nosimd.
15015 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
15016 (cortex-a8): Add fpu. Add option for nofp.
15017 (cortex-a9): Add fpu. Add options for nosimd and nofp.
15018 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
15019 (cortex-r4f): Add fpu.
15020 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
15021 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
15023 (cortex-r8): Likewise.
15024 (cortex-m4): Add fpu. Add option for nofp.
15025 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
15026 (cortex-a17.cortex-a7): Likewise.
15027 (cortex-a32): Add fpu. Add options for crypto and nofp.
15028 (cortex-a35, cortex-a53): Likewise.
15029 (cortex-a57): Add fpu. Add option for crypto.
15030 (cortex-a72, cortex-a73): Likewise.
15031 (exynos-m1): Likewise.
15032 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
15033 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
15034 (cortex-m33): Add fpu. Add option for nofp.
15035 * config/arm/arm-cpu-cdata.h: Regenerated
15036 * config/arm/arm-cpu-data.h: Regenerated.
15038 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15040 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
15041 (armv5te, armv5tej): Likewise.
15042 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
15043 (armv7): Add options fp and vfpv3-d16.
15044 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
15045 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
15047 (armv7ve): Likewise.
15048 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
15049 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
15050 (armv8-a): Add nocrypto option.
15051 (armv8.1-a, armv8.2-a): Likewise.
15052 (armv8-m.main): add options fp, fp.dp and nofp.
15054 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15056 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
15058 (armv8-a+crc): Delete.
15059 (armv8.1-a): Add options simd, crypto and nofp.
15060 (armv8.2-a): Add options fp16, simd, crypto and nofp.
15061 (armv8.2-a+fp16): Delete.
15062 (armv8-m.main): Add option dsp.
15063 (armv8-m.main+dsp): Delete.
15064 (cortex-a8): Add fpu. Add nofp option.
15065 (cortex-a9): Add fpu. Add nofp and nosimd options.
15066 * config/arm/parsecpu.awk (gen_data): Generate option tables and
15067 link to main cpu and architecture data structures.
15068 (gen_comm_data): Only put isa attributes from the main architecture
15070 (option): New statement for architecture and CPU entries.
15071 * arm.c (struct cpu_option): New structure.
15072 (struct processors): Add entry for options.
15073 (arm_unrecognized_feature): New function.
15074 (arm_parse_arch_cpu_name): Ignore any characters after the first
15076 (arm_parse_arch_cpu_feature): New function.
15077 (arm_configure_build_target): Separate out any CPU and architecture
15078 features and parse separately. Don't error out if -mfpu=auto is
15079 used with only an architecture string.
15080 (arm_print_asm_arch_directives): New function.
15081 (arm_file_start): Call it.
15082 * config/arm/arm-cpu-cdata.h: Regenerated.
15083 * config/arm/arm-cpu-data.h: Likewise.
15084 * config/arm/arm-tables.opt: Likewise.
15086 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15088 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
15089 assembler when it is not -mfpu=auto.
15091 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15093 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
15094 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
15095 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
15096 (ASM_CPU_SPEC): Rewrite.
15097 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
15098 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
15099 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
15100 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
15101 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
15102 copied string is NUL-terminated. Also strip any characters prefixed
15104 (arm_rewrite_selected_arch): New function.
15105 (arm_rewrite_march): New function.
15107 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15109 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
15110 (x_arm_cpu_string, x_arm_tune_string): Likewise.
15111 (march, mcpu, mtune): Convert to string-based options.
15112 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
15113 (arm_parse_arch_cpu_name): New function.
15114 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
15115 identify selected architecture or CPU.
15116 (arm_option_save): New function.
15117 (TARGET_OPTION_SAVE): Redefine.
15118 (arm_option_restore): Restore string options.
15119 (arm_option_print): Print string options.
15121 2017-06-16 Martin Sebor <msebor@redhat.com>
15123 PR tree-optimization/80933
15124 PR tree-optimization/80934
15125 * builtins.c (fold_builtin_3): Do not handle bcmp here.
15126 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
15127 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
15128 (gimple_fold_builtin): Call them.
15130 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15132 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
15133 as unlikely; update profile.
15135 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15137 * predict.c (force_edge_cold): Handle declaring edges impossible
15140 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15142 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
15144 (try_unroll_loop_completely): Fix reporting.
15146 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15148 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
15150 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
15153 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
15154 if given a non-constant argument for an intrinsic which requires a
15157 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15159 * profile.c (compare_freqs): New function.
15160 (branch_prob): Sort edge list.
15161 (find_spanning_tree): Assume that the list is priority sorted.
15163 2017-06-16 Richard Biener <rguenther@suse.de>
15165 PR tree-optimization/81090
15166 * passes.def (pass_record_bounds): Remove.
15167 * tree-pass.h (make_pass_record_bounds): Likewise.
15168 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
15169 make_pass_record_bounds): Likewise.
15170 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
15171 not free niter estimates at the beginning but at the end.
15172 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
15174 2017-06-16 Richard Biener <rguenther@suse.de>
15176 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
15177 initializer to workaround ICE in host GCC 4.8.
15179 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15181 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
15183 (clone_inlined_nodes): Update.
15185 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15187 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
15188 prefetch settings, and enable prefetching by default at -O3.
15190 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15192 * config/aarch64/aarch64.c (aarch64_override_options_internal):
15193 Set flag_prefetch_loop_arrays according to tuning data.
15195 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15197 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
15198 New tune structure.
15199 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
15200 [Unrelated to main purpose of the patch] Place the pointer field last
15201 to enable type checking errors when tune structure are wrongly merged.
15202 * config/aarch64/aarch64.c (generic_prefetch_tune,)
15203 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
15204 (thunderx2t99_prefetch_tune): New tune constants.
15205 (tune_params *_tunings): Update all tunings (no functional change).
15206 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
15207 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
15208 from tunings structures.
15210 2017-06-16 Jakub Jelinek <jakub@redhat.com>
15213 * ubsan.c (instrument_null): Add T argument, use it instead
15214 of computing it based on IS_LHS.
15215 (instrument_object_size): Likewise.
15216 (pass_ubsan::execute): Adjust instrument_null and
15217 instrument_object_size callers to pass gimple_get_lhs or
15218 gimple_assign_rhs1 result to it. Use instrument_null instead of
15219 calling get_base_address and instrument_mem_ref. Handle
15220 aggregate call arguments for object-size sanitization.
15222 2017-06-16 Yury Gribov <tetra2005@gmail.com>
15224 PR tree-optimization/81089
15225 * tree-vrp.c (is_masked_range_test): Validate operands of
15228 2017-06-15 Martin Sebor <msebor@redhat.com>
15231 * dumpfile.c (dump_register): Avoid calling memset to initialize
15232 a class with a default ctor.
15233 * gcc.c (struct compiler): Remove const qualification.
15234 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
15235 * hash-table.h: Ditto.
15236 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
15238 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
15239 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
15241 * params.h (struct param_info): Make struct members non-const.
15242 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
15243 with default initialization.
15244 * vec.h (vec_copy_construct, vec_default_construct): New helper
15246 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
15247 with vec_copy_construct.
15248 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
15249 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
15250 * doc/invoke.texi (-Wclass-memaccess): Document.
15252 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15254 * emit-rtl.h (is_leaf): Update comment about local
15255 register allocator.
15257 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15260 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
15261 for a variable to have a section before checking if the section has a
15263 Set section to.persistent if persistent attribute is set.
15264 Warn if .persistent attribute is used on an automatic variable.
15266 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
15268 PR rtl-optimization/80474
15269 * reorg.c (update_block): Do not ignore instructions in a delay slot.
15271 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
15273 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
15276 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
15278 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
15279 (casesi): Emit bounds checking as RTL.
15280 (casesi_internal_mips16_<mode>): Remove bounds checking.
15282 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
15284 * config/xtensa/xtensa.c (xtensa_option_override): Append
15285 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
15286 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
15287 xtensa_doloop_hooks): Define unconditionally.
15288 (xtensa_reorg_loops): Only call reorg_loops in the presence of
15290 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
15291 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
15292 for it in xtensa_option_override.
15293 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
15294 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
15296 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
15298 * doc/cppopts.texi: Document '-' special value to -MF.
15300 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
15302 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
15303 (cortex_a53_fconst): Likewise.
15304 (cortex_a53_fpmul): Likewise.
15305 (cortex_a53_f_load_64): Likewise.
15306 (cortex_a53_f_load_many): Likewise.
15307 (cortex_a53_advsimd_alu): Likewise.
15308 (cortex_a53_advsimd_alu_q): Likewise.
15309 (cortex_a53_advsimd_mul): Likewise.
15310 (cortex_a53_advsimd_mul_q): Likewise.
15311 (fpmac bypass): Add new bypass for fpmac-fpmac case.
15312 Add missing fmul, r2f_cvt and fconst cases.
15314 2017-06-14 Richard Biener <rguenther@suse.de>
15316 PR middle-end/81088
15317 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
15319 (fold_binary_loc): When associating do not treat pre-existing
15320 TREE_OVERFLOW on literal constants as a reason to allow
15321 TREE_OVERFLOW on associated literal constants.
15323 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15325 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
15326 (MASK_FEATURES): New macro.
15327 * config/sparc/sparc.c (sparc_option_override): Remove the special
15328 handling of -mfpu and generalize it to all MASK_FEATURES switches.
15330 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15332 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
15333 a division of 0 if non-call exceptions are enabled.
15335 2017-06-14 Andrew Pinski <apinski@cavium.com>
15336 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15339 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
15340 Improve vector initialization code gen for only variable case.
15342 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15344 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
15346 2017-06-14 Richard Biener <rguenther@suse.de>
15348 PR tree-optimization/81083
15349 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
15352 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15354 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
15355 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
15356 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
15357 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
15358 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
15359 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
15361 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15363 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
15364 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
15366 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15368 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
15370 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15372 * config/rs6000/t-rtems: Don't handle SPE.
15374 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15376 * config/rs6000/t-linux: Don't handle SPE.
15378 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15380 * config/rs6000/eabispe.h: Delete file.
15382 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15384 * config/rs6000/t-spe: Delete file.
15386 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15388 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
15389 (rs6000_legitimate_offset_address_p): Return false for anything in
15390 V2SImode or V2SFmode.
15392 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15394 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
15395 except V2SF and V2SI. Rearrange the vector modes, and add comments.
15396 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
15398 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
15399 (rs6000_legitimate_offset_address_p): Ditto.
15400 (rs6000_emit_move): Ditto.
15401 (rs6000_init_builtins): Remove V4HI_type_node.
15403 2017-06-13 Martin Liska <mliska@suse.cz>
15406 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
15407 (gate_asan): Likewise.
15408 * asan.h (asan_no_sanitize_address_p): Remove the function.
15409 (sanitize_flags_p): New function.
15410 * builtins.def: Fix coding style.
15411 * common.opt: Use renamed enum value.
15412 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
15413 * doc/extend.texi: Document no_sanitize attribute.
15414 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
15415 to SANITIZE_UNDEFINED_NONDEFAULT.
15416 * gcc.c (sanitize_spec_function): Use the renamed enum value.
15417 * gimple-fold.c (optimize_atomic_compare_exchange_p):
15418 Use sanitize_flags_p.
15419 * gimplify.c (gimplify_function_tree): Likewise.
15420 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
15421 * opts.c (parse_no_sanitize_attribute): New function.
15422 (common_handle_option): Use renamed enum value.
15423 * opts.h (parse_no_sanitize_attribute): Declare.
15424 * tree.c (sanitize_flags_p): New function.
15425 * tree.h: Declared here.
15426 * tsan.c: Use sanitize_flags_p.
15427 * ubsan.c (ubsan_expand_null_ifn): Likewise.
15428 (instrument_mem_ref): Likewise.
15429 (instrument_bool_enum_load): Likewise.
15430 (do_ubsan_in_current_function): Remove the function.
15431 (pass_ubsan::execute): Use sanitize_flags_p.
15432 * ubsan.h: Remove do_ubsan_in_current_function
15433 * tree-cfg.c (print_no_sanitize_attr_value): New function.
15434 (dump_function_to_file): Use it here.
15436 2017-06-13 Martin Jambor <mjambor@suse.cz>
15438 PR tree-optimization/80803
15439 PR tree-optimization/81063
15440 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
15441 (propagate_subaccesses_across_link): Enqueue subtree whenever
15442 necessary instead of relying on the caller.
15444 2017-06-13 Martin Jambor <mjambor@suse.cz>
15446 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
15447 that have a first_link.
15448 (sort_and_splice_var_accesses): Do not check first_link before
15450 (subtree_mark_written_and_enqueue): Likewise.
15451 (propagate_all_subaccesses): Likewise and do not stop at first
15452 parent with a first_link.
15454 2017-06-13 Martin Jambor <mjambor@suse.cz>
15456 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
15459 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15461 * match.pd: New pattern.
15463 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15465 * tree-vrp.c (is_masked_range_test): New function.
15466 (register_edge_assert_for): Determine ranges for
15469 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15471 PR tree-optimization/67328
15472 * fold-const.c (maskable_range_p): New function.
15473 (build_range_check): Generate bittests if possible.
15475 2017-06-13 Martin Liska <mliska@suse.cz>
15477 * gimple-pretty-print.c (dump_probability): Add new argument.
15478 (dump_edge_probability): Dump both probability and count.
15479 (dump_gimple_label): Likewise.
15480 (dump_gimple_bb_header): Likewise.
15482 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
15485 * config/avr/avr-devices.c: Fix indentation.
15486 * config/avr/gen-avr-mmcu-specs.c: Dito.
15488 2017-06-13 Richard Biener <rguenther@suse.de>
15490 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
15491 instead get vector type from stmt_info.
15492 (vectorizable_reduction): Adjust. Remove dead code.
15494 2017-06-13 Richard Biener <rguenther@suse.de>
15496 PR middle-end/81065
15497 * fold-const.c (extract_muldiv_1): Remove bogus distribution
15498 case of C * (x * C2 + C3).
15499 (fold_addr_of_array_ref_difference): Properly fold index difference.
15501 2017-06-12 David S. Miller <davem@davemloft.net>
15504 * config/sparc/sparc.md (return expander): Emit frame blockage if
15505 function uses alloca.
15507 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
15509 * combine.c (make_field_assignment): Check len rather than the mode
15510 precision when calling force_to_mode.
15512 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
15514 Support multilibs and devices that see flash in RAM address range.
15517 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
15518 (avr_mcu_t) <flash_pm_offset>: New field.
15519 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
15520 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
15521 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
15522 (AVR_TINY_PM_OFFSET): Remove macro.
15523 * config/avr/avr.opt (-mshort-calls): New option.
15524 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15525 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
15526 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
15527 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
15528 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
15529 instead of avr_arch->have_jmp_call.
15530 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
15531 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
15532 avr_arch->flash_pm_offset to define.
15533 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
15534 new field flash_pm_offset. Add entry for avrxmega3.
15535 (avr_texinfo): Add entry for avrxmega3.
15536 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
15537 attiny212, attiny214,
15538 attiny412, attiny414, attiny416, attiny417,
15539 attiny814, attiny816, attiny817,
15540 attiny1614, attiny1616, attiny1617,
15541 attiny3214, attiny3216, attiny3217.
15542 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
15543 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
15544 (avr_print_operand_address) [AVR_TINY]: Same.
15545 (avr_asm_init_sections) <readonly_data_section>: Only patch
15546 callback if avr_arch->flash_pm_offset = 0.
15547 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
15548 for rodata if avr_arch->flash_pm_offset != 0.
15549 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
15550 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
15551 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
15552 (m_options): Append opt_rcall.
15553 (m_dirnames): Append dir_rcall.
15554 * config/avr/t-multilib: Regenerate.
15556 * configure.ac [target=avr]: Check whether avrxmega3 default
15557 linker description file works as needed.
15558 * configure: Regenerate.
15559 * doc/avr-mmcu.texi: Regenerate.
15560 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
15561 <__AVR_ARCH__>: Document avrxmega3 and 103.
15562 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
15563 <__AVR_SHORT_CALLS__>: Document it.
15564 <__AVR_PM_BASE_ADDRESS__>: Document it.
15565 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
15566 (AVR Variable Attributes) <progmem>: Document this is
15567 not needed for avrxmega3.
15568 (AVR Named Address Spaces) <__flash>: Dito.
15570 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
15572 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
15574 2017-06-12 Doug Rupp <rupp@adacore.com>
15576 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
15577 Append vxworks-stdint.h to the tm_file list.
15578 * config/vxworks-stdint.h: New file.
15580 2017-06-12 Martin Liska <mliska@suse.cz>
15582 PR tree-optimization/81041
15583 * tree-profile.c (gimple_gen_ic_func_profiler):
15584 Create an extra BB in profile-generate
15585 (gimple_gen_time_profiler): Likewise.
15587 2017-06-12 Jakub Jelinek <jakub@redhat.com>
15589 PR tree-optimization/81003
15590 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
15591 (update_range_test): Use it instead of force_gimple_operand_gsi.
15593 2017-06-12 Richard Biener <rguenther@suse.de>
15595 PR tree-optimization/81053
15596 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
15597 with backedge value not defined in loop. Simplify def stmt
15600 2017-06-11 Tom de Vries <tom@codesourcery.com>
15603 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
15605 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
15606 nvptx_cannot_force_const_mem.
15608 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15610 * opts.c (finish_options): Move test for flag_split_stack after
15611 it has been initialized.
15613 2017-06-11 Jason Merrill <jason@redhat.com>
15615 * tree.h (id_equal): New.
15616 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
15617 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
15618 instead of strcmp of IDENTIFIER_POINTER.
15620 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15622 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
15623 (mark_all_inlined_calls_cdtor): Fix formating.
15624 (inline_transform): Rescale profile before inlining.
15626 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15628 * cgraph.h (cgraph_edge::clone): Update prototype.
15629 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
15630 (cgraph_node::create_clone): Update.
15631 (cgraph_node::create_version_clone): Update.
15632 * tree-inline.c (copy_bb): Update.
15633 (expand_call_inline): Update.
15635 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15637 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
15638 factored out from ...
15639 (rs6000_emit_prologue): ... here.
15641 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15643 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
15644 factored out from ...
15645 (rs6000_emit_prologue): ... here.
15647 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15649 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
15651 (handle_missing_profiles): Fix computation of tp_first_run.
15652 (counts_to_freqs): Do not touch freqs when count is 0.
15654 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15656 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
15659 2017-06-10 Tom de Vries <tom@codesourcery.com>
15661 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
15662 attributes): Document signal effective target.
15664 2017-06-10 Tom de Vries <tom@codesourcery.com>
15666 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
15667 Document effective target stack_size.
15669 2017-06-09 David Malcolm <dmalcolm@redhat.com>
15671 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
15672 to the edit_context if they can be auto-applied.
15674 2017-06-9 Ian Lance Taylor <iant@golang.org>
15676 * opts.c (finish_options): If -fsplit-stack, disable implicit
15677 -forder-blocks-and-partition.
15678 * doc/invoke.texi (Optimize Options): Document that when using
15679 -fsplit-stack -forder-blocks-and-partition is not implicitly
15682 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15684 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
15685 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
15686 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
15687 * builtins.def (abort, trap, unreachable): Declare cold.
15688 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
15689 * tree-core.h (ECF_COLD): New.
15690 * tree.c (set_call_expr_flags): Handle ECF_COLD.
15691 (build_common_builtin_nodes): Mark unreachable and abort as cold.
15693 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15695 * predict.c (unlikely_executed_stmt_p): Cleanup.
15697 2017-06-09 Richard Biener <rguenther@suse.de>
15699 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
15700 model if the ref is always written to.
15702 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15704 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
15706 2017-06-09 Tamar Christina <tamar.christina@arm.com>
15708 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
15711 2017-06-09 Tom de Vries <tom@codesourcery.com>
15714 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
15715 "target cannot support label values" when encountering LABEL_REF.
15717 2017-06-09 Martin Liska <mliska@suse.cz>
15719 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
15720 (gimple_gen_ic_func_profiler): Emit direct comparison
15721 of __gcov_indirect_call_callee with NULL.
15722 (gimple_gen_time_profiler): Change probability from
15723 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
15725 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15727 * profile.c (edge_gcov_counts): Turn to pointer.
15728 (compute_branch_probabilities, compute_branch_probabilities): Update.
15729 (branch_prob): Do not clear edge_gcov_count.
15730 * profile.h (edge_gcov_counts): Turn to pointer.
15731 (edge_gcov_count): Update.
15733 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15735 * gimple.h (gimple_check_failed): Mark cold.
15737 2017-06-09 Richard Biener <rguenther@suse.de>
15739 PR tree-optimization/66623
15740 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
15741 refactor check_reduction into two parts, properly computing
15742 whether we have to check reduction validity for outer loop
15745 2017-06-09 Richard Biener <rguenther@suse.de>
15747 PR tree-optimization/79483
15748 * graphite-scop-detection.c (order): New global.
15749 (get_order): Compute bb to order mapping that satisfies code
15750 generation constraints.
15751 (cmp_pbbs): New helper.
15752 (build_scops): Start domwalk at entry block, sort generated
15755 2017-06-09 Richard Biener <rguenther@suse.de>
15757 PR middle-end/81007
15758 * ipa-polymorphic-call.c
15759 (ipa_polymorphic_call_context::restrict_to_inner_class):
15760 Skip FIELD_DECLs with error_mark_node type.
15761 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
15764 2017-06-09 Martin Liska <mliska@suse.cz>
15766 * predict.c (struct branch_predictor): New struct.
15767 (test_prediction_value_range): New test.
15768 (predict_c_tests): New function.
15769 * selftest-run-tests.c (selftest::run_tests): Run the function.
15770 * selftest.h: Declare new tests.
15772 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
15775 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
15776 gen_add3_insn did not fail.
15777 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
15778 r0, construct that number in a temporary reg and add that reg to r0.
15779 If asked to put the result in r0 as well, fail.
15781 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
15783 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15784 for early expansion of vec_eqv.
15786 2017-06-08 Jakub Jelinek <jakub@redhat.com>
15788 PR middle-end/81005
15789 * ubsan.c (instrument_null): Avoid pointless code temporary.
15790 (pass_ubsan::execute): Instrument aggregate arguments of calls.
15792 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
15796 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
15799 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
15802 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15804 * predict.c (unlikely_executed_edge_p): Move ahead.
15805 (probably_never_executed_edge_p): Use it.
15807 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15809 PR middle-end/79988
15810 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
15811 gimple_call_builtin_p call.
15813 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15815 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
15816 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
15817 rtl_check_failed_type2, rtl_check_failed_code1,
15818 rtl_check_failed_code2, rtl_check_failed_code_mode,
15819 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
15820 rtvec_check_failed_bounds, rtl_check_failed_flag,
15821 _fatal_insn_not_found, _fatal_insn): Likewise.
15822 * tree.h (tree_contains_struct_check_failed,
15823 tree_check_failed, tree_not_check_failed,
15824 tree_class_check_failed, tree_range_check_failed,
15825 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
15826 tree_vec_elt_check_failed, phi_node_elt_check_failed,
15827 tree_operand_check_failed, omp_clause_check_failed,
15828 omp_clause_operand_check_failed, omp_clause_range_check_failed):
15831 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15833 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
15834 flag_branch_probabilities.
15835 * ipa-inline.c (edge_badness): Likewise.
15836 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
15837 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
15838 * predict.c (maybe_hot_frequency_p): Likewise.
15839 (probably_never_executed): Likewise.
15840 * sched-ebb.c (schedule_ebbs): Likewise.
15841 * sched-rgn.c (find_single_block_region): Likewise.
15842 * tracer.c (tail_duplicate): Likewise.
15844 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15846 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
15847 longer requires x_flag_profile_use.
15849 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15851 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
15852 instead of flag_reorder_blocks_and_partition.
15853 * dbxout.c (dbxout_function_end): Likewise.
15854 * dwarf2out.c (gen_subprogram_die): Likewise.
15855 * haifa-sched.c (sched_create_recovery_edges): Likewise.
15856 * hw-doloop.c (reorg_loops): Likewise.
15857 * varasm.c (assemble_start_function,
15858 assemble_end_function): Likewise.
15859 (decide_function_section): Do not check for
15860 flag_reorder_blocks_and_partition.
15862 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
15864 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
15866 (chkp_get_hard_register_fake_addr_expr): Ditto.
15867 (chkp_build_addr_expr): Add check for hard reg case.
15868 (chkp_parse_array_and_component_ref): Ditto.
15869 (chkp_find_bounds_1): Ditto.
15870 (chkp_process_stmt): Don't generate bounds store for
15873 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15875 * predict.c (maybe_hot_bb_p): Do not check profile status.
15876 (maybe_hot_edge_p): Likewise.
15877 (probably_never_executed): Check for zero counts even if profile
15879 (unlikely_executed_edge_p): New function.
15880 (unlikely_executed_stmt_p): New function.
15881 (unlikely_executed_bb_p): New function.
15882 (set_even_probabilities): Use unlikely predicates.
15883 (combine_predictions_for_bb): Likewise.
15884 (predict_paths_for_bb): Likewise.
15885 (predict_paths_leading_to_edge): Likewise.
15886 (determine_unlikely_bbs): New function.
15887 (estimate_bb_frequencies): Use it.
15888 (compute_function_frequency): Use zero counts even if profile is
15890 * profile-count.h: Fix typo.
15892 2017-08-08 Julia Koval <julia.koval@intel.com>
15894 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
15895 _mm512_mask_cvtsepi16_storeu_epi8,
15896 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
15897 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
15898 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
15899 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
15900 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
15901 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
15902 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
15903 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
15904 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
15905 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
15906 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
15907 __builtin_ia32_pmovuswb256mem_mask,
15908 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
15909 __builtin_ia32_pmovwb512mem_mask): New builtins.
15911 2017-08-08 Julia Koval <julia.koval@intel.com>
15913 PR target/73350,80862
15914 * config/i386/subst.md (round): Fix round pattern.
15915 * config/i386/i386.c (ix86_erase_embedded_rounding):
15916 Fix erasing rounding for the fixed pattern.
15918 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
15920 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
15922 2017-06-08 Martin Liska <mliska@suse.cz>
15924 PR gcov-profile/80911
15925 * gcov.c (block_info::block_info): New constructor.
15927 2017-06-07 Carl Love <cel@us.ibm.com>
15929 * config/rs6000/rs6000-c: The return type of the following
15930 built-in functions was implemented as int not long long. Fix sign
15931 of return value for the unsigned version of vec_mulo and vec_mule.
15932 vector unsigned long long vec_bperm (vector unsigned long long,
15933 vector unsigned char)
15934 vector signed long long vec_mule (vector signed int,
15936 vector unsigned long long vec_mule (vector unsigned int,
15937 vector unsigned int)
15938 vector signed long long vec_mulo (vector signed int,
15940 vector unsigned long long vec_mulo (vector unsigned int,
15941 vector unsigned int)
15942 * doc/extend.texi: Fix the documentation for the built-in
15945 2017-06-07 Carl Love <cel@us.ibm.com>
15948 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
15951 2017-06-07 Carl Love <cel@us.ibm.com>
15953 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
15954 support, Generate doublehv for signed int/float for BE case only.
15956 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
15958 * doc/invoke.texi (mcx16): Rewrite.
15960 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15962 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
15963 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
15964 *mov<mode>_softfloat, and an anonymous splitter): Use
15965 nonimmediate_operand instead of rs6000_nonimmediate_operand.
15967 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15969 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
15971 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15972 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
15973 (rs6000_debug_reg_global): Adjust.
15974 (rs6000_init_hard_regno_mode_ok): Adjust.
15975 (rs6000_dbx_register_number): Adjust.
15976 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
15977 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15978 Remove SPE_ACC and SPEFSCR.
15979 (REG_ALLOC_ORDER): Ditto.
15980 (FRAME_POINTER_REGNUM): Change to 111.
15981 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
15982 (REG_CLASS_NAMES): Ditto.
15983 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
15984 (REGISTER_NAMES): Ditto.
15985 (ADDITIONAL_REG_NAMES): Ditto.
15986 (rs6000_reg_names): Ditto.
15987 * config/rs6000/rs6000.md: Renumber some register number
15990 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
15992 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
15994 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
15995 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
15997 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
15998 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
15999 Delete the SPE high registers.
16000 (REG_ALLOC_ORDER): Ditto.
16001 (enum reg_class): Remove SPE_HIGH_REGS.
16002 (REG_CLASS_NAMES): Ditto.
16003 (REG_CLASS_CONTENTS): Delete the SPE high registers.
16004 (REGISTER_NAMES): Ditto.
16005 (rs6000_reg_names): Ditto.
16006 * doc/tm.texi.in: Remove SPE as example.
16007 * doc/tm.texi: Regenerate.
16009 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16011 * config/rs6000/8540.md (ppc8540_brinc): Delete.
16012 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
16013 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
16014 * config/rs6000/rs6000.md (type): Remove "brinc".
16016 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16018 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
16019 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
16020 * config/rs6000/linuxspe.h: Delete file.
16021 * config/rs6000/rs6000.md: Don't include spe.md.
16022 * config/rs6000/spe.h: Delete file.
16023 * config/rs6000/spe.md: Delete file.
16024 * config/rs6000/t-rs6000: Remove spe.md.
16026 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16028 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
16029 (reg_or_none500mem_operand): Delete.
16030 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
16031 instead of reg_or_none500mem_operand.
16033 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16035 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16036 handling of SPE flags.
16037 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
16039 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16041 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
16043 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
16044 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
16045 paired_divv2sf3): Similar.
16046 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
16047 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
16048 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
16050 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
16051 Rename the paired_* instruction patterns.
16052 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
16054 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
16055 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
16056 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
16057 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
16058 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
16059 PAIRED_VECTOR_MODE.
16060 (struct machine_function): Delete field spe_insn_chain_scanned_p.
16061 (spe_func_has_64bit_regs_p): Delete.
16062 (spe_expand_predicate_builtin): Delete.
16063 (spe_expand_evsel_builtin): Delete.
16064 (TARGET_DWARF_REGISTER_SPAN): Do not define.
16065 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
16066 (invalid_e500_subreg): Delete.
16067 (rs6000_legitimize_address): Always force_reg op2 as well, for
16068 paired single memory accesses.
16069 (rs6000_member_type_forces_blk): Delete.
16070 (rs6000_spe_function_arg): Delete.
16071 (rs6000_expand_unop_builtin): Delete SPE handling.
16072 (rs6000_expand_binop_builtin): Ditto.
16073 (spe_expand_stv_builtin): Delete.
16074 (bdesc_2arg_spe): Delete.
16075 (spe_expand_builtin): Delete.
16076 (spe_expand_predicate_builtin): Delete.
16077 (spe_expand_evsel_builtin): Delete.
16078 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
16079 (spe_init_builtins): Delete.
16080 (spe_func_has_64bit_regs_p): Delete.
16081 (savres_routine_name): Delete "info" parameter. Adjust callers.
16082 (rs6000_emit_stack_reset): Ditto.
16083 (rs6000_dwarf_register_span): Delete.
16084 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
16085 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
16086 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
16088 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
16090 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
16091 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
16092 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
16093 mulv2sf3, divv2sf3): Delete expanders.
16095 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16097 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
16099 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16101 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
16102 * config/rs6000/rs6000.c: Ditto.
16104 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16106 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
16107 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
16108 comparison_operator.
16110 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16112 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
16113 * config/rs6000/rs6000.opt: Ditto.
16114 * config/rs6000/t-rtems: Ditto.
16116 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16118 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
16119 TARGET_E500_SINGLE by 0, simplify.
16120 * config/rs6000/rs6000.c: Ditto.
16121 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
16122 (spe_build_register_parallel): Delete.
16123 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
16124 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
16125 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
16126 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
16127 (E500_CONVERT): Delete.
16128 * config/rs6000/spe.md: Remove many patterns and all define_constants.
16130 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16132 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
16133 * config/rs6000/dfp.md: Ditto.
16134 (negdd2, *negdd2_fpr): Merge.
16135 (absdd2, *absdd2_fpr): Merge.
16136 (negtd2, *negtd2_fpr): Merge.
16137 (abstd2, *abstd2_fpr): Merge.
16138 * config/rs6000/e500.h: Delete file.
16139 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
16140 TARGET_FPRS by 1 and simplify.
16141 * config/rs6000/rs6000-c.c: Ditto.
16142 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
16143 TARGET_DF_SPE by 0.
16144 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
16146 * config/rs6000/rs6000.md: Ditto.
16147 (floatdidf2, *floatdidf2_fpr): Merge.
16148 (move_from_CR_gt_bit): Delete.
16149 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
16150 (E500_CR_IOR_COMPARE): Delete.
16151 (All patterns that require !TARGET_FPRS): Delete.
16152 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
16154 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16156 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
16158 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16160 * graphds.c (add_edge): Intitialize edge's attached data.
16161 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
16162 pointer parameter. Call pointed function on each edge during
16163 graph traversing. Skip traversing the edge when the function
16165 (graphds_dfs, graphds_scc): Ditto.
16166 (for_each_edge): New parameter. Pass the new parameter to callback
16168 * graphds.h (skip_edge_callback): New function pointer type.
16169 (graphds_dfs, graphds_scc): New function pointer parameter.
16170 (graphds_edge_callback, for_each_edge): New parameter.
16172 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16174 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
16175 out code checking if runtime alias check is possible to below ...
16176 Call the new function.
16177 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
16178 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
16180 2017-06-07 Marek Polacek <polacek@redhat.com>
16183 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
16184 TYPE_OVERFLOW_WRAPS check.
16186 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16188 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
16189 if versioning is required.
16190 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
16191 peeling with the check for versioning.
16193 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16195 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
16196 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
16197 Set true to new parameter if new ssa variable is defined.
16198 (vect_gen_vector_loop_niters): Refactor. Set range information
16199 for the new vector loop bound variable.
16200 (vect_do_peeling): Ditto.
16202 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16204 * tree-affine.c (ssa.h): Include header file.
16205 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
16206 has wrapping overflow behavior.
16208 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16210 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
16212 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16214 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
16215 (tree_to_aff_combination): ... here.
16217 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16219 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
16220 reg_pressure model function.
16221 (ivopts_global_cost_for_size): Delete.
16222 (determine_set_costs, iv_ca_recount_cost): Call new model function
16223 ivopts_estimate_reg_pressure.
16225 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16227 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
16228 expensive than udiv. Remove floating point cases from mod.
16230 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16232 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
16233 Increase idiv cost.
16235 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16237 * config/aarch64/aarch64.md
16238 (copysignsf3): Fix mask generation.
16240 2017-06-07 Jakub Jelinek <jakub@redhat.com>
16242 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
16244 (class dump_manager): Add register_dumps method.
16245 * dumpfile.c: Include langhooks.h.
16246 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
16247 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
16248 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
16249 (dump_manager::dump_register): Start with 512 entries instead of 32.
16250 (dump_manager::register_dumps): New method.
16251 * toplev.c (general_init): Instead of invoking register_dumps
16252 langhook, invoke register_dumps method on the dump manager.
16253 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
16256 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
16258 * doc/md.texi: Clarify the restrictions on a define_insn condition.
16259 Say that # requires an associated define_split to exist, and that
16260 the define_split must be suitable for use after register allocation.
16262 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16264 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
16265 (compute_outgoing_frequencies): Also initialize zero counts.
16266 (find_many_sub_basic_blocks): Do not produce uninitialized profile
16267 around loops; preserve more of profile when nothing changes.
16269 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
16271 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
16273 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
16274 * config/arm/arm-cpu-cdata.h: Regenerate.
16275 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
16276 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
16277 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
16278 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
16279 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
16281 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
16283 * config/arm/t-rmprofile: Likewise.
16284 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
16286 2017-06-06 David S. Miller <davem@davemloft.net>
16289 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
16290 blockage if function uses alloca.
16292 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16294 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
16295 New "uid" fields to hold pretty-print IDs of group and ref.
16296 Memory references are now identified as <group_id>:<ref_id>
16297 instead of using [random] addresses.
16298 (dump_mem_details): Simplify, no functional change.
16299 (dump_mem_ref): Simplify and make output more concise.
16300 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
16301 (find_or_create_group): Initialize group uid.
16302 (record_ref): Initialize ref uid. Improve debug output.
16303 (prune_group_by_reuse, should_issue_prefetch_p,)
16304 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
16305 (mark_nontemporal_store, determine_loop_nest_reuse):
16306 Improve debug output.
16308 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16310 * dbgcnt.def (prefetch): New debug counter.
16311 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
16312 (schedule_prefetches): Stop issueing prefetches if debug counter
16315 2017-06-06 Tom de Vries <tom@codesourcery.com>
16317 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
16318 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
16320 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16322 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
16323 Use aarch64_reg_or_zero predicate for operand 4.
16324 (aarch64_compare_and_swap<mode> define_insn_and_split):
16325 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
16326 (aarch64_store_exclusive<mode>): Likewise for operand 2.
16328 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16330 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
16331 (arm_compute_save_core_reg_mask): This.
16332 (thumb1_compute_save_reg_mask): Rename into ...
16333 (thumb1_compute_save_core_reg_mask): This.
16334 (arm_compute_save_reg0_reg12_mask): Adapt comment.
16335 (arm_compute_frame_layout): Likewise.
16337 2017-06-06 Richard Biener <rguenther@suse.de>
16339 PR tree-optimization/80974
16340 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
16341 keep or clear leaders SSA info.
16343 2017-06-06 Tom de Vries <tom@codesourcery.com>
16345 * config/nvptx/nvptx.c (split_mode_p): New function.
16346 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
16348 2017-06-06 Tom de Vries <tom@codesourcery.com>
16350 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
16352 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16355 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
16358 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16360 * shrink-wrap.c (handle_simple_exit): Update profile.
16361 (try_shrink_wrapping): Upate profile.
16363 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16365 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
16366 (tree_guess_outgoing_edge_probabilities): New.
16367 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
16368 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
16370 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16372 * ipa-split.c (split_function): Initialize return bb profile.
16374 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16376 * profile.c (compute_branch_probabilities): Also initialize
16377 EXIT_BLOCK profile.
16379 2017-06-06 Richard Biener <rguenther@suse.de>
16381 PR tree-optimization/80928
16382 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
16383 (vect_analyze_loop_operations): Properly guard analysis for
16385 (vect_transform_loop): Likewise.
16386 (vect_analyze_loop_2): Also reset SLP type on PHIs.
16387 (vect_model_induction_cost): Do not cost for pure SLP.
16388 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
16389 of induction in inner loop vectorization.
16390 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
16391 (vect_get_and_check_slp_defs): Handle vect_induction_def.
16392 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
16394 (vect_analyze_slp_cost_1): Cost induction.
16395 (vect_detect_hybrid_slp_stmts): Handle PHIs.
16396 (vect_get_slp_vect_defs): Likewise.
16397 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
16398 (vect_transform_stmt): Handle SLP reductions.
16399 * tree-vectorizer.h (vectorizable_induction): Adjust.
16401 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16403 * config/rs6000/rs6000.c (make_resolver_func): Update
16404 init_lowered_empty_function call.
16406 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
16408 * doc/invoke.texi: Document the -fprofile-abs-path option.
16409 * common.opt (fprofile-abs-path): New option.
16410 * gcov-io.h (gcov_write_filename): Declare.
16411 * gcov-io.c (gcov_write_filename): New function.
16412 * coverage.c (coverage_begin_function): Use gcov_write_filename.
16413 * profile.c (output_location): Likewise.
16415 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16417 * shring-wrap.c: Revert accidental commit.
16419 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
16421 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
16423 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16425 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
16427 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
16428 profile in callgraph edge.
16429 * profile-count.h (apply_probability): If THIS is 0, then result is 0
16430 (apply_scale): Likewise.
16431 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
16432 Also scale profile when inlining function with zero profile.
16433 (initialize_cfun): Update exit block profile even when it is zero.
16434 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
16435 when profile is read.
16437 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16439 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
16440 (CLONE_*): New constants to define the processors we can generate
16441 code for with the target_clone attribute.
16442 (rs6000_clone_map): New array to identify which clone processors
16443 the current program is running on.
16444 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
16445 target_clone attribute.
16446 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
16447 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
16448 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
16449 (cpu_expand_builtin): Add support for target_clone attribute.
16450 (rs6000_valid_attribute_p): Allow "default" attribute.
16451 (get_decl_name): New debug function to simplify printing the
16452 current function name in debugging statements.
16453 (rs6000_clone_priority): New functions to support the target_clone
16454 attribute, and be able to generate code to switch between ISA 2.05
16455 through ISA 3.0 (power6 through power9).
16456 (rs6000_compare_version_priority): Likewise.
16457 (rs6000_get_function_versions_dispatcher): Likewise.
16458 (make_resolver_func): Likewise.
16459 (add_condition_to_bb): Likewise.
16460 (dispatch_function_versions): Likewise.
16461 (rs6000_generate_version_dispatcher_body): Likewise.
16462 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
16463 (fusion_gpr_load_p): Fix a spacing issue.
16464 * doc/extend.texi (Common Function Attributes): Document that the
16465 PowerPC supports the target_clone attribute.
16467 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16469 * config/arm/arm.h: explain F symbol found in description of ARM
16470 register allocation in its legend.
16472 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16474 * config/mips/frame-header-opt.c: Include profile-count.h.
16475 * config/riscv/riscv.c: Include profile-count.h
16477 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16479 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
16481 (sm_set_flag_if_changed): Add bbs field.
16482 (execute_sm_if_changed_flag_set): Pass BBS.
16483 (execute_sm): Update.
16485 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16487 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
16490 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16492 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
16493 (peephole2): New peephole2 to emit the above.
16494 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
16496 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16498 * config/aarch64/aarch64.c (define_peephole2 above
16499 *sub_<shift>_<mode>): New peephole.
16501 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16503 * config/i386/i386.c (make_resolver_func): Update.
16504 * Makefile.in: Add profile-count.h and profile-count.o
16505 * auto-profile.c (afdo_indirect_call): Update to new API.
16506 (afdo_set_bb_count): Update.
16507 (afdo_propagate_edge): Update.
16508 (afdo_propagate_circuit): Update.
16509 (afdo_calculate_branch_prob): Update.
16510 (afdo_annotate_cfg): Update.
16511 * basic-block.h: Include profile-count.h
16512 (struct edge_def): Turn count to profile_count.
16513 (struct basic_block_def): Likewie.
16514 (REG_BR_PROB_BASE): Move to profile-count.h
16515 (RDIV): Move to profile-count.h
16516 * bb-reorder.c (max_entry_count): Turn to profile_count.
16517 (find_traces): Update.
16518 (rotate_loop):Update.
16519 (connect_traces):Update.
16520 (sanitize_hot_paths):Update.
16521 * bt-load.c (migrate_btr_defs): Update.
16522 * cfg.c (RDIV): Remove.
16523 (init_flow): Use alloc_block.
16524 (alloc_block): Uninitialize count.
16525 (unchecked_make_edge): Uninitialize count.
16526 (check_bb_profile): Update.
16527 (dump_edge_info): Update.
16528 (dump_bb_info): Update.
16529 (update_bb_profile_for_threading): Update.
16530 (scale_bbs_frequencies_int): Update.
16531 (scale_bbs_frequencies_gcov_type): Update.
16532 (scale_bbs_frequencies_profile_count): New.
16533 * cfg.h (update_bb_profile_for_threading): Update.
16534 (scale_bbs_frequencies_profile_count): Declare.
16535 * cfgbuild.c (compute_outgoing_frequencies): Update.
16536 (find_many_sub_basic_blocks): Update.
16537 * cfgcleanup.c (try_forward_edges): Update.
16538 (try_crossjump_to_edge): Update.
16539 * cfgexpand.c (expand_gimple_tailcall): Update.
16540 (construct_exit_block): Update.
16541 * cfghooks.c (verify_flow_info): Update.
16542 (dump_bb_for_graph): Update.
16543 (split_edge): Update.
16544 (make_forwarder_block): Update.
16545 (duplicate_block): Update.
16546 (account_profile_record): Update.
16547 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
16548 (get_estimated_loop_iterations): Update.
16549 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
16550 (single_likely_exit): Update.
16551 * cfgloopmanip.c (scale_loop_profile): Update.
16553 (set_zero_probability): Update.
16554 (lv_adjust_loop_entry_edge): Update.
16555 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16556 (purge_dead_edges): Update.
16557 (rtl_account_profile_record): Update.
16558 * cgraph.c (cgraph_node::create): Uninitialize count.
16559 (symbol_table::create_edge): Uninitialize count.
16560 (cgraph_update_edges_for_call_stmt_node): Update.
16561 (cgraph_edge::dump_edge_flags): Update.
16562 (cgraph_node::dump): Update.
16563 (cgraph_edge::maybe_hot_p): Update.
16564 * cgraph.h: Include profile-count.h
16565 (create_clone), create_edge, create_indirect_edge): Update.
16566 (cgraph_node): Turn count to profile_count.
16567 (cgraph_edge0: Likewise.
16568 (make_speculative, clone): Update.
16569 (create_edge): Update.
16570 (init_lowered_empty_function): Update.
16571 * cgraphclones.c (cgraph_edge::clone): Update.
16572 (duplicate_thunk_for_node): Update.
16573 (cgraph_node::create_clone): Update.
16574 * cgraphunit.c (cgraph_node::analyze): Update.
16575 (cgraph_node::expand_thunk): Update.
16576 * final.c (dump_basic_block_info): Update.
16577 * gimple-streamer-in.c (input_bb): Update.
16578 * gimple-streamer-out.c (output_bb): Update.
16579 * graphite.c (print_global_statistics): Update.
16580 (print_graphite_scop_statistics): Update.
16581 * hsa-brig.c: Include basic-block.h.
16582 * hsa-dump.c: Include basic-block.h.
16583 * hsa-gen.c (T sum_slice): Update.
16584 (convert_switch_statements):Update.
16585 * hsa-regalloc.c: Include basic-block.h.
16586 * ipa-chkp.c (chkp_produce_thunks): Update.
16587 * ipa-cp.c (struct caller_statistics): Update.
16588 (init_caller_stats): Update.
16589 (gather_caller_stats): Update.
16590 (ipcp_cloning_candidate_p): Update.
16591 (good_cloning_opportunity_p): Update.
16592 (get_info_about_necessary_edges): Update.
16593 (dump_profile_updates): Update.
16594 (update_profiling_info): Update.
16595 (update_specialized_profile): Update.
16596 (perhaps_add_new_callers): Update.
16597 (decide_about_value): Update.
16598 (ipa_cp_c_finalize): Update.
16599 * ipa-devirt.c (struct odr_type_warn_count): Update.
16600 (struct decl_warn_count): Update.
16601 (struct final_warning_record): Update.
16602 (possible_polymorphic_call_targets): Update.
16603 (ipa_devirt): Update.
16604 * ipa-fnsummary.c (redirect_to_unreachable): Update.
16605 * ipa-icf.c (sem_function::merge): Update.
16606 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
16607 * ipa-inline.c (compute_uninlined_call_time): Update.
16608 (compute_inlined_call_time): Update.
16609 (want_inline_small_function_p): Update.
16610 (want_inline_self_recursive_call_p): Update.
16611 (edge_badness): Update.
16612 (lookup_recursive_calls): Update.
16613 (recursive_inlining): Update.
16614 (inline_small_functions): Update.
16615 (dump_overall_stats): Update.
16616 (dump_inline_stats): Update.
16617 * ipa-profile.c (ipa_profile_generate_summary): Update.
16618 (ipa_propagate_frequency): Update.
16619 (ipa_profile): Update.
16620 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16621 * ipa-utils.c (ipa_merge_profiles): Update.
16622 * loop-doloop.c (doloop_modify): Update.
16623 * loop-unroll.c (report_unroll): Update.
16624 (unroll_loop_runtime_iterations): Update.
16625 * lto-cgraph.c (lto_output_edge): Update.
16626 (lto_output_node): Update.
16627 (input_node): Update.
16628 (input_edge): Update.
16629 (merge_profile_summaries): Update.
16630 * lto-streamer-in.c (input_cfg): Update.
16631 * lto-streamer-out.c (output_cfg): Update.
16632 * mcf.c (create_fixup_graph): Update.
16633 (adjust_cfg_counts): Update.
16634 (sum_edge_counts): Update.
16635 * modulo-sched.c (sms_schedule): Update.
16636 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
16637 * predict.c (maybe_hot_count_p): Update.
16638 (probably_never_executed): Update.
16639 (dump_prediction): Update.
16640 (combine_predictions_for_bb): Update.
16641 (propagate_freq): Update.
16642 (handle_missing_profiles): Update.
16643 (counts_to_freqs): Update.
16644 (rebuild_frequencies): Update.
16645 (force_edge_cold): Update.
16646 * predict.h: Include profile-count.h
16647 (maybe_hot_count_p, counts_to_freqs): UPdate.
16648 * print-rtl-function.c: Do not include cfg.h
16649 * print-rtl.c: Include basic-block.h
16650 * profile-count.c: New file.
16651 * profile-count.h: New file.
16652 * profile.c (is_edge_inconsistent): Update.
16653 (correct_negative_edge_counts): Update.
16654 (is_inconsistent): Update.
16655 (set_bb_counts): Update.
16656 (read_profile_edge_counts): Update.
16657 (compute_frequency_overlap): Update.
16658 (compute_branch_probabilities): Update; Initialize and deinitialize
16660 (branch_prob): Update.
16661 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
16662 (edge_gcov_count): New.
16663 (bb_gcov_count): New.
16664 * shrink-wrap.c (try_shrink_wrapping): Update.
16665 * tracer.c (better_p): Update.
16666 * trans-mem.c (expand_transaction): Update.
16667 (ipa_tm_insert_irr_call): Update.
16668 (ipa_tm_insert_gettmclone_call): Update.
16669 * tree-call-cdce.c: Update.
16670 * tree-cfg.c (gimple_duplicate_sese_region): Update.
16671 (gimple_duplicate_sese_tail): Update.
16672 (gimple_account_profile_record): Update.
16673 (execute_fixup_cfg): Update.
16674 * tree-inline.c (copy_bb): Update.
16675 (copy_edges_for_bb): Update.
16676 (initialize_cfun): Update.
16677 (freqs_to_counts): Update.
16678 (copy_cfg_body): Update.
16679 (expand_call_inline): Update.
16680 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16681 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
16682 (try_unroll_loop_completely): Update.
16683 (try_peel_loop): Update.
16684 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16685 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
16686 * tree-ssa-loop-split.c (connect_loops): Update.
16687 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
16688 * tree-ssa-reassoc.c (branch_fixup): Update.
16689 * tree-ssa-tail-merge.c (replace_block_by): Update.
16690 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
16691 (compute_path_counts): Update.
16692 (update_profile): Update.
16693 (recompute_probabilities): Update.
16694 (update_joiner_offpath_counts): Update.
16695 (estimated_freqs_path): Update.
16696 (freqs_to_counts_path): Update.
16697 (clear_counts_path): Update.
16698 (ssa_fix_duplicate_block_edges): Update.
16699 (duplicate_thread_path): Update.
16700 * tree-switch-conversion.c (case_bit_test_cmp): Update.
16701 (struct switch_conv_info): Update.
16702 * tree-tailcall.c (decrease_profile): Update.
16703 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
16704 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16705 * value-prof.c (check_counter): Update.
16706 (gimple_divmod_fixed_value): Update.
16707 (gimple_mod_pow2): Update.
16708 (gimple_mod_subtract): Update.
16709 (gimple_ic_transform): Update.
16710 (gimple_stringop_fixed_value): Update.
16711 * value-prof.h (gimple_ic): Update.
16713 2017-06-02 Carl Love <cel@us.ibm.com>
16715 * config/rs6000/rs6000-c: Add support for built-in functions
16716 vector double vec_doublee (vector signed int);
16717 vector double vec_doublee (vector unsigned int);
16718 vector double vec_doublee (vector float);
16719 vector double vec_doubleh (vector signed int);
16720 vector double vec_doubleh (vector unsigned int);
16721 vector double vec_doubleh (vector float);
16722 vector double vec_doublel (vector signed int);
16723 vector double vec_doublel (vector unsigned int);
16724 vector double vec_doublel (vector float);
16725 vector double vec_doubleo (vector signed int);
16726 vector double vec_doubleo (vector unsigned int);
16727 vector double vec_doubleo (vector float);.
16728 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
16729 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
16731 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
16732 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
16733 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
16735 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
16736 vec_doublel, vec_doubleh.
16737 * doc/extend.texi: Update the built-in documentation file for the
16738 new built-in functions.
16740 2017-06-02 David Malcolm <dmalcolm@redhat.com>
16743 * ipa-inline-analysis.c (free_growth_caches): Set
16744 edge_removal_hook_holder to NULL after removing it.
16746 2017-06-02 Sudakshina Das <sudi.das@arm.com>
16748 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
16749 comparision with zero.
16751 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16752 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16753 for early expansion of vec_min and vec_max builtins.
16754 (builtin_function_type): Add min/max unsigned variants to those
16755 identified as having unsigned arguments.
16757 2017-06-02 Olivier Hainque <hainque@adacore.com>
16759 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
16761 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16763 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
16764 Use VALL_F16 iterator rather than VALL.
16766 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16768 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
16769 Emit CBNZ inside loop when doing a strong exchange and comparing
16770 against zero. Generate the CC flags after the loop.
16772 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
16774 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
16775 (dl_section_ref): New.
16776 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
16777 On AIX, append an expression to subtract the size of the
16778 section length to dl_section_ref.
16780 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
16782 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16783 for early expansion of vector absolute builtins.
16785 2017-06-02 Richard Biener <rguenther@suse.de>
16787 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
16788 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
16790 2017-06-02 Richard Biener <rguenther@suse.de>
16792 PR tree-optimization/80948
16793 * tree-tailcall.c (find_tail_calls): Track stmts to move in
16794 stmt order as well.
16796 2017-06-02 Richard Biener <rguenther@suse.de>
16798 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
16800 * tree-vect-stmts.c (process_use): Do not mark backedge defs
16801 for inductions as relevant.
16803 2017-06-02 Richard Biener <rguenther@suse.de>
16805 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
16806 (vectorizable_induction): ... this. Remove dead code.
16808 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
16810 * builtins. (expand_builtin_alloca): Remove second parameter and
16811 infer its value from the first parameter instead.
16812 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
16814 2017-06-02 Jakub Jelinek <jakub@redhat.com>
16816 PR rtl-optimization/80903
16817 * loop-doloop.c (add_test): Unshare sequence.
16819 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
16821 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
16823 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
16825 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
16827 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
16828 xlogue_layout::get_instance, logue_layout::xlogue_layout,
16829 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
16830 (xlogue_layout::get_stub_rtx): Make static.
16831 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
16832 (xlogue_layout::compute_stub_managed_regs): Rename to...
16833 (xlogue_layout::count_stub_managed_regs): ...this.
16834 (xlogue_layout::is_stub_managed_reg): New function.
16835 (xlogue_layout::m_stub_names): Rename to...
16836 (xlogue_layout::s_stub_names): ...this, make static.
16837 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
16838 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
16839 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
16840 xlogue_layout::s_stub_names): Instantiate statics.
16841 (stub_managed_regs): Remove.
16842 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
16843 (disable_call_ms2sysv_xlogues): Rename to...
16844 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
16845 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
16847 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
16848 change after reload_completed.
16849 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
16851 (ix86_expand_prologue): Likewise.
16852 (ix86_expand_epilogue): Likewise.
16853 (ix86_expand_split_stack_prologue): Likewise.
16854 (ix86_compute_frame_layout): Remove frame parameter ...
16855 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
16856 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
16858 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
16859 (ix86_frame): Move from here ...
16860 * config/i386/i386.h (ix86_frame): ... to here.
16861 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
16862 complete ix86_frame data structure instead. Remove some_ld_name.
16864 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16866 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
16867 symbols that hold a DECL_VALUE_EXPR.
16869 2017-06-01 Martin Jambor <mjambor@suse.cz>
16871 PR tree-optimization/80898
16872 * tree-sra.c (process_subtree_disqualification): Removed.
16873 (disqualify_candidate): Do not acll
16874 process_subtree_disqualification.
16875 (subtree_mark_written_and_enqueue): New function.
16876 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
16877 RHS has been disqualified and re-queue LHS if necessary. Apart
16878 from that, ignore disqualified RHS.
16880 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16882 * config/s390/s390.c (s390_emit_epilogue): Disable early return
16883 address fetch for z10 or later.
16885 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16887 * config/arc/arc.md (tst_movb): Add guard when splitting.
16889 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16891 * config/arc/arc.c (arc_can_eliminate): Test against
16892 arc_frame_pointer_needed.
16894 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16896 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
16897 to prevent store reordering.
16898 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
16899 (type): Add block type.
16900 (stack_tie): Define special instruction to be used in
16903 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16905 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
16906 constraint. It is not valid for the pattern.
16907 (noncommutative_binary_comparison): Likewise.
16909 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16911 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
16914 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16916 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
16917 be used by the reg-alloc.
16919 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16921 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
16922 reg-alloc when having mul64 or mul32x16 instructions.
16923 (mulsidi3): Likewise.
16924 (umulsidi3): Likewise.
16925 (mulsi32x16): New pattern.
16926 (mulsi64): Likewise.
16927 (mulsidi64): Likewise.
16928 (umulsidi64): Likewise.
16929 (MUL32x16_REG): Define.
16930 (mul64_600): Use MUL32x16_REG.
16931 (mac64_600): Likewise.
16932 (umul64_600): Likewise.
16933 (umac64_600): Likewise.
16935 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
16937 * config/arc/arc.md (mulsi3_700): Make it commutative.
16939 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
16941 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
16943 (*sign_extendsidi2_insn): Likewise for movstosw.
16945 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
16947 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
16948 the type of the input discriminant value. Convert the
16949 discriminant value of signedness vary.
16951 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
16953 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
16954 Add to -Wall section.
16956 2017-06-01 Richard Biener <rguenther@suse.de>
16958 PR middle-end/66313
16959 * fold-const.c (fold_plusminus_mult_expr): If the factored
16960 factor may be zero use a wrapping type for the inner operation.
16961 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
16962 and handle moved defs.
16963 (process_assignment): Properly guard the unary op case. Return a
16964 tri-state indicating that moving the stmt before the call may allow
16965 to continue. Pass through to_move.
16966 (find_tail_calls): Handle moving unrelated defs before
16969 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
16972 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
16973 splitter result in the canonical way.
16975 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
16977 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
16978 also for 32bit target. Update insn attributes.
16979 (zero-extendsidi2 splitter): Allow all registers for operand 1.
16981 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
16983 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
16984 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
16985 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
16986 (_mm_maskz_min_ss): New intrinsics.
16988 2017-05-31 Martin Liska <mliska@suse.cz>
16990 * tree-vect-loop.c (vect_create_epilog_for_reduction):
16991 Change comment style to one we normally use.
16992 (vectorizable_reduction): Likewise.
16993 (vectorizable_induction): Likewise.
16994 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
16995 (vectorizable_call): Likewise.
16996 (vectorizable_simd_clone_call): Likewise.
16997 (vectorizable_conversion): Likewise.
16998 (vectorizable_assignment): Likewise.
16999 (vectorizable_shift): Likewise.
17000 (vectorizable_operation): Likewise.
17001 (vectorizable_store): Likewise.
17002 (vectorizable_load): Likewise.
17003 * tree-vectorizer.h: Likewise.
17005 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
17007 * passes.c (emergency_dump_function): New.
17008 * tree-pass.h (emergency_dump_function): Declare.
17009 * plugin.c (plugins_internal_error_function): Remove.
17010 * plugin.h (plugins_internal_error_function): Remove declaration.
17011 * toplev.c (internal_error_function): New static function. Use it...
17012 (general_init): ...here.
17014 2017-05-31 Graham Markall <graham.markall@embecosm.com>
17016 * config/arc/arc.c (arc_print_operand): Handle constant operands.
17017 (arc_rtx_costs): Add costs for new patterns.
17018 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
17019 * config/arc/predicates.md: Add _1_2_3_operand predicate.
17021 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
17023 * tree-ssa-strlen.c (get_next_strinfo): New function.
17024 (get_stridx_plus_constant): Use it.
17025 (zero_length_string): Likewise.
17026 (adjust_related_strinfos): Likewise.
17027 (adjust_last_stmt): Likewise.
17029 2017-05-31 Richard Biener <rguenther@suse.de>
17032 * config/i386/i386.c (ix86_expand_builtin): Remove assert
17033 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
17035 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
17037 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
17038 loop_vinfo argument and use of dependence distance vectors.
17039 Check instead whether the two references differ only in their
17040 initial value and assume that they have the same alignment if the
17041 difference is a multiple of the vector alignment.
17042 (vect_analyze_data_refs_alignment): Update call accordingly.
17044 2017-05-31 Martin Liska <mliska@suse.cz>
17047 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
17049 2017-05-31 Bin Cheng <bin.cheng@arm.com>
17051 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
17052 (create_intersect_range_checks): Move from ...
17053 * tree-data-ref.c (create_intersect_range_checks_index)
17054 (create_intersect_range_checks): ... to here.
17055 (create_runtime_alias_checks): New function factored from ...
17056 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
17057 here. Call above function.
17058 * tree-data-ref.h (create_runtime_alias_checks): New function.
17060 2017-05-31 Bin Cheng <bin.cheng@arm.com>
17062 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
17063 segment length for dr_b and compute it in wide_int.
17065 2017-05-31 Richard Biener <rguenther@suse.de>
17067 PR tree-optimization/80906
17068 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
17069 and pass through iv_map.
17070 (copy_bb_and_scalar_dependences): Adjust.
17071 (translate_pending_phi_nodes): Likewise.
17072 (copy_loop_close_phi_args): Handle code-generating IVs instead
17075 2017-05-30 David Malcolm <dmalcolm@redhat.com>
17077 * diagnostic-color.c (color_dict): Add "type-diff".
17078 (parse_gcc_colors): Update comment.
17079 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
17080 -fdiagnostics-show-template-tree and -fno-elide-type.
17081 (GCC_COLORS): Add type-diff to example.
17083 (-fdiagnostics-show-template-tree): New.
17084 (-fno-elide-type): New.
17085 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
17086 the pp_format_decoder callback. Call any m_format_postprocessor's
17088 (pretty_printer::pretty_printer): Initialize
17089 m_format_postprocessor.
17090 (pretty_printer::~pretty_printer): Delete any
17091 m_format_postprocessor.
17092 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
17093 (class format_postprocessor): New class.
17094 (struct pretty_printer::format_decoder): Document the new parameters.
17095 (struct pretty_printer::m_format_postprocessor): New field.
17096 * tree-diagnostic.c (default_tree_printer): Update for new
17097 bool and const char ** params.
17098 * tree-diagnostic.h (default_tree_printer): Likewise.
17100 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
17102 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
17103 (lwa_operand): Delete rs6000_gen_cell_microcode test.
17104 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17105 rs6000_gen_cell_microcode code.
17106 (rs6000_final_prescan_insn): Delete.
17107 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
17108 "warn-cell-microcode" entries.
17109 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
17110 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
17111 throughout. Change cc_reg_not_micro_cr0_operand to
17112 cc_reg_not_cr0_operand throughout.
17113 (*extendhi<mode>2_noload): Delete.
17114 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
17115 (mwarn-cell-microcode): Delete.
17116 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
17117 -mgen-cell-microcode and -mwarn-cell-microcode.
17119 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
17122 * config/i386/constraints.md (Yd): New constraint.
17124 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
17125 and (?Yd, r) alternatives. Update insn attributes.
17126 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
17127 and (?*Yd, r) alternatives. Update insn attributes.
17128 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
17130 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
17132 * gimplify.c (gimplify_modify_expr): Don't create a
17133 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
17136 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
17138 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
17140 2017-05-30 Richard Biener <rguenther@suse.de>
17142 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
17143 and reduc_def fields.
17144 (STMT_VINFO_REDUC_TYPE): New define.
17145 (STMT_VINFO_REDUC_DEF): Likewise.
17146 (vect_force_simple_reduction): Adjust prototype.
17147 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
17148 (vect_is_simple_reduction): Remove check_reduction argument.
17149 (vect_force_simple_reduction): Adjust and set
17150 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
17151 (vectorizable_reduction): Do not re-do reduction analysis
17152 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
17153 * tree-parloops.c (gather_scalar_reductions): Adjust.
17155 2017-05-30 Richard Biener <rguenther@suse.de>
17157 PR middle-end/80901
17158 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
17161 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
17163 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
17164 Introduce unknown_misalignment parameter and remove vf.
17165 (vect_peeling_hash_get_lowest_cost):
17166 Pass unknown_misalignment parameter.
17167 (vect_enhance_data_refs_alignment):
17168 Fix unsupportable data ref treatment.
17170 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17172 * tree-vect-data-refs.c (vect_get_data_access_cost):
17173 Workaround for SLP handling.
17174 (vect_enhance_data_refs_alignment):
17175 Compute costs for doing no peeling at all, compare to the best
17176 peeling costs so far and avoid peeling if cheaper.
17178 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17180 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
17181 Return peeling info and set costs to zero for unlimited cost
17183 (vect_enhance_data_refs_alignment): Also inspect all datarefs
17184 with unknown misalignment. Compute and costs for unknown
17185 misalignment, compare them to the costs for known misalignment
17186 and choose the cheapest for peeling.
17188 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17190 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
17191 (vect_get_peeling_costs_all_drs): Create function.
17192 (vect_peeling_hash_get_lowest_cost):
17193 Use vect_get_peeling_costs_all_drs.
17194 (vect_peeling_supportable): Create function.
17195 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
17197 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17199 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
17200 DR_HAS_NEGATIVE_STEP.
17201 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
17202 (vect_enhance_data_refs_alignment): Use.
17203 (vect_duplicate_ssa_name_ptr_info): Use.
17204 * tree-vectorizer.h (dr_misalignment): Use.
17205 (known_alignment_for_access_p): Use.
17207 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17210 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
17212 (has_section_name): New function.
17214 2017-05-30 Martin Liska <mliska@suse.cz>
17217 * auto-profile.c (get_function_decl_from_block): Fix
17220 2017-05-30 Richard Biener <rguenther@suse.de>
17222 PR middle-end/80876
17223 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
17225 2017-05-30 Martin Liska <mliska@suse.cz>
17227 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
17228 * dumpfile.h (struct dump_file_info): Remove ctors.
17230 2017-05-30 Martin Liska <mliska@suse.cz>
17232 * predict.def: Fix GNU coding style.
17234 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
17236 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17237 Mark 'to' argument with ATTRIBUTE_UNUSED.
17239 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
17241 * config/xtensa/xtensa.c (xtensa_emit_call): Use
17242 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
17243 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
17246 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
17248 * doc/install.texi (Options specification): Restore entry of
17249 --enable-sjlj-exceptions.
17251 2017-05-27 Michael Eager <eager@eagercon.com>
17254 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
17256 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
17258 * config/microblaze/microblaze.h
17259 (FIXED_REGISTERS): Update in macro.
17260 (CALL_USED_REGISTERS): Update in macro.
17262 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
17264 * doc/install.texi: Add links to macOS binary distributions.
17266 2017-05-27 Jakub Jelinek <jakub@redhat.com>
17270 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17272 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
17274 2017-05-26 Martin Liska <mliska@suse.cz>
17276 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
17278 2017-05-26 Martin Liska <mliska@suse.cz>
17280 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
17281 always leading ';; '.
17282 (dump_bb_info): Likewise.
17283 (brief_dump_cfg): Likewise.
17284 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
17285 * dumpfile.c: Remove usage of TDF_VERBOSE.
17286 * dumpfile.h (enum dump_kind): Likewise.
17287 (dump_gimple_bb_header): Do not use TDF_COMMENT.
17288 * print-tree.c (debug_verbose): Remove.
17289 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
17290 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
17291 * tree-diagnostic.c (default_tree_printer): Replace
17292 TDF_DIAGNOSTIC with TDF_SLIM.
17294 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17296 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
17297 in parameter loop, rather than loop_vinfo.
17298 (create_intersect_range_checks): Ditto.
17299 (vect_create_cond_for_alias_checks): Update call to above functions.
17301 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17303 PR tree-optimization/80815
17304 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
17305 for merging runtime alias checks. Handle negative DR_STEPs.
17307 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17309 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
17311 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
17312 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
17313 out code pruning runtime alias checks.
17314 * tree-data-ref.c (prune_runtime_alias_test_list): New function
17315 factored out from above.
17316 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17318 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17320 (prune_runtime_alias_test_list): New decalaration.
17322 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17324 * tree-vect-data-refs.c (compare_tree): Rename and move ...
17325 * tree-data-ref.c (data_ref_compare_tree): ... to here.
17326 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
17327 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
17328 (operator==, comp_dr_with_seg_len_pair): Ditto.
17329 (vect_prune_runtime_alias_test_list): Ditto.
17331 2017-05-26 Martin Liska <mliska@suse.cz>
17334 * params.def: Bound partial-inlining-entry-probability param.
17336 2017-05-26 Marek Polacek <polacek@redhat.com>
17339 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
17342 2017-05-26 Richard Biener <rguenther@suse.de>
17344 PR tree-optimization/80842
17345 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
17348 2017-05-26 Richard Biener <rguenther@suse.de>
17350 PR tree-optimization/80844
17351 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
17353 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
17355 * doc/md.texi (Machine Constraints): Update x86 family
17356 machine constraints section to match 'config/i386/constraints.md'.
17358 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
17360 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
17362 2017-05-25 Nathan Sidwell <nathan@acm.org>
17364 * doc/invoke.texi (--enable-languages): Update documentation.
17366 2017-05-25 Martin Liska <mliska@suse.cz>
17368 * dumpfile.c: Add TDF_FOLDING.
17369 * dumpfile.h (enum dump_kind): Likewise.
17370 * genmatch.c (dt_simplify::gen_1): Use it.
17372 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17374 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
17376 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17378 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
17379 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
17381 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17383 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
17384 * match.pd (X == C): Rewrite it here.
17385 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
17386 with_certain_nonzero_bits2): New predicates.
17387 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
17389 2017-05-24 Nathan Sidwell <nathan@acm.org>
17391 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
17394 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
17397 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17399 * config/powerpcspe: New port. Files are copied from the rs6000
17400 port, with "rs6000" in filenames replaced by "powerpcspe".
17402 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
17404 PR rtl-optimization/80754
17405 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
17407 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
17409 * config/sparc/sparc.md (length): Return the correct value for -mflat
17410 sibcalls to match output_sibcall.
17412 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17416 * config/rs6000/rs6000.c (struct machine_function): Add new field
17418 (rs6000_get_separate_components): Init that field, use it.
17419 (rs6000_components_for_bb): Use the field.
17421 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
17423 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
17425 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
17427 PR middle-end/80823
17428 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
17430 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17433 * config/s390/s390.c (s390_check_qrst_address): Check incoming
17434 address against address_operand predicate.
17435 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
17437 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
17439 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
17440 parameters passed indirectly.
17442 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
17444 * config/i386/i386.md (*movdi_internal): Remove SSE4
17445 alternative 18 (?r, *v). Update insn attributes.
17446 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
17447 Update insn attributes.
17448 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
17449 Update insn attributes.
17450 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
17451 alternative 1 (r, v). Remove isa attribute.
17452 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
17453 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
17454 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
17456 2017-05-23 Tom de Vries <tom@codesourcery.com>
17458 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
17461 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17463 * cgraphunit.c (symbol_table::process_new_functions): Update.
17464 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
17465 (inline_generate_summary): Rename to ...
17466 (ipa_fn_summary_generate): ... this one.
17467 (inline_read_summary): Rename to ...
17468 (ipa_fn_summary_read): ... this one.
17469 (inline_write_summary): Rename to ...
17470 (ipa_fn_summary_write): ... this one.
17471 (inline_free_summary): Rename to ...
17472 (ipa_free_fn_summary): ... this one.
17473 (pass_data_local_fn_summary, pass_local_fn_summary,
17474 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
17475 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
17476 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
17477 make_pass_ipa_fn_summary): New.
17478 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
17479 inline_write_summary, inline_free_summary): Remove.
17480 (ipa_free_fn_summary) : New.
17481 * ipa-inline.c (ipa_inline): Update.
17482 (pass_ipa_inline): Do not generate summaries.
17483 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
17485 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
17486 and add pass_ipa_fn_summary.
17487 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
17489 (make_pass_inline_parameters): Remove.
17491 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
17493 * omp-low.c (struct omp_context): Remove "default_kind" member.
17496 * omp-offload.c (execute_oacc_device_lower): Remove the
17497 parallelism dimensions function attributes for unparallelized
17498 OpenACC kernels constructs.
17500 2017-05-23 Martin Liska <mliska@suse.cz>
17502 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
17504 (cgraph_edge::make_speculative): Likewise.
17505 (cgraph_edge::resolve_speculation): Likewise.
17506 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
17507 (cgraph_node::dump): Likewise.
17508 * cgraph.h: Likewise.
17509 * cgraphunit.c (analyze_functions): Likewise.
17510 (symbol_table::compile): Likewise.
17511 * ipa-cp.c (print_all_lattices): Likewise.
17512 (determine_versionability): Likewise.
17513 (initialize_node_lattices): Likewise.
17514 (ipcp_verify_propagated_values): Likewise.
17515 (estimate_local_effects): Likewise.
17516 (update_profiling_info): Likewise.
17517 (create_specialized_node): Likewise.
17518 (perhaps_add_new_callers): Likewise.
17519 (decide_about_value): Likewise.
17520 (decide_whether_version_node): Likewise.
17521 (identify_dead_nodes): Likewise.
17522 (ipcp_store_bits_results): Likewise.
17523 * ipa-devirt.c (dump_targets): Likewise.
17524 (ipa_devirt): Likewise.
17525 * ipa-icf.c (sem_item::dump): Likewise.
17526 (sem_function::equals): Likewise.
17527 (sem_variable::equals): Likewise.
17528 (sem_item_optimizer::read_section): Likewise.
17529 (sem_item_optimizer::execute): Likewise.
17530 (congruence_class::dump): Likewise.
17531 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
17532 (dump_inline_summary): Likewise.
17533 (estimate_node_size_and_time): Likewise.
17534 (inline_analyze_function): Likewise.
17535 * ipa-inline-transform.c (inline_call): Likewise.
17536 * ipa-inline.c (report_inline_failed_reason): Likewise.
17537 (want_early_inline_function_p): Likewise.
17538 (edge_badness): Likewise.
17539 (update_edge_key): Likewise.
17540 (inline_small_functions): Likewise.
17541 * ipa-profile.c (ipa_profile): Likewise.
17542 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
17543 (ipa_make_edge_direct_to_target): Likewise.
17544 (remove_described_reference): Likewise.
17545 (ipa_impossible_devirt_target): Likewise.
17546 (propagate_controlled_uses): Likewise.
17547 (ipa_print_node_params): Likewise.
17548 (ipcp_transform_function): Likewise.
17549 * ipa-pure-const.c (pure_const_read_summary): Likewise.
17550 (propagate_pure_const): Likewise.
17551 * ipa-reference.c (generate_summary): Likewise.
17552 (read_write_all_from_decl): Likewise.
17553 (propagate): Likewise.
17554 (ipa_reference_read_optimization_summary): Likewise.
17555 * ipa-utils.c (ipa_merge_profiles): Likewise.
17556 * ipa.c (walk_polymorphic_call_targets): Likewise.
17557 (symbol_table::remove_unreachable_nodes): Likewise.
17558 (ipa_single_use): Likewise.
17559 * passes.c (execute_todo): Likewise.
17560 * predict.c (drop_profile): Likewise.
17561 * symtab.c (symtab_node::get_dump_name): New function.
17562 (symtab_node::dump_name): Likewise.
17563 (symtab_node::dump_asm_name): Likewise.
17564 (symtab_node::dump_references): Likewise.
17565 (symtab_node::dump_referring): Likewise.
17566 (symtab_node::dump_base): Likewise.
17567 (symtab_node::debug_symtab): Likewise.
17568 * tree-sra.c (convert_callers_for_node): Likewise.
17569 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
17570 * value-prof.c (init_node_map): Likewise.
17572 2017-05-23 Martin Liska <mliska@suse.cz>
17574 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
17575 and symtab_node::debug_symtab to symbol_table::debug.
17576 * cgraphunit.c (analyze_functions): Use the renamed function.
17577 (symbol_table::compile): Likewise.
17578 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
17579 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
17580 * passes.c (execute_todo): Likewise.
17581 * symtab.c (symbol_table::dump): New function.
17582 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
17584 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17586 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
17587 that nonconst implies exec.
17589 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17591 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17592 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17593 (inline_edge_summary_vec): Turn into ...
17594 (ipa_call_summaries): ... this one.
17595 (redirect_to_unreachable, edge_set_predicate,
17596 evaluate_properties_for_edge, inline_summary_alloc,
17597 reset_ipa_call_summary, reset_inline_summary,
17598 inline_summary_t::duplicate): Update.
17599 (inline_edge_duplication_hook): Turn to ...
17600 (ipa_call_summary_t::duplicate): ... this one.
17601 (inline_edge_removal_hook): Turn to ...
17602 (ipa_call_summary_t::remove): ... this one.
17603 (dump_inline_edge_summary): Turn to ...
17604 (dump_ipa_call_summary): ... this one.
17605 (estimate_function_body_sizes): Update.
17606 (inline_update_callee_summaries): Update.
17607 (remap_edge_change_prob): Update.
17608 (remap_edge_summaries): Update.
17609 (inline_merge_summary): Update.
17610 (do_estimate_edge_time): Update.
17611 (inline_generate_summary): Update.
17612 (inline_read_section): Update.
17613 (inline_read_summary): Update.
17614 (inline_free_summary): Update.
17615 * ipa-inline.c (can_inline_edge_p): Update.
17616 (compute_inlined_call_time): Update.
17617 (want_inline_small_function_p): Update.
17618 (edge_badness): Update.
17619 (early_inliner): Update.
17620 * ipa-inline.h (inline_edge_summary): Turn to ...
17621 (ipa_call_summary): ... this one.
17622 (ipa_call_summary_t): New class.
17623 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17624 (ipa_call_summaries): New.
17625 (inline_edge_summary): Remove.
17626 (estimate_edge_growth): Update.
17627 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17628 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17629 * ipa-split.c (execute_split_functions): Update.
17630 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17632 2017-05-23 Tom de Vries <tom@codesourcery.com>
17634 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17635 attributes): Document rdrand effective target.
17637 2017-05-23 Tom de Vries <tom@codesourcery.com>
17639 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17640 attributes): Sort alphabetically.
17642 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
17644 * config/avr/genmultilib.awk: Use gsub instead of gensub.
17646 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
17649 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
17650 V2DF/V2DI splat into two separate patterns, one that handles
17651 registers, and the other that only handles memory. Drop support
17652 for splatting from a GPR on ISA 2.07 and then splitting the
17653 splat into direct move and splat.
17654 (vsx_splat_<mode>_reg): Likewise.
17655 (vsx_splat_<mode>_mem): Likewise.
17657 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
17659 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
17661 2017-05-22 Jakub Jelinek <jakub@redhat.com>
17663 PR middle-end/80809
17664 * omp-low.c (finish_taskreg_remap): New function.
17665 (finish_taskreg_scan): If unit size of ctx->record_type
17666 is non-constant, unshare the size expression and replace
17667 decls in it with possible outer var refs.
17669 PR middle-end/80809
17670 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
17671 GOVD_SHARED rather than GOVD_PRIVATE with it.
17672 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
17673 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
17675 PR middle-end/80853
17676 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
17677 as last argument to build_outer_var_ref for pointer bases of array
17678 section reductions.
17680 2017-05-19 Martin Sebor <msebor@redhat.com>
17682 * print-tree.c (print_node): Print DECL_READ_P flag.
17684 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17686 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
17687 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
17688 * cgraph.c: Likewise.
17689 * cgraphunit.c: Likewise.
17690 * gengtype.c: Likewise.
17691 * ipa-cp.c: Likewise.
17692 * ipa-devirt.c: Likewise.
17693 * ipa-icf.c: Likewise.
17694 * ipa-predicate.c: Likewise.
17695 * ipa-profile.c: Likewise.
17696 * ipa-prop.c: Likewise.
17697 * ipa-split.c: Likewise.
17699 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
17700 edge_predicate_pool, dump_inline_hints,
17701 inline_summary::account_size_time, redirect_to_unreachable,
17702 edge_set_predicate, set_hint_predicate,
17703 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
17704 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
17705 inline_summary_t::remove, remap_hint_predicate_after_duplication,
17706 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
17707 ipa_call_summary_t::remove, initialize_growth_caches,
17708 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
17709 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
17710 mark_modified, unmodified_parm_1, unmodified_parm,
17711 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
17712 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
17713 compute_bb_predicates, will_be_nonconstant_expr_predicate,
17714 will_be_nonconstant_predicate, record_modified_bb_info,
17715 get_minimal_bb, record_modified, param_change_prob,
17716 phi_result_unknown_predicate, predicate_for_phi_result,
17717 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
17718 estimate_function_body_sizes, compute_inline_parameters,
17719 compute_inline_parameters_for_curren, pass_data_inline_parameters,
17720 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
17721 inline_update_callee_summaries, remap_edge_change_prob,
17722 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
17723 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
17724 inline_analyze_function, inline_summary_t::insert,
17725 inline_generate_summary, read_ipa_call_summary, inline_read_section,
17726 inline_read_summary, write_ipa_call_summary, inline_write_summary,
17727 inline_free_summary): Move to ipa-fnsummary.h
17728 (predicate_t): Remove.
17729 * ipa-fnsummary.c: New file.
17730 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
17731 (enum inline_hints_vals, inline_hints, agg_position_info,
17732 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
17733 inline_summaries, ipa_call_summary, ipa_call_summary_t,
17734 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
17735 dump_inline_summary, dump_inline_hints, inline_generate_summary,
17736 inline_read_summary, inline_write_summary, inline_free_summary,
17737 inline_analyze_function, initialize_inline_failed,
17738 inline_merge_summary, inline_update_overall_summary,
17739 compute_inline_parameters): Move to ipa-fnsummary.h
17740 * ipa-fnsummary.h: New file.
17741 * ipa-inline-transform.h: Include ipa-inline.h.
17742 * ipa-inline.c: LIkewise.
17744 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17746 * ipa-inline.c (edge_badness): Use inlined_time instead of
17747 inline_summaries->get.
17749 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17751 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
17753 2017-05-22 Nathan Sidwell <nathan@acm.org>
17755 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
17756 (fdump-lang): Document 'raw' option.
17757 * dumpfile.h (TDI_tu): Delete.
17758 * dumpfile.c (dump_files): Remove translation-unit.
17759 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
17761 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
17763 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
17764 command option from $(AWK) call.
17765 * config/avr/genmultilib.awk: Simplify and rewrite so that it
17766 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
17767 [FORMAT]: Remove handling of variable.
17768 * config/avr/t-multilib: Regenerate.
17770 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17772 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
17774 (dump_inline_summary): Do not print self_time.
17775 (estimate_function_body_sizes): Do not set self_time.
17776 (compute_inline_parameters): Likewise.
17777 (inline_read_section, inline_write_summary): Do not stream self_time.
17778 * ipa-inline.h (inline_summary): Drop self_time.
17780 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17782 * ipa-inline-analysis.c (account_size_time): Rename to ...
17783 (inline_summary::account_size_time): ... this one.
17784 (reset_ipa_call_summary): Turn to ...
17785 (ipa_call_summary::reset): ... this one.
17786 (reset_inline_summary): Turn to ...
17787 (inline_summary::reset): ... this one.
17788 (inline_summary_t::remove): Update.
17789 (inline_summary_t::duplicate): Update.
17790 (ipa_call_summary_t::remove): Update.
17791 (dump_inline_summary): Update.
17792 (estimate_function_body_sizes): Update.
17793 (compute_inline_parameters): Update.
17794 (estimate_node_size_and_time): Update.
17795 (inline_merge_summary): Update.
17796 (inline_update_overall_summary): Update.
17797 (inline_read_section): Update.
17798 (inline_write_summary): Update.
17799 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
17800 add account_size_time and reset member functions.
17801 (ipa_call_summary): Add reset function.
17802 * ipa-predicate.h (predicate::operator &): Constify.
17804 2017-05-22 Richard Biener <rguenther@suse.de>
17806 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
17808 2017-05-19 Jason Merrill <jason@redhat.com>
17810 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
17812 2017-05-19 Marek Polacek <polacek@redhat.com>
17815 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
17816 TYPE_OVERFLOW_WRAPS checks.
17818 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
17820 * tree-core.h (enum omp_clause_default_kind): Add
17821 "OMP_CLAUSE_DEFAULT_PRESENT".
17822 * tree-pretty-print.c (dump_omp_clause): Handle it.
17823 * gimplify.c (enum gimplify_omp_var_data): Add
17824 "GOVD_MAP_FORCE_PRESENT".
17825 (gimplify_adjust_omp_clauses_1): Map it to
17826 "GOMP_MAP_FORCE_PRESENT".
17827 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
17829 * gimplify.c (oacc_default_clause): Clarify.
17831 2017-05-19 Nathan Sidwell <nathan@acm.org>
17833 LANG_HOOK_REGISTER_DUMPS
17834 * toplev.c (general_init): Call register dump lang hook.
17835 * doc/invoke.texi: Document -fdump-lang option family.
17836 * dumpfile.c (dump_files): Remove class dump here.
17837 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
17838 * dumpfile.h (tree_dump_index): Remove TDI_class.
17839 * langhooks-def.h (lhd_register_dumps): Declare.
17840 (LANG_HOOKS_REGISTER_DUMPS): Define.
17841 (LANG_HOOKS_INITIALIZER): Add it.
17842 * langhooks.c (lhd_register_dumps): Define.
17843 * langhooks.h (struct lang_hooks): Add register_dumps.
17845 2017-05-19 Nathan Sidwell <nathan@acm.org>
17847 * context.h (context::set_passes): New.
17848 * context.c (context::context): Do not create pass manager.
17849 * toplev.c (general_init): Create pass manager here.
17851 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
17853 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
17854 use this splitter if two add or or instructions would also work for
17855 the constant we want to generate.
17857 2017-05-19 Richard Biener <rguenther@suse.de>
17860 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
17861 predicate evaluation.
17863 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17865 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
17867 * ipa-inline.c (want_inline_small_function_p): Do not cast to
17870 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
17872 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17873 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17874 (inline_edge_summary_vec): Turn into ...
17875 (ipa_call_summaries): ... this one.
17876 (redirect_to_unreachable, edge_set_predicate,
17877 evaluate_properties_for_edge, inline_summary_alloc,
17878 reset_ipa_call_summary, reset_inline_summary,
17879 inline_summary_t::duplicate): Update.
17880 (inline_edge_duplication_hook): Turn to ...
17881 (ipa_call_summary_t::duplicate): ... this one.
17882 (inline_edge_removal_hook): Turn to ...
17883 (ipa_call_summary_t::remove): ... this one.
17884 (dump_inline_edge_summary): Turn to ...
17885 (dump_ipa_call_summary): ... this one.
17886 (estimate_function_body_sizes): Update.
17887 (inline_update_callee_summaries): Update.
17888 (remap_edge_change_prob): Update.
17889 (remap_edge_summaries): Update.
17890 (inline_merge_summary): Update.
17891 (do_estimate_edge_time): Update.
17892 (inline_generate_summary): Update.
17893 (inline_read_section): Update.
17894 (inline_read_summary): Update.
17895 (inline_free_summary): Update.
17896 * ipa-inline.c (can_inline_edge_p): Update.
17897 (compute_inlined_call_time): Update.
17898 (want_inline_small_function_p): Update.
17899 (edge_badness): Update.
17900 (early_inliner): Update.
17901 * ipa-inline.h (inline_edge_summary): Turn to ...
17902 (ipa_call_summary): ... this one.
17903 (ipa_call_summary_t): New class.
17904 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17905 (ipa_call_summaries): New.
17906 (inline_edge_summary): Remove.
17907 (estimate_edge_growth): Update.
17908 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17909 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17910 * ipa-split.c (execute_split_functions): Update.
17911 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17913 2017-05-19 Richard Biener <rguenther@suse.de>
17915 PR middle-end/80764
17916 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
17918 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
17920 * config/rs6000/rs6000.c (struct machine_function): Add field
17921 fpr_is_wrapped_separately.
17922 (rs6000_get_separate_components): Use 64 components. Handle the
17923 new FPR components.
17924 (rs6000_components_for_bb): Handle the FPR components.
17925 (rs6000_emit_prologue_components): Handle the FPR components.
17926 (rs6000_emit_epilogue_components): Handle the FPR components.
17927 (rs6000_set_handled_components): Handle the FPR components.
17928 (rs6000_emit_prologue): Don't output prologue code for those FPRs
17929 that are already separately shrink-wrapped.
17930 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
17931 that are already separately shrink-wrapped.
17933 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17936 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
17939 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
17940 (define_peephole2 for Altivec d-form load): Add peepholes to catch
17941 cases where the register allocator uses a move and an offsettable
17942 memory operation to/from a FPR register on ISA 2.06/2.07.
17943 (define_peephole2 for Altivec d-form store): Likewise.
17945 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
17948 * config/i386/mmx.md (*mov<mode>_internal): Enable
17949 alternatives 11, 12, 13 and 14 also for 32bit targets.
17950 Remove alternatives 15, 16, 17 and 18.
17951 * config/i386/sse.md (vec_concatv2di): Change
17952 alternative (!x, *y) to (x, ?!*Yn).
17954 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
17956 * dumpfile.h (enum dump_kind): Remove stray comma.
17958 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
17960 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
17961 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
17962 predicate::num_conditions
17963 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
17964 (CHANGED): turn into predicate::changed.
17965 (agg_position_info): Move to ipa-predicate.h
17966 (add_condition, predicate::add_clause, predicate::operator &=,
17967 predicate::or_with, predicate::evaluate, predicate::probability,
17968 dump_condition, dump_clause, predicate::dump,
17969 predicate::remap_after_duplication, predicate::remap_after_inlining,
17970 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
17971 (evaluate_conditions_for_known_args): Update.
17972 (set_cond_stmt_execution_predicate): Update.
17973 * ipa-inline.h: Include ipa-predicate.h
17974 (condition, inline_param_summary, conditions, agg_position_info,
17975 predicate): Move to ipa-predicate.h
17976 * ipa-predicate.c: New file.
17977 * ipa-predicate.h: New file.
17979 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
17981 * final.c (leaf_function_p): Check we are not in a sequence.
17983 2017-05-18 Martin Liska <mliska@suse.cz>
17985 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
17986 * dumpfile.c (dump_register): Use new enum dump_kind.
17987 (get_dump_file_name): Likewise.
17988 (dump_enable_all): Likewise.
17989 (dump_switch_p_1): Likewise.
17990 (enable_rtl_dump_file): Remove usage of TDF_RTL.
17991 * dumpfile.h (enum dump_kind): New enum type.
17992 (struct dump_file_info): Create constructor and
17993 format fields and comments.
17994 * passes.c (pass_manager::register_one_dump_file):
17996 * statistics.c (statistics_early_init): Likewise.
17997 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
17998 TDF_TREE with TDF_SLIM.
17999 (gather_memory_references_ref): Likewise.
18001 2017-05-18 Martin Liska <mliska@suse.cz>
18003 * vec.h (struct vnull): Use it.
18005 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
18007 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
18008 (true_predicate, false_predicate, true_predicate_p,
18009 false_predicate_p): Remove.
18010 (single_cond_predicate, not_inlined_predicate): Turn to member function
18012 (add_condition): Update.
18013 (add_clause): Turn to...
18014 (predicate::add_clause): ... this one; update; allow passing NULL
18016 (and_predicates): Turn to ...
18017 (predicate::operator &=): ... this one.
18018 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
18019 (or_predicates): Turn to ...
18020 (predicate::or_with): ... this one.
18021 (evaluate_predicate): Turn to ...
18022 (predicate::evaluate): ... this one.
18023 (predicate_probability): Turn to ...
18024 (predicate::probability): ... this one.
18025 (dump_condition): Update.
18026 (dump_predicate): Turn to ...
18027 (predicate::dump): ... this one.
18028 (account_size_time): Update.
18029 (edge_set_predicate): Update.
18030 (set_hint_predicate): UPdate.
18031 (evaluate_conditions_for_known_args): Update.
18032 (evaluate_properties_for_edge): Update.
18033 (remap_predicate_after_duplication): Turn to...
18034 (predicate::remap_after_duplication): ... this one.
18035 (remap_hint_predicate_after_duplication): Update.
18036 (inline_summary_t::duplicate): UPdate.
18037 (dump_inline_edge_summary): Update.
18038 (dump_inline_summary): Update.
18039 (set_cond_stmt_execution_predicate): Update.
18040 (set_switch_stmt_execution_predicate): Update.
18041 (compute_bb_predicates): Update.
18042 (will_be_nonconstant_expr_predicate): Update.
18043 (will_be_nonconstant_predicate): Update.
18044 (phi_result_unknown_predicate): Update.
18045 (predicate_for_phi_result): Update.
18046 (array_index_predicate): Update.
18047 (estimate_function_body_sizes): Update.
18048 (estimate_node_size_and_time): Update.
18049 (estimate_ipcp_clone_size_and_time): Update.
18050 (remap_predicate): Rename to ...
18051 (predicate::remap_after_inlining): ... this one.
18052 (remap_hint_predicate): Update.
18053 (inline_merge_summary): Update.
18054 (inline_update_overall_summary): Update.
18055 (estimate_size_after_inlining): Update.
18056 (read_predicate): Rename to ...
18057 (predicate::stream_in): ... this one.
18058 (read_inline_edge_summary): Update.
18059 (write_predicate): Rename to ...
18060 (predicate::stream_out): ... this one.
18061 (write_inline_edge_summary): Update.
18062 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
18063 (clause_t): Turn to uint32_t
18064 (predicate): Turn to class; implement constructor and operators
18066 (size_time_entry): Update.
18067 (inline_summary): Update.
18068 (inline_edge_summary): Update.
18070 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
18072 * fold-const.c (fold_binary_loc): Move transformation...
18073 * match.pd (C - X CMP X): ... here.
18075 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
18077 * config/sparc/sparc.c (sparc_option_override): Set function
18078 alignment for -mcpu=niagara7 to 64 to match the I$ line.
18079 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
18081 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
18083 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
18085 2017-05-18 Marek Polacek <polacek@redhat.com>
18088 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
18089 (pass_ubsan::execute): Call gimple_assign_single_p instead of
18090 gimple_assign_load_p.
18092 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
18094 PR middle-end/80692
18095 * real.c (do_compare): Give decimal_do_compare preference over
18096 comparing just the signs.
18098 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
18100 * doc/md.texi (Canonicalization of Instructions): Describe the
18101 canonical form of instructions that inherently set a condition
18104 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
18106 PR middle-end/80775
18107 * tree-cfg.c: Move deletion of unreachable case statements to after
18108 the merging of consecutive case labels.
18110 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
18112 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
18113 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
18114 restoring of callee-saved registers.
18116 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
18118 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
18119 * config/visium/visium.c (single_set_and_flags): Likewise.
18120 * config/visium/visium.md (Substitutions): Likewise.
18122 2017-05-17 Martin Liska <mliska@suse.cz>
18124 * cfg.c: Introduce dump_flags_t type and
18125 use it instead of int type.
18127 * cfghooks.c: Likewise.
18128 * cfghooks.h (struct cfg_hooks): Likewise.
18129 * cfgrtl.c: Likewise.
18130 * cfgrtl.h: Likewise.
18131 * cgraph.c (cgraph_node::get_body): Likewise.
18132 * coretypes.h: Likewise.
18133 * domwalk.c: Likewise.
18134 * domwalk.h: Likewise.
18135 * dumpfile.c (struct dump_option_value_info): Likewise.
18136 (dump_enable_all): Likewise.
18137 (dump_switch_p_1): Likewise.
18138 (opt_info_switch_p): Likewise.
18139 * dumpfile.h (enum tree_dump_index): Likewise.
18140 (struct dump_file_info): Likewise.
18141 * genemit.c: Likewise.
18142 * generic-match-head.c: Likewise.
18143 * gengtype.c (open_base_files): Likewise.
18144 * gimple-pretty-print.c: Likewise.
18145 * gimple-pretty-print.h: Likewise.
18146 * graph.c (print_graph_cfg): Likewise.
18147 * graphite-scop-detection.c (dot_all_sese): Likewise.
18148 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
18149 * loop-unroll.c (report_unroll): Likewise.
18150 * passes.c (pass_manager::register_one_dump_file): Likewise.
18151 * print-tree.c: Likewise.
18152 * statistics.c: Likewise.
18153 * tree-cfg.c: Likewise.
18154 * tree-cfg.h: Likewise.
18155 * tree-dfa.c: Likewise.
18156 * tree-dfa.h: Likewise.
18157 * tree-dump.c (dump_function): Likewise.
18158 * tree-dump.h (struct dump_info): Likewise.
18159 * tree-pretty-print.c: Likewise.
18160 * tree-pretty-print.h: Likewise.
18161 * tree-ssa-live.c: Likewise.
18162 * tree-ssa-live.h: Likewise.
18163 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
18164 * tree-vect-loop.c: Likewise.
18165 * tree-vect-slp.c: Likewise.
18167 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
18168 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18170 PR tree-optimization/80457
18171 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
18172 of all arguments to a statement as scalar_to_vec operations.
18173 (vectorizable_call): Adjust call to vect_model_simple_cost for
18175 (vectorizable_conversion): Likewise.
18176 (vectorizable_assignment): Likewise.
18177 (vectorizable_shift): Likewise.
18178 (vectorizable_operation): Likewise.
18179 (vectorizable_comparison): Likewise.
18180 (vect_is_simple_cond): Record the def types for operands.
18181 (vectorizable_condition): Likewise, call vect_model_simple_cost.
18182 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
18183 for statement argument count.
18185 2017-05-16 Carl Love <cel@us.ibm.com>
18187 * config/rs6000/rs6000-c: Add support for built-in functions
18188 vector unsigned long long vec_bperm (vector unsigned long long,
18189 vector unsigned char)
18190 vector signed long long vec_mule (vector signed int,
18192 vector unsigned long long vec_mule (vector unsigned int,
18193 vector unsigned int)
18194 vector signed long long vec_mulo (vector signed int,
18196 vector unsigned long long vec_mulo (vector unsigned int,
18197 vector unsigned int)
18198 vector signed char vec_sldw (vector signed char,
18199 vector signed char,
18201 vector unsigned char vec_sldw (vector unsigned char,
18202 vector unsigned char,
18204 vector signed short vec_sldw (vector signed short,
18205 vector signed short,
18207 vector unsigned short vec_sldw (vector unsigned short,
18208 vector unsigned short,
18210 vector signed int vec_sldw (vector signed int,
18213 vector unsigned int vec_sldw (vector unsigned int,
18214 vector unsigned int,
18216 vector signed long long vec_sldw (vector signed long long,
18217 vector signed long long,
18219 vector unsigned long long vec_sldw (vector unsigned long long,
18220 vector unsigned long long,
18222 * config/rs6000/rs6000-c: Add support for built-in functions
18223 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
18224 * config/rs6000/altivec.h: Add defintion for vec_sldw.
18225 * doc/extend.texi: Update the built-in documentation for the
18226 new built-in functions.
18228 2017-05-16 Marek Polacek <polacek@redhat.com>
18232 * tree.c (save_expr): Don't fold the expression.
18234 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
18236 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
18237 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
18238 and (?*y,m). Update insn attributes.
18240 2017-05-16 Martin Liska <mliska@suse.cz>
18242 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
18243 flags argument of print_gimple_stmt, print_gimple_expr,
18244 print_generic_stmt and print_generic_expr.
18245 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
18246 * coretypes.h: Likewise.
18247 * except.c (dump_eh_tree): Likewise.
18248 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
18249 * gimple-pretty-print.h: Likewise.
18250 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
18251 (backprop::push_to_worklist): Likewise.
18252 (backprop::pop_from_worklist): Likewise.
18253 (backprop::process_use): Likewise.
18254 (backprop::intersect_uses): Likewise.
18255 (note_replacement): Likewise.
18256 * gimple-ssa-store-merging.c
18257 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
18258 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
18259 (pass_store_merging::execute): Likewise.
18260 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
18261 (ssa_base_cand_dump_callback): Likewise.
18262 (dump_incr_vec): Likewise.
18263 (replace_refs): Likewise.
18264 (replace_mult_candidate): Likewise.
18265 (create_add_on_incoming_edge): Likewise.
18266 (create_phi_basis): Likewise.
18267 (insert_initializers): Likewise.
18268 (all_phi_incrs_profitable): Likewise.
18269 (introduce_cast_before_cand): Likewise.
18270 (replace_one_candidate): Likewise.
18271 * gimplify.c (gimplify_expr): Likewise.
18272 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
18273 (set_rename): Likewise.
18274 (rename_uses): Likewise.
18275 (copy_loop_phi_nodes): Likewise.
18276 (add_close_phis_to_merge_points): Likewise.
18277 (copy_loop_close_phi_args): Likewise.
18278 (copy_cond_phi_args): Likewise.
18279 (graphite_copy_stmts_from_block): Likewise.
18280 (translate_pending_phi_nodes): Likewise.
18281 * graphite-poly.c (print_pdr): Likewise.
18282 (dump_gbb_cases): Likewise.
18283 (dump_gbb_conditions): Likewise.
18284 (print_scop_params): Likewise.
18285 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
18286 (build_cross_bb_scalars_use): Likewise.
18287 (gather_bbs::before_dom_children): Likewise.
18288 * hsa-dump.c (dump_hsa_immed): Likewise.
18289 * ipa-cp.c (print_ipcp_constant_value): Likewise.
18290 (get_replacement_map): Likewise.
18291 * ipa-inline-analysis.c (dump_condition): Likewise.
18292 (estimate_function_body_sizes): Likewise.
18293 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
18294 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
18295 * ipa-prop.c (ipa_dump_param): Likewise.
18296 (ipa_print_node_jump_functions_for_edge): Likewise.
18297 (ipa_modify_call_arguments): Likewise.
18298 (ipa_modify_expr): Likewise.
18299 (ipa_dump_param_adjustments): Likewise.
18300 (ipa_dump_agg_replacement_values): Likewise.
18301 (ipcp_modif_dom_walker::before_dom_children): Likewise.
18302 * ipa-pure-const.c (check_stmt): Likewise.
18303 (pass_nothrow::execute): Likewise.
18304 * ipa-split.c (execute_split_functions): Likewise.
18305 * omp-offload.c (dump_oacc_loop_part): Likewise.
18306 (dump_oacc_loop): Likewise.
18307 * trans-mem.c (tm_log_emit): Likewise.
18308 (tm_memopt_accumulate_memops): Likewise.
18309 (dump_tm_memopt_set): Likewise.
18310 (dump_tm_memopt_transform): Likewise.
18311 * tree-cfg.c (gimple_verify_flow_info): Likewise.
18312 (print_loop): Likewise.
18313 * tree-chkp-opt.c (chkp_print_addr): Likewise.
18314 (chkp_gather_checks_info): Likewise.
18315 (chkp_get_check_result): Likewise.
18316 (chkp_remove_check_if_pass): Likewise.
18317 (chkp_use_outer_bounds_if_possible): Likewise.
18318 (chkp_reduce_bounds_lifetime): Likewise.
18319 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
18320 (chkp_mark_completed_bounds): Likewise.
18321 (chkp_register_incomplete_bounds): Likewise.
18322 (chkp_mark_invalid_bounds): Likewise.
18323 (chkp_maybe_copy_and_register_bounds): Likewise.
18324 (chkp_build_returned_bound): Likewise.
18325 (chkp_get_bound_for_parm): Likewise.
18326 (chkp_build_bndldx): Likewise.
18327 (chkp_get_bounds_by_definition): Likewise.
18328 (chkp_generate_extern_var_bounds): Likewise.
18329 (chkp_get_bounds_for_decl_addr): Likewise.
18330 * tree-chrec.c (chrec_apply): Likewise.
18331 * tree-data-ref.c (dump_data_reference): Likewise.
18332 (dump_subscript): Likewise.
18333 (dump_data_dependence_relation): Likewise.
18334 (analyze_overlapping_iterations): Likewise.
18335 * tree-inline.c (expand_call_inline): Likewise.
18336 (tree_function_versioning): Likewise.
18337 * tree-into-ssa.c (dump_defs_stack): Likewise.
18338 (dump_currdefs): Likewise.
18339 (dump_names_replaced_by): Likewise.
18340 (dump_update_ssa): Likewise.
18341 (update_ssa): Likewise.
18342 * tree-object-size.c (pass_object_sizes::execute): Likewise.
18343 * tree-parloops.c (build_new_reduction): Likewise.
18344 (try_create_reduction_list): Likewise.
18345 (ref_conflicts_with_region): Likewise.
18346 (oacc_entry_exit_ok_1): Likewise.
18347 (oacc_entry_exit_single_gang): Likewise.
18348 * tree-pretty-print.h: Likewise.
18349 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
18350 (get_scalar_evolution): Likewise.
18351 (add_to_evolution): Likewise.
18352 (get_loop_exit_condition): Likewise.
18353 (analyze_evolution_in_loop): Likewise.
18354 (analyze_initial_condition): Likewise.
18355 (analyze_scalar_evolution): Likewise.
18356 (instantiate_scev): Likewise.
18357 (number_of_latch_executions): Likewise.
18358 (gather_chrec_stats): Likewise.
18359 (final_value_replacement_loop): Likewise.
18360 (scev_const_prop): Likewise.
18361 * tree-sra.c (dump_access): Likewise.
18362 (disqualify_candidate): Likewise.
18363 (create_access): Likewise.
18364 (reject): Likewise.
18365 (maybe_add_sra_candidate): Likewise.
18366 (create_access_replacement): Likewise.
18367 (analyze_access_subtree): Likewise.
18368 (analyze_all_variable_accesses): Likewise.
18369 (sra_modify_assign): Likewise.
18370 (initialize_constant_pool_replacements): Likewise.
18371 (find_param_candidates): Likewise.
18372 (decide_one_param_reduction): Likewise.
18373 (replace_removed_params_ssa_names): Likewise.
18374 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
18375 * tree-ssa-copy.c (dump_copy_of): Likewise.
18376 (copy_prop_visit_cond_stmt): Likewise.
18377 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
18378 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
18379 (record_equivalences_from_stmt): Likewise.
18380 * tree-ssa-dse.c (compute_trims): Likewise.
18381 (delete_dead_call): Likewise.
18382 (delete_dead_assignment): Likewise.
18383 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
18384 (forward_propagate_into_cond): Likewise.
18385 (pass_forwprop::execute): Likewise.
18386 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
18387 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
18389 (move_computations_worker): Likewise.
18390 (execute_sm): Likewise.
18391 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
18392 (remove_exits_and_undefined_stmts): Likewise.
18393 (remove_redundant_iv_tests): Likewise.
18394 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
18395 (adjust_iv_update_pos): Likewise.
18396 * tree-ssa-math-opts.c (bswap_replace): Likewise.
18397 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
18398 (value_replacement): Likewise.
18399 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
18400 * tree-ssa-pre.c (print_pre_expr): Likewise.
18401 (get_representative_for): Likewise.
18402 (create_expression_by_pieces): Likewise.
18403 (insert_into_preds_of_block): Likewise.
18404 (eliminate_insert): Likewise.
18405 (eliminate_dom_walker::before_dom_children): Likewise.
18406 (eliminate): Likewise.
18407 (remove_dead_inserted_code): Likewise.
18408 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
18409 * tree-ssa-reassoc.c (get_rank): Likewise.
18410 (eliminate_duplicate_pair): Likewise.
18411 (eliminate_plus_minus_pair): Likewise.
18412 (eliminate_not_pairs): Likewise.
18413 (undistribute_ops_list): Likewise.
18414 (eliminate_redundant_comparison): Likewise.
18415 (update_range_test): Likewise.
18416 (optimize_range_tests_var_bound): Likewise.
18417 (optimize_vec_cond_expr): Likewise.
18418 (rewrite_expr_tree): Likewise.
18419 (rewrite_expr_tree_parallel): Likewise.
18420 (linearize_expr): Likewise.
18421 (break_up_subtract): Likewise.
18422 (linearize_expr_tree): Likewise.
18423 (attempt_builtin_powi): Likewise.
18424 (attempt_builtin_copysign): Likewise.
18425 (transform_stmt_to_copy): Likewise.
18426 (transform_stmt_to_multiply): Likewise.
18427 (dump_ops_vector): Likewise.
18428 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
18429 (print_scc): Likewise.
18430 (set_ssa_val_to): Likewise.
18431 (visit_reference_op_store): Likewise.
18432 (visit_use): Likewise.
18433 (sccvn_dom_walker::before_dom_children): Likewise.
18434 (run_scc_vn): Likewise.
18435 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18437 (expr_hash_elt::print): Likewise.
18438 (const_and_copies::pop_to_marker): Likewise.
18439 (const_and_copies::record_const_or_copy_raw): Likewise.
18440 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
18441 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
18442 (dump_predicates): Likewise.
18443 (find_uninit_use): Likewise.
18444 (warn_uninitialized_phi): Likewise.
18445 (pass_late_warn_uninitialized::execute): Likewise.
18446 * tree-ssa.c (verify_vssa): Likewise.
18447 (verify_ssa): Likewise.
18448 (maybe_optimize_var): Likewise.
18449 * tree-vrp.c (dump_value_range): Likewise.
18450 (dump_all_value_ranges): Likewise.
18451 (dump_asserts_for): Likewise.
18452 (register_edge_assert_for_2): Likewise.
18453 (vrp_visit_cond_stmt): Likewise.
18454 (vrp_visit_switch_stmt): Likewise.
18455 (vrp_visit_stmt): Likewise.
18456 (vrp_visit_phi_node): Likewise.
18457 (simplify_cond_using_ranges_1): Likewise.
18458 (fold_predicate_in): Likewise.
18459 (evrp_dom_walker::before_dom_children): Likewise.
18460 (evrp_dom_walker::push_value_range): Likewise.
18461 (evrp_dom_walker::pop_value_range): Likewise.
18462 (execute_early_vrp): Likewise.
18464 2017-05-16 Richard Biener <rguenther@suse.de>
18466 * dwarf2out.c (loc_list_from_tree_1): Do not create
18467 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
18469 2017-05-16 Richard Biener <rguenther@suse.de>
18471 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
18473 (note_variable_value_in_expr): If we resolved the decl ref
18474 do not push to the stack.
18476 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
18478 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
18479 operations in fast-math mode.
18480 (vaddq_f16): Likewise.
18481 (vmul_f16): Likewise.
18482 (vmulq_f16): Likewise.
18483 (vsub_f16): Likewise.
18484 (vsubq_f16): Likewise.
18485 * config/arm/neon.md (add<mode>3): New.
18487 (fma:<VH:mode>3): New. Also remove outdated comment.
18490 2017-05-16 Martin Liska <mliska@suse.cz>
18494 * ipa-devirt.c (warn_types_mismatch): Fix typo.
18495 (odr_types_equivalent_p): Likewise.
18497 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
18499 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
18501 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
18504 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
18505 non-interunit SSE move alternatives with '?'.
18506 (zero-extendsidi peephole2): New peephole to skip intermediate
18507 general register in SSE zero-extend sequence.
18509 2017-05-15 Jeff Law <law@redhat.com>
18511 * reorg.c (relax_delay_slots): Create a new variable to hold
18512 the temporary target rather than clobbering TARGET_LABEL.
18514 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
18515 missing argument to extract_bit_field call.
18516 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
18518 2017-05-15 Martin Liska <mliska@suse.cz>
18521 * gcc.c (process_command): Do not allow empty argument of -o option.
18523 2017-05-15 Renlin Li <renlin.li@arm.com>
18525 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
18526 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
18527 * config/aarch64/constraints.md (Usf): Add long call check.
18528 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
18529 (call_value): Likewise.
18530 (sibcall): Likewise.
18531 (sibcall_value): Likewise.
18533 (call_value_insn): New.
18534 (sibcall_insn): Update rtx pattern.
18535 (sibcall_value_insn): Likewise.
18536 (call_internal): Remove.
18537 (call_value_internal): Likewise.
18538 (sibcall_internal): Likewise.
18539 (sibcall_value_internal): Likewise.
18540 (call_reg): Likewise.
18541 (call_symbol): Likewise.
18542 (call_value_reg): Likewise.
18543 (call_value_symbol): Likewise.
18545 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
18548 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
18550 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
18552 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
18553 compatible with CCGOCmode and with CCZmode.
18555 2017-05-14 Martin Sebor <msebor@redhat.com>
18557 PR middle-end/77671
18558 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
18559 (gimple_fold_builtin_snprintf): Same.
18560 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
18561 (gimple_fold_builtin_snprintf): Same.
18562 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
18563 of character types.
18564 (is_call_safe): New function.
18565 (try_substitute_return_value): Call it.
18566 (try_simplify_call): New function.
18567 (pass_sprintf_length::handle_gimple_call): Call it.
18569 2017-05-14 Martin Sebor <msebor@redhat.com>
18571 PR middle-end/80669
18572 * builtins.c (expand_builtin_stpncpy): Simplify.
18574 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
18576 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
18577 * config/i386/i386.h
18578 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
18579 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
18580 (struct machine_function): Add new members call_ms2sysv,
18581 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
18582 (struct machine_frame_state): New fields sp_realigned and
18583 sp_realigned_offset.
18584 * config/i386/i386.c
18585 (enum xlogue_stub): New enum.
18586 (enum xlogue_stub_sets): New enum.
18587 (class xlogue_layout): New class.
18588 (struct ix86_frame): New fields stack_realign_allocate_offset,
18589 stack_realign_offset and outlined_save_offset. Modify comments to
18590 detail stack layout when using out-of-line stubs.
18591 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
18592 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
18593 -mcall-ms2sysv-xlogues.
18594 (stub_managed_regs): New static variable.
18595 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
18596 registers managed by out-of-line stub.
18597 (disable_call_ms2sysv_xlogues): New function.
18598 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
18599 m->call_ms2sysv when appropriate and compute frame layout for
18601 (sp_valid_at, fp_valid_at): New inline functions.
18602 (choose_basereg): New function.
18603 (choose_baseaddr): Add align parameter, use choose_basereg and modify
18605 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
18606 Use align parameter of choose_baseaddr to generated aligned SSE movs
18608 (pro_epilogue_adjust_stack): Modify to track
18609 machine_frame_state::sp_realigned.
18610 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
18611 (ix86_nsaved_sseregs): Likewise.
18612 (ix86_emit_save_regs): Likewise.
18613 (ix86_emit_save_regs_using_mov): Likewise.
18614 (ix86_emit_save_sse_regs_using_mov): Likewise.
18615 (get_scratch_register_on_entry): Likewise.
18616 (gen_frame_set): New function.
18617 (gen_frame_load): Likewise.
18618 (gen_frame_store): Likewise.
18619 (emit_outlined_ms2sysv_save): Likewise.
18620 (emit_outlined_ms2sysv_restore): Likewise.
18621 (ix86_expand_prologue): Modify stack re-alignment code and call
18622 emit_outlined_ms2sysv_save when appropriate.
18623 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
18624 parameter rtx_insn *insn, which allows the function to be used to only
18625 generate the notes.
18626 (ix86_expand_epilogue): Modify validity checks of frame and stack
18627 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
18628 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
18629 * config/i386/predicates.md
18630 (save_multiple): New predicate.
18631 (restore_multiple): Likewise.
18632 * config/i386/sse.md
18633 (save_multiple<mode>): New pattern.
18634 (save_multiple_realign<mode>): Likewise.
18635 (restore_multiple<mode>): Likewise.
18636 (restore_multiple_and_return<mode>): Likewise.
18637 (restore_multiple_leave_return<mode>): Likewise.
18638 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
18640 2017-05-14 Julia Koval <julia.koval@intel.com>
18642 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
18643 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
18644 (__builtin_ia32_xsetbv): New builtins.
18645 * config/i386/i386.c (ix86_expand_special_args_builtin):
18647 (ix86_expand_builtin): Special expand for new intrinsics.
18648 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
18649 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
18650 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
18652 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18654 * cfganal.c (inverted_post_order_compute): Change argument type
18656 * cfganal.h (inverted_post_order_compute): Adjust prototype.
18657 * df-core.c (rest_of_handle_df_initialize): Adjust.
18658 (rest_of_handle_df_finish): Likewise.
18659 (df_analyze_1): Likewise.
18660 (df_analyze): Likewise.
18661 (loop_inverted_post_order_compute): Change argument to be a vec *.
18662 (df_analyze_loop): Adjust.
18663 (df_get_n_blocks): Likewise.
18664 (df_get_postorder): Likewise.
18665 * df.h (struct df_d): Change field to be a vec.
18666 * lcm.c (compute_laterin): Adjust.
18667 (compute_available): Likewise.
18668 * lra-lives.c (lra_create_live_ranges_1): Likewise.
18669 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
18670 * tree-ssa-pre.c (compute_antic): Likewise.
18672 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18674 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
18675 (depth_first_search::depth_first_search): Change structure init
18676 function to this constructor.
18677 (depth_first_search::add_bb): Rename function to this member.
18678 (depth_first_search::execute): Likewise.
18679 (flow_dfs_compute_reverse_finish): Adjust.
18681 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18683 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
18684 (longest_simple_path): Likewise.
18685 * shrink-wrap.c (spread_components): Likewise.
18686 (disqualify_problematic_components): Likewise.
18687 (emit_common_heads_for_components): Likewise.
18688 (emit_common_tails_for_components): Likewise.
18689 (insert_prologue_epilogue_for_components): Likewise.
18691 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18693 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
18696 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18698 * df-core.c (df_set_blocks): Start using auto_bitmap.
18699 (df_compact_blocks): Likewise.
18700 * df-problems.c (df_rd_confluence_n): Likewise.
18701 * df-scan.c (df_insn_rescan_all): Likewise.
18702 (df_process_deferred_rescans): Likewise.
18703 (df_update_entry_block_defs): Likewise.
18704 (df_update_exit_block_uses): Likewise.
18705 (df_entry_block_bitmap_verify): Likewise.
18706 (df_exit_block_bitmap_verify): Likewise.
18707 (df_scan_verify): Likewise.
18708 * lra-constraints.c (lra_constraints): Likewise.
18709 (undo_optional_reloads): Likewise.
18710 (lra_undo_inheritance): Likewise.
18711 * lra-remat.c (calculate_gen_cands): Likewise.
18712 (do_remat): Likewise.
18713 * lra-spills.c (assign_spill_hard_regs): Likewise.
18714 (spill_pseudos): Likewise.
18715 * tree-ssa-pre.c (bitmap_set_and): Likewise.
18716 (bitmap_set_subtract_values): Likewise.
18718 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18720 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
18721 management with auto_bitmap.
18722 (fix_inter_tick): Likewise.
18723 (fix_recovery_deps): Likewise.
18724 * ira.c (add_store_equivs): Likewise.
18725 (find_moveable_pseudos): Likewise.
18726 (split_live_ranges_for_shrink_wrap): Likewise.
18727 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
18728 (rtx_reuse_manager::seen_def_p): Likewise.
18729 (rtx_reuse_manager::set_seen_def): Likewise.
18730 * print-rtl.h (class rtx_reuse_manager): Likewise.
18732 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18734 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
18736 (migrate_btr_def): Likewise.
18737 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
18738 * df-core.c (loop_post_order_compute): Likewise.
18739 (loop_inverted_post_order_compute): Likewise.
18740 * hsa-common.h: Likewise.
18741 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
18742 * init-regs.c (initialize_uninitialized_regs): Likewise.
18743 * ipa-inline.c (resolve_noninline_speculation): Likewise.
18744 (inline_small_functions): Likewise.
18745 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
18746 * ira.c (combine_and_move_insns): Likewise.
18747 (build_insn_chain): Likewise.
18748 * loop-invariant.c (find_invariants): Likewise.
18749 * lower-subreg.c (propagate_pseudo_copies): Likewise.
18750 * predict.c (tree_predict_by_opcode): Likewise.
18751 (predict_paths_leading_to): Likewise.
18752 (predict_paths_leading_to_edge): Likewise.
18753 (estimate_loops_at_level): Likewise.
18754 (estimate_loops): Likewise.
18755 * shrink-wrap.c (try_shrink_wrapping): Likewise.
18756 (spread_components): Likewise.
18757 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
18758 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
18759 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
18760 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
18761 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
18762 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
18763 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
18764 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
18765 (mark_threaded_blocks): Likewise.
18766 (thread_through_all_blocks): Likewise.
18767 * tree-ssa.c (verify_ssa): Likewise.
18768 (execute_update_addresses_taken): Likewise.
18769 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
18771 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18773 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
18775 (post_order_compute): Likewise.
18776 (inverted_post_order_compute): Likewise.
18777 (pre_and_rev_post_order_compute_fn): Likewise.
18779 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18781 * genrecog.c (int_set::int_set): Explicitly construct our
18782 auto_vec base class.
18783 * vec.h (auto_vec::auto_vec): New constructor.
18785 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18787 * bitmap.h (class auto_bitmap): New constructor taking
18788 bitmap_obstack * argument.
18790 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18792 * bitmap.h (class auto_bitmap): Change type of m_bits to
18793 bitmap_head, and adjust ctor / dtor and member operators.
18795 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
18797 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
18798 when returned register mode doesn't match original mode.
18800 2017-05-12 Jeff Law <law@redhat.com>
18801 Jakub Jelinek <jakub@redhat.com>
18803 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
18804 we look for cc setter after the compare-elim changes.
18805 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
18806 within the vector to match what compare-elim now expects.
18807 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
18808 (xorsi3_flags, one_cmplsi2_flags): Likewise.
18810 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
18811 after the compare-elim changes.
18812 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
18813 the vector to match what compare-elim now expects.
18814 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
18815 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
18816 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
18817 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
18818 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
18820 * config/visium/visium.c (single_set_and_flags): Fix where
18821 we look for cc setter after the compare-elim changes.
18822 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
18823 with the vector to match what compare-elim now expects.
18824 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
18825 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
18826 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
18827 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
18828 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
18829 (neg<mode>2_insn_set_overflow): Likewise.
18831 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
18833 PR middle-end/79794
18834 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
18835 maybe_expand_insn call, set ops[0].target. If still set after call,
18836 set alt_rtl. Add extra arg to recursive calls.
18837 (extract_bit_field): Add alt_rtl argument. Pass to
18839 * expmed.h (extract_bit_field): Fix prototype.
18840 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
18841 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
18842 to extract_bit_field_calls.
18843 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
18844 Pass alt_rtl to extract_bit_field calls.
18845 * calls.c (store_unaligned_arguments_into_psuedos)
18846 load_register_parameters): Pass extra NULL to extract_bit_field calls.
18847 * optabs.c (maybe_legitimize_operand): Clear op->target when call
18849 * optabs.h (struct expand_operand): Add target bitfield.
18851 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18853 * compare-elim.c (try_eliminate_compare): Canonicalize
18854 operation with embedded compare to
18855 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
18856 (set (reg) (operation)].
18858 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
18860 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
18863 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
18864 cost of adding a carry flag for ADC instruction.
18865 [case MINUS]: Ignore the cost of subtracting a carry flag
18866 for SBB instruction.
18868 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
18870 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
18872 * config/rs6000/bmiintrin.h: New file.
18873 * config/rs6000/bmi2intrin.h: New file.
18874 * config/rs6000/x86intrin.h: New file.
18876 2017-05-12 Jeff Law <law@redhat.com>
18878 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
18881 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
18883 PR middle-end/80707
18884 * tree-cfg.c: Remove cfg edges of unreachable case statements.
18886 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18888 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18889 early expansion of vector divide builtins.
18890 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
18891 builtins identified as having unsigned arguments.
18893 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18895 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
18896 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
18897 expansion of vector logical operations (and, andc, or, xor,
18900 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18902 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
18903 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
18905 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
18907 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
18908 early GIMPLE expansion of vector multiplies.
18910 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18912 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
18913 TARGET_HAVE_MOVT conditional.
18914 (movt splitter): Likewise.
18916 2017-05-12 Richard Biener <rguenther@suse.de>
18918 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
18919 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
18920 Fold all stmts not inplace.
18922 2017-05-12 Richard Biener <rguenther@suse.de>
18924 PR tree-optimization/80713
18925 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
18926 inserted_exprs bit for not removed stmts.
18928 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
18930 PR middle-end/69921
18931 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
18932 parallelized" attribute for parallelized OpenACC kernels.
18933 * omp-offload.c (execute_oacc_device_lower): Use it.
18935 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
18936 Set "oacc kernels" attribute.
18937 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
18938 parameter. Adjust all users.
18939 (oacc_fn_attrib_kernels_p): Remove function.
18940 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
18941 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
18942 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
18943 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
18944 assert "oacc kernels" attribute is set.
18946 2017-05-11 Carl Love <cel@us.ibm.com>
18948 * config/rs6000/rs6000-c: Add support for built-in functions
18949 vector unsigned char vec_popcnt (vector signed char)
18950 vector unsigned char vec_popcnt (vector unsigned char)
18951 vector unsigned short vec_popcnt (vector signed short)
18952 vector unsigned short vec_popcnt (vector unsigned short)
18953 vector unsigned int vec_popcnt (vector signed int)
18954 vector unsigned int vec_popcnt (vector unsigned int)
18955 vector unsigned long long vec_popcnt (vector signed long long)
18956 vector unsigned long long vec_popcnt (vector unsigned long long)
18957 vector signed long long vec_slo (vector signed long long,
18958 vector signed char)
18959 vector signed long long vec_slo (vector signed long long,
18960 vector unsigned char)
18961 vector unsigned long long vec_slo (vector unsigned long long,
18962 vector signed char)
18963 vector unsigned long long vec_slo (vector unsigned long long,
18964 vector unsigned char)
18965 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
18966 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
18967 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
18968 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
18969 * doc/extend.texi: Update the built-in documentation file for the
18970 new built-in functions.
18972 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
18974 * attribs.h (sorted_attr_string): Move machine independent
18975 functions for target clone support from the i386 port to common
18976 code. Rename ix86_function_versions to common_function_versions.
18977 Rename make_name to make_unique_name.
18978 (common_function_versions): Likewise.
18979 (make_unique_name): Likewise.
18980 (make_dispatcher_decl): Likewise.
18981 (is_function_default_version): Likewise.
18982 * attribs.c (attr_strcmp): Likewise.
18983 (sorted_attr_string): Likewise.
18984 (common_function_versions): Likewise.
18985 (make_unique_name): Likewise.
18986 (make_dispatcher_decl): Likewise.
18987 (is_function_default_version): Likewise.
18988 * config/i386/i386.c (attr_strcmp): Likewise.
18989 (sorted_attr_string): Likewise.
18990 (ix86_function_versions): Likewise.
18991 (make_name): Likewise.
18992 (make_dispatcher_decl): Likewise.
18993 (is_function_default_version): Likewise.
18994 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
18996 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18999 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
19000 Account for direct move costs for vec_construct of integer
19003 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
19006 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
19007 (UNSPEC_STX_ATOMIC): Ditto.
19008 (loaddi_via_sse): New insn.
19009 (storedi_via_sse): Ditto.
19010 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
19011 Update corresponding peephole2 patterns.
19012 (atomic_storedi_fpu): Ditto.
19014 2017-05-11 Julia Koval <julia.koval@intel.com>
19016 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
19017 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
19019 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
19020 (__builtin_ia32_rsqrt14ss_mask): New builtins.
19021 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
19023 2017-05-11 Nathan Sidwell <nathan@acm.org>
19025 * graphite-poly.c: Include dumpfile.h.
19027 * dumpfle.h (dump_function): Declare here ...
19028 * tree-dump.h (dump_function): ... not here.
19029 * dumpfile.c: #include tree-cfg.h.
19030 (dump_function): Move here from ...
19031 * tree-dump.c (dump_function): ... here.
19032 * gimplify.c: #include splay-tree.h, not tree-dump.h.
19033 * graphite-poly.c: Don't include tree-dump.h.
19034 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
19035 * print-tree.c: Likewise.
19036 * stor-layout.c: Likewise.
19037 * tree-nested.c: Likewise.
19039 * dumpfile.c (dump_start): Use TDF_FLAGS.
19040 (dump_enable_all): Fix TDF_KIND check thinko.
19042 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
19044 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19045 array entries to represent two legal parameterizations of the
19046 overloaded __builtin_cmpb function, as represented by the
19047 P6_OV_BUILTIN_CMPB constant.
19048 (altivec_resolve_overloaded_builtin): Add special case handling
19049 for the __builtin_cmpb function, as represented by the
19050 P6_OV_BUILTIN_CMPB constant.
19051 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
19052 (BU_P6_64BIT_2): New macro.
19053 (BU_P6_OVERLOAD_2): New macro
19054 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
19055 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
19056 (CMPB): Add overload support to represent both 32-bit and 64-bit
19057 compare-bytes function.
19058 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
19059 support for TARGET_CMPB.
19060 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
19061 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19062 documentation of the __builtin_cmpb overloaded built-in function.
19064 2017-05-11 Richard Biener <rguenther@suse.de>
19066 PR tree-optimization/80705
19067 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
19068 bases are not vectorizable.
19070 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19072 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
19073 when counting register pressure.
19075 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19077 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
19078 (iv_ca_more_deps): Renamed to ...
19079 (iv_ca_compare_deps): ... this.
19080 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
19082 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19084 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
19086 (determine_group_iv_costs): ... here.
19087 (find_inv_vars_cb): Record inv var if it's not recorded before.
19089 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19091 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
19092 (get_shiftadd_cost): Ditto.
19094 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19096 * tree-ssa-address.c: Include header file.
19097 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
19099 (add_to_parts): Refactor.
19100 (addr_to_parts): New parameter. Update use of move_hint_to_base.
19101 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
19104 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19106 PR tree-optimization/53090
19107 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
19109 (extract_cond_operands): Detect condition with IV on both sides
19110 and return COMP_IV_EXPR_2.
19111 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
19112 (rewrite_use_compare): Simplify by removing call to function
19113 extract_cond_operands.
19115 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19117 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
19118 (extract_cond_operands): Detect condition comparing against non-
19119 invariant bound and return appropriate enum value.
19120 (find_interesting_uses_cond): Update use of extract_cond_operands.
19121 Handle its return value accordingly.
19122 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
19124 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19126 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
19127 nonlinear iv_use computation in loop invariant sensitive way.
19129 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19131 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
19132 (find_iv_candidates): Call relate_compare_use_with_all_cands.
19134 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19136 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
19137 (dump_cand): Support iv_cand.inv_exprs.
19138 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
19140 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
19143 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19145 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
19147 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
19148 as local function. Include necessary header files.
19149 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
19151 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19153 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
19155 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19157 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
19158 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
19159 RSHIFT_EXPR and BIT_NOT_EXPR.
19161 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19163 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
19164 (adjust_setup_cost): New parameter supporting round up adjustment.
19165 (struct address_cost_data): Delete.
19166 (force_expr_to_var_cost): Don't bound cost with spill_cost.
19167 (split_address_cost, ptr_difference_cost): Delete.
19168 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
19169 (struct ainc_cost_data): New struct.
19170 (get_address_cost_ainc): New function.
19171 (get_address_cost, get_computation_cost): Reimplement.
19172 (determine_group_iv_cost_address): Record inv_expr for all uses of
19174 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
19175 (iv_ca_has_deps): Reimplemented to ...
19176 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
19178 (iv_ca_extend): Call iv_ca_more_deps.
19180 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19182 * tree-ssa-address.c (struct mem_address): Move to header file.
19183 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
19184 * tree-ssa-address.h (struct mem_address): Move from C file.
19185 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
19187 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19189 * tree-affine.h (aff_combination_type): New interface.
19190 (aff_combination_zero_p): Remove static.
19191 (aff_combination_const_p): New interface.
19192 (aff_combination_singleton_var_p): New interfaces.
19194 2017-05-11 Richard Biener <rguenther@suse.de>
19196 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
19197 Skip unreachable blocks and destinations.
19198 (eliminate): Move stmt removal and fixup ...
19199 (fini_eliminate): ... here. Skip inserted exprs.
19200 (pass_pre::execute): Move fini_pre after fini_eliminate.
19201 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
19202 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
19203 PRE to get rid of dead code that has invalid SSA form and
19204 split critical edges again.
19206 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19208 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19210 2017-05-11 Richard Biener <rguenther@suse.de>
19212 * passes.c (execute_function_todo): Verify loops if they are
19213 said to be up-to-date.
19214 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
19215 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
19217 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
19220 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
19221 handle calling assemble_external ourself.
19224 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
19225 modes with zero size. Enhance comment.
19227 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19229 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
19230 built-ins for vec_xl and vec_xst with short and char pointer
19233 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
19235 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
19236 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
19237 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
19238 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
19239 (_mm_maskz_min_round_ss): New intrinsics.
19240 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
19241 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
19242 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
19243 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
19244 (__builtin_ia32_minss_mask_round): New builtins.
19245 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19246 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19247 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
19249 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
19250 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
19252 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
19255 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
19257 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
19258 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
19259 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
19260 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
19261 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
19262 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
19263 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
19264 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19265 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19266 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
19267 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
19268 (__builtin_ia32_mulss_mask_round): New builtins.
19269 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19270 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19271 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
19273 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
19274 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19276 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19279 2017-05-10 Julia Koval <julia.koval@intel.com>
19281 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
19282 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
19283 (_mm256_setr_m128i): New intrinsics.
19285 2017-05-10 Julia Koval <julia.koval@intel.com>
19287 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
19288 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
19289 (_mm_maskz_rcp14_ss): New intrinsics.
19290 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
19291 (__builtin_ia32_rcp14ss_mask): New builtins.
19292 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
19294 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
19296 PR tree-optimization/51513
19297 * tree-cfg.c (gimple_seq_unreachable_p): New function.
19298 (assert_unreachable_fallthru_edge_p): Use it.
19299 (group_case_labels_stmt): Likewise.
19300 * tree-cfg.h: Prototype it.
19301 * stmt.c: Include cfghooks.h and tree-cfg.h.
19302 (emit_case_dispatch_table) <gap_label>: New local variable.
19303 Use it to fill dispatch table gaps.
19304 Test for default_label before updating probabilities.
19305 (expand_case) <default_label>: Remove unneeded initialization.
19306 Test for unreachable default case statement and remove its edge.
19307 Set default_label accordingly.
19308 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
19310 2017-05-10 Carl Love <cel@us.ibm.com>
19312 * config/rs6000/rs6000-c: Add support for built-in functions
19313 vector signed char vec_neg (vector signed char)
19314 vector signed short int vec_neg (vector short int)
19315 vector signed int vec_neg (vector signed int)
19316 vector signed long long vec_neg (vector signed long long)
19317 vector float vec_neg (vector float)
19318 vector double vec_neg (vector double)
19319 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
19321 * config/rs6000/altivec.h: Add define for vec_neg
19322 * doc/extend.texi: Update the built-in documentation for the
19323 new built-in functions.
19325 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19327 PR tree-optimization/77644
19328 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
19330 2017-05-10 Nathan Sidwell <nathan@acm.org>
19332 * dumpfile.h (TDI_lang_all): New.
19333 (TDF_KIND): New. Renumber others
19334 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
19336 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
19338 (get_dump_file_name): Adjust suffix generation.
19339 (dump_enable_all): Use TDF_KIND.
19340 * doc/invoke.texi (-fdump-lang-all): Document.
19342 * dumpfile.h: Tabify.
19344 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
19347 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
19348 Move member access before delete.
19350 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
19352 * tree-inline.c (expand_call_inline): Split block at stmt
19355 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
19358 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
19359 are now unused after splitting mov{sf,sd}_hardfloat.
19360 (f32_lr2): Likewise.
19361 (f32_lm): Likewise.
19362 (f32_lm2): Likewise.
19363 (f32_li): Likewise.
19364 (f32_li2): Likewise.
19365 (f32_lv): Likewise.
19366 (f32_sr): Likewise.
19367 (f32_sr2): Likewise.
19368 (f32_sm): Likewise.
19369 (f32_sm2): Likewise.
19370 (f32_si): Likewise.
19371 (f32_si2): Likewise.
19372 (f32_sv): Likewise.
19373 (f32_dm): Likewise.
19374 (f32_vsx): Likewise.
19375 (f32_av): Likewise.
19376 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
19377 For movsf, order stores so the VSX stores occur before the GPR
19378 store which encourages the register allocator to use a traditional
19379 FPR instead of a GPR. For movsd, order the stores so that the GPR
19380 store comes before the VSX stores to allow the power6 to work.
19381 This is due to the power6 not having a 32-bit integer store
19382 instruction from a FPR.
19383 (movsf_hardfloat): Likewise.
19384 (movsd_hardfloat): Likewise.
19386 2017-05-09 Martin Sebor <msebor@redhat.com>
19388 PR translation/80280
19389 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
19392 PR translation/80280
19393 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
19394 data member added in r247778.
19395 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
19397 2017-05-09 Nathan Sidwell <nathan@acm.org>
19399 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
19401 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
19404 2017-05-09 Marek Polacek <polacek@redhat.com>
19406 * doc/invoke.texi: Fix typo.
19408 2017-05-09 Richard Biener <rguenther@suse.de>
19410 * tree-vrp.c (vrp_val_is_max): Adjust comment.
19411 (vrp_val_is_min): Likewise.
19412 (set_value_range_to_value): Likewise.
19413 (set_value_range_to_nonnegative): Likewise.
19414 (gimple_assign_nonzero_p): Likewise.
19415 (gimple_stmt_nonzero_p): Likewise.
19416 (vrp_int_const_binop): Likewise. Remove unreachable case.
19417 (adjust_range_with_scev): Adjust comments.
19418 (compare_range_with_value): Likewise.
19419 (extract_range_from_phi_node): Likewise.
19420 (test_for_singularity): Likewise.
19422 2017-05-09 Richard Biener <rguenther@suse.de>
19424 * tree-vrp.c (get_single_symbol): Add assert that we don't
19425 get overflowed constants as invariant part.
19426 (compare_values_warnv): Add comment before the TREE_NO_WARNING
19427 checks. Use wi::cmp instead of recursing for integer constants.
19428 (compare_values): Just ignore whether we assumed undefined
19429 overflow instead of failing the compare.
19430 (extract_range_for_var_from_comparison_expr): Add comment before the
19431 TREE_NO_WARNING sets.
19432 (test_for_singularity): Likewise.
19433 (extract_range_from_comparison): Do not disable optimization
19434 when we assumed undefined overflow.
19435 (extract_range_basic): Remove init of unused var.
19437 2017-05-09 Richard Biener <rguenther@suse.de>
19439 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
19440 (extract_range_from_multiplicative_op_1): Adjust.
19441 (extract_range_from_binary_expr_1): Use int_const_binop.
19443 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
19446 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
19447 rs6000_store_data_bypass_p in seven define_bypass directives and
19448 in several comments.
19449 * config/rs6000/rs6000-protos.h: Add prototype for
19450 rs6000_store_data_bypass_p function.
19451 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
19452 function implements slightly different (rs6000-specific) semantics
19453 than store_data_bypass_p, returning false rather than aborting
19454 with assertion error when arguments do not satisfy the
19455 requirements of store data bypass.
19456 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
19457 rs6000_store_data_bypass_p.
19459 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
19461 * config/xtensa/xtensa-protos.h
19462 (xtensa_initial_elimination_offset): New declaration.
19463 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
19464 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
19465 macro definition, add case for FRAME_POINTER_REGNUM when
19466 FRAME_GROWS_DOWNWARD.
19467 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
19468 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
19469 xtensa_initial_elimination_offset.
19471 2017-05-08 Nathan Sidwell <nathan@acm.org>
19473 * doc/invoke.texi: Alphabetize -fdump options.
19475 2017-05-08 Martin Sebor <msebor@redhat.com>
19477 PR translation/80280
19478 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
19480 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
19482 * target.def (compute_frame_layout): New optional target hook.
19483 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
19484 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
19485 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
19487 * reload1.c (verify_initial_elim_offsets): Likewise.
19488 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
19489 (use_simple_return_p): Call arm_compute_frame_layout if needed.
19490 (arm_get_frame_offsets): Split up into this ...
19491 (arm_compute_frame_layout): ... and this function.
19493 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
19495 * config/aarch64/constraints.md (Usa): New constraint.
19496 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
19498 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
19500 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
19501 with_multilib_list after it has been checked.
19503 2017-05-08 Richard Biener <rguenther@suse.de>
19505 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
19506 (bitmap_set_subtract_values): Likewise.
19508 2017-05-08 Richard Biener <rguenther@suse.de>
19510 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
19511 (gimple_assign_nonzero): ... this and remove strict_overflow_p
19513 (gimple_stmt_nonzero_warnv_p): Rename to ...
19514 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
19516 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
19517 (extract_range_basic): Adjust, do not disable propagation on
19518 strict overflow sensitive simplification.
19519 (vrp_visit_cond_stmt): Likewise.
19521 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
19523 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
19524 body size unconditionally.
19526 2017-05-07 Jeff Law <law@redhat.com>
19529 2017-05-06 Jeff Law <law@redhat.com>
19530 PR tree-optimization/78496
19531 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19534 PR tree-optimization/78496
19535 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19536 (simplify_stmt_using_ranges): Call it.
19537 (vrp_dom_walker::before_dom_children): Extract equivalences
19538 from an ASSERT_EXPR with an equality comparison against a
19541 2017-05-06 Jeff Law <law@redhat.com>
19543 PR tree-optimization/78496
19544 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19547 PR tree-optimization/78496
19548 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19549 (simplify_stmt_using_ranges): Call it.
19550 (vrp_dom_walker::before_dom_children): Extract equivalences
19551 from an ASSERT_EXPR with an equality comparison against a
19554 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19556 * lra-constraints.c (lra_copy_reg_equiv): New function.
19557 (split_reg): Use it to copy equivalence information from the
19558 original register to the spill register.
19560 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19562 PR rtl-optimization/75964
19563 * simplify-rtx.c (simplify_const_relational_operation): Remove
19564 invalid handling of comparisons of integer ABS.
19566 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
19568 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
19569 initialize to zero.
19570 (init_regs): Remove declaration.
19571 (function_arg_advance_32): Initialize error_p as boolean variable.
19573 2017-05-05 Nathan Sidwell <nathan@acm.org>
19575 * store-motion.c (remove_reachable_equiv_notes): Reformat long
19576 lines. Use for (;;).
19578 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19580 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
19581 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
19582 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
19583 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
19584 VF=2 that require versioning.
19586 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19588 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
19591 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19593 * diagnostic.h (diagnostic_override_option_index): Convert from
19594 macro to inline function.
19596 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19598 * diagnostic.c (last_module_changed_p): New function.
19599 (set_last_module): New function.
19600 (diagnostic_report_current_module): Convert macro usage to
19601 the above functions.
19602 * diagnostic.h (diagnostic_context::last_module): Strengthen
19603 from const line_map * to const line_map_ordinary *.
19604 (diagnostic_last_module_changed): Delete macro.
19605 (diagnostic_set_last_module): Delete macro.
19607 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19609 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
19610 with diagnostic_report_diagnostic.
19611 (diagnostic_n_impl_richloc): Likewise.
19612 * diagnostic.h (report_diagnostic): Delete macro.
19613 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
19614 with diagnostic_report_diagnostic.
19615 * substring-locations.c (format_warning_va): Likewise.
19617 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19619 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
19620 save/restor of format_spec. Move option-printing code to...
19621 (print_option_information): ...this new function, and
19622 reimplement by simply printing to the pretty_printer,
19623 rather than appending to the format string.
19625 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19627 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
19628 handling logic into...
19629 (update_effective_level_from_pragmas): ...this new function.
19631 2017-05-04 Andrew Waterman <andrew@sifive.com>
19633 * config/riscv/riscv.opt (mstrict-align): New option.
19634 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
19635 (SLOW_UNALIGNED_ACCESS): Define.
19636 (riscv_slow_unaligned_access): Declare.
19637 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
19639 (riscv_slow_unaligned_access): New variable.
19640 (rocket_tune_info): Set slow_unaligned_access to true.
19641 (optimize_size_tune_info): Set slow_unaligned_access to false.
19642 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
19643 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
19644 (riscv_option_override): Set riscv_slow_unaligned_access.
19645 * doc/invoke.texi: Add -mstrict-align to RISC-V.
19647 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
19649 * config/riscv/riscv.md: Unify indentation.
19651 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19656 * config/rs6000/rs6000.md (u code attribute): Add FIX and
19658 (extendsi<mode>2): Add support for doing sign extension via
19659 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
19660 don't have ISA 3.0 instructions.
19661 (extendsi<mode>2 splitter): Likewise.
19662 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
19663 generate the normal insns since SImode can now go in vector
19664 registers. Disallow the special UNSPECs needed for previous
19665 machines to hide SImode being used. Add new insns
19666 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
19667 (fix_trunc<mode>si2_stfiwx): Likewise.
19668 (fix_trunc<mode>si2_internal): Likewise.
19669 (fixuns_trunc<mode>si2): Likewise.
19670 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19671 (fctiw<u>z_<mode>_smallint): Likewise.
19672 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
19673 of floating point to 32-bit integer from doing a direct move to
19674 the GPR registers to do a store.
19675 (fctiwz_<mode>): Break long line.
19677 2017-05-05 Bin Cheng <bin.cheng@arm.com>
19679 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
19680 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
19681 (addr_list, addr_offset_valid_p): New.
19682 (split_address_groups): Check offset validity with above function.
19683 (gt-tree-ssa-loop-ivopts.h): Include header file.
19685 2017-05-05 Nathan Sidwell <nathan@acm.org>
19687 * config.gcc (arm*-*-*): Add missing 'fi'.
19689 2017-05-05 Steve Ellcey <sellcey@cavium.com>
19691 * doc/invoke.texi (-fopt-info): Explicitly say order of options
19692 included in -fopt-info does not matter.
19693 * doc/optinfo.texi (-fopt-info): Fix description of default
19694 behavour. Explicitly say order of options included in -fopt-info
19697 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19699 * config.gcc: Allow combinations of aprofile and rmprofile values for
19700 --with-multilib-list.
19701 * config/arm/t-multilib: New file.
19702 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
19703 variables. Remove setting of ISA and floating-point ABI in
19704 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
19705 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
19706 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
19707 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
19709 * config/arm/t-rmprofile: Likewise except for the matches changes.
19710 * doc/install.texi (--with-multilib-list): Document the combination of
19711 aprofile and rmprofile values and warn about pitfalls in doing that.
19713 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19715 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
19716 (movdi_aarch64): Likewise.
19718 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19720 PR tree-optimization/80632
19721 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
19723 (build_arrays): Initialize it for virtual phis.
19724 (fix_phi_nodes): Use it for virtual phis.
19726 PR tree-optimization/80558
19727 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
19728 [x, y] op z into [x op, y op z] for op & or | if conditions
19731 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
19732 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19735 * config/arm/arm.md (use_literal_pool): Remove.
19736 (64-bit immediate split): No longer takes cost into consideration
19737 if arm_disable_literal_pool is enabled.
19738 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
19739 used when arm_disable_literal_pool is enabled.
19740 (arm_max_const_double_inline_cost): Remove use of
19741 arm_disable_literal_pool.
19742 (push_minipool_fix): Add assert.
19743 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
19744 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
19745 (no_literal_pool_sf_immediate): New.
19747 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19749 PR tree-optimization/80613
19750 * tree-ssa-dce.c (propagate_necessity): Remove cases for
19751 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
19753 2017-05-05 Richard Biener <rguenther@suse.de>
19755 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
19757 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
19759 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
19760 of this flag from insn conditions due to removal from r247495.
19762 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
19764 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
19766 (arm_early_store_addr_dep_ptr): Likewise.
19767 * config/arm/aarch-common-protos.h
19768 (arm_early_load_addr_dep_ptr): Add prototype.
19769 (arm_early_store_addr_dep_ptr): Likewise.
19770 * config/arm/cortex-a53.md: Add new bypasses.
19772 2017-05-05 Jakub Jelinek <jakub@redhat.com>
19774 * tree.c (next_type_uid): Change type to unsigned.
19775 (type_hash_canon): Decrement back next_type_uid if
19776 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
19777 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
19780 2017-05-04 Martin Sebor <msebor@redhat.com>
19782 * builtins.c: Fix a trivial typo in a comment.
19784 PR middle-end/79234
19785 * builtins.c (check_sizes): Adjust to handle reading past the end.
19786 Avoid printing excessive upper bound of ranges. Use %E to print
19787 tree nodes instead of converting them to %wu.
19788 (expand_builtin_memchr): New function.
19789 (compute_dest_size): Rename...
19790 (compute_objsize): ...to this.
19791 (expand_builtin_memcpy): Adjust.
19792 (expand_builtin_mempcpy): Adjust.
19793 (expand_builtin_strcat): Adjust.
19794 (expand_builtin_strcpy): Adjust.
19795 (check_strncat_sizes): Adjust.
19796 (expand_builtin_strncat): Adjust.
19797 (expand_builtin_strncpy): Adjust and simplify.
19798 (expand_builtin_memset): Adjust.
19799 (expand_builtin_bzero): Adjust.
19800 (expand_builtin_memcmp): Adjust.
19801 (expand_builtin): Handle memcmp.
19802 (maybe_emit_chk_warning): Check strncat just once.
19804 2017-05-04 Martin Sebor <msebor@redhat.com>
19806 PR preprocessor/79214
19807 PR middle-end/79222
19808 PR middle-end/79223
19809 * builtins.c (check_sizes): Add inlining context and issue
19810 warnings even when -Wno-system-headers is set.
19811 (check_strncat_sizes): Same.
19812 (expand_builtin_strncat): Same.
19813 (expand_builtin_memmove): New function.
19814 (expand_builtin_stpncpy): Same.
19815 (expand_builtin): Handle memmove and stpncpy.
19817 2017-05-04 Bin Cheng <bin.cheng@arm.com>
19819 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
19820 which is not used any more.
19822 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19824 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
19826 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19828 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
19829 (cortexa53_tunings): Likewise.
19830 (cortexa57_tunings): Likewise.
19831 (cortexa72_tunings): Likewise.
19832 (cortexa73_tunings): Likewise.
19834 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19836 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
19837 Set loop alignment to 8.
19839 2017-05-04 Martin Sebor <msebor@redhat.com>
19841 PR translation/80280
19842 * builtins.c (expand_builtin_object_size): Add missing quoting to
19843 %D and like directives.
19844 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
19845 (hsa_type_for_tree_type): Same.
19846 (verify_function_arguments): Same.
19847 * symtab.c (symbol_table::change_decl_assembler_name): Same.
19848 * varasm.c (get_section): Same.
19851 2017-05-04 Martin Sebor <msebor@redhat.com>
19853 PR translation/80280
19854 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
19856 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
19858 * config/aarch64/aarch64.c (generic_addrcost_table):
19859 Change HI/TI mode setting.
19861 2017-05-04 Martin Jambor <mjambor@suse.cz>
19863 PR tree-optimization/80622
19864 * tree-sra.c (comes_initialized_p): New function.
19865 (build_accesses_from_assign): Only set write lazily when
19866 comes_initialized_p is false.
19867 (analyze_access_subtree): Use comes_initialized_p.
19868 (propagate_subaccesses_across_link): Assert !comes_initialized_p
19869 instead of testing for PARM_DECL.
19871 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19873 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
19874 constraint on operand 0 to allow more general addressing modes.
19875 Adjust output template.
19876 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
19878 * config/aarch64/aarch64-protos.h
19879 (aarch64_address_valid_for_prefetch_p): Declare prototype.
19880 * config/aarch64/constraints.md (Dp): New address constraint.
19881 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
19884 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
19886 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
19887 update use of estimate_ipcp_clone_size_and_time.
19888 (estimate_local_effects): Update use of
19889 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
19890 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
19891 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
19892 Return nonspecialized time.
19894 2017-05-04 Richard Biener <rguenther@suse.de>
19896 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
19897 for the last VUSE which def dominates the PHI. Directly call
19899 (get_continuation_for_phi_1): Remove.
19901 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
19903 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
19904 to explain the use of truncating division. Cap the number of
19905 iterations to the maximum given by nb_iterations_upper_bound,
19908 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
19910 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
19911 * configure: Regenerate.
19912 * config.in: Regenerate.
19913 * config/i386/driver-mingw32.c: new file.
19914 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
19915 * config.host: Link driver-mingw32.o on MinGW host.
19916 * doc/install.texi: Document new --enable-mingw-wildcard configure
19919 2017-05-04 Marek Polacek <polacek@redhat.com>
19921 PR tree-optimization/80612
19922 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
19924 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19925 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
19927 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
19928 (movt splitter): Likewise.
19929 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
19930 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
19931 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
19932 block for Thumb-1 with MOVT.
19933 (thumb2_legitimate_address_p): Move code block ...
19934 (can_avoid_literal_pool_for_label_p): ... into this new function.
19935 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
19937 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
19938 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
19939 "M-profile targets with the MOVT instruction".
19941 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19943 * config/arm/arm-builtins.c (arm_init_builtins): Rename
19944 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
19945 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
19947 2017-05-04 Martin Liska <mliska@suse.cz>
19949 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
19950 variable cond_code.
19952 2017-05-04 Richard Biener <rguenther@suse.de>
19954 * tree.c (array_at_struct_end_p): Handle arrays at struct
19955 end with flexarrays more conservatively. Refactor and treat
19956 arrays of arrays or aggregates more strict. Fix
19957 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
19958 * tree.c (array_at_struct_end_p): Adjust prototype.
19959 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
19960 * gimple-fold.c (get_range_strlen): Likewise.
19961 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
19963 2017-05-04 Richard Biener <rguenther@suse.de>
19965 PR tree-optimization/31130
19966 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
19968 (supports_overflow_infinity): Likewise.
19969 (is_negative_overflow_infinity): Likewise.
19970 (is_positive_overflow_infinity): Likewise.
19971 (is_overflow_infinity): Likewise.
19972 (stmt_overflow_infinity): Likewise.
19973 (overflow_infinity_range_p): Likewise.
19974 (usable_range_p): Remove as always returning true.
19975 (make_overflow_infinity): Remove.
19976 (negative_overflow_infinity): Likewise.
19977 (positive_overflow_infinity): Likewise.
19978 (avoid_overflow_infinity): Likewise.
19979 (set_value_range): Adjust accordingly.
19980 (set_value_range_to_nonnegative): Likewise, remove now unused
19981 overflow_infinity arg.
19982 (vrp_operand_equal_p): Adjust.
19983 (update_value_range): Likewise.
19984 (range_int_cst_singleton_p): Likewise.
19985 (operand_less_p): Likewise.
19986 (compare_values_warnv): Likewise.
19987 (extract_range_for_var_from_comparison_expr): Likewise.
19988 (vrp_int_const_binop): Likewise.
19989 (zero_nonzero_bits_from_vr): Likewise.
19990 (extract_range_from_multiplicative_op_1): Likewise.
19991 (extract_range_from_binary_expr_1): Likewise.
19992 (extract_range_from_unary_expr): Likewise.
19993 (extract_range_from_comparison): Likewise.
19994 (extract_range_basic): Likewise.
19995 (adjust_range_with_scev): Likewise.
19996 (compare_ranges): Likewise.
19997 (compare_range_with_value): Likewise.
19998 (dump_value_range): Likewise.
19999 (test_for_singularity): Likewise, remove strict_overflow_p parameter
20001 (simplify_cond_using_ranges): Adjust.
20003 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20005 * brig-builtins.def: Added a builtin for class_f64.
20006 * builtin-types.def: Added a builtin type needed by class_f64.
20008 2017-05-03 Jason Merrill <jason@redhat.com>
20010 * timevar.def: Add TV_CONSTEXPR.
20012 2017-05-03 David Malcolm <dmalcolm@redhat.com>
20014 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
20016 2017-05-03 Martin Jambor <mjambor@suse.cz>
20018 * ipa-prop.c (ipa_update_after_lto_read): Removed.
20019 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
20020 * ipa-cp.c (ipcp_propagate_stage): Do not call
20021 ipa_update_after_lto_read.
20022 * ipa-inline.c (ipa_inline): Likewise.
20024 2017-05-03 Martin Jambor <mjambor@suse.cz>
20026 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
20027 tag. Added a default constructor and a destructor.
20028 (ipa_edge_args_sum_t): New class;
20029 (ipa_edge_args_sum): Declare.
20030 (ipa_edge_args_vector): Remove declaration.
20031 (IPA_EDGE_REF): Use ipa_edge_args_sum.
20032 (ipa_free_edge_args_substructures): Remove declaration.
20033 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
20034 (ipa_edge_args_info_available_for_edge_p): Likewise.
20035 * ipa-prop.c (ipa_edge_args_vector): Removed.
20036 (edge_removal_hook_holder): Likewise.
20037 (edge_duplication_hook_holder): Likewise.
20038 (ipa_edge_args_sum): New variable.
20039 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
20040 ipa_edge_args_vector.
20041 (ipa_free_edge_args_substructures): Likewise.
20042 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
20043 ipa_edge_args_vector.
20044 (ipa_edge_removal_hook): Turned into method
20045 ipa_edge_args_sum_t::remove.
20046 (ipa_edge_duplication_hook): Turned into method
20047 ipa_edge_args_sum_t::duplicate.
20048 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
20049 registering edge hooks.
20050 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
20051 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
20052 ipa_edge_args_sum instead of ipa_edge_args_vector.
20053 * ipa-profile.c (ipa_profile): Likewise.
20055 2017-05-03 Martin Jambor <mjambor@suse.cz>
20057 * symbol-summary.h (function_summary): New method exists.
20058 (function_summary::symtab_removal): Deallocate through release.
20059 (call_summary): New class.
20060 (gt_ggc_mx): New overload.
20061 (gt_pch_nx): Likewise.
20062 (gt_pch_nx): Likewise.
20064 2017-05-03 Jeff Law <law@redhat.com>
20066 PR tree-optimization/78496
20067 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
20068 from simplify_cond_using_ranges. Split off code to walk
20069 backwards through casts into ...
20070 (simplify_cond_using_ranges_2): New function.
20071 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
20072 (execute_vrp): After identifying jump threads, call
20073 simplify_cond_using_ranges_2.
20075 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
20078 * ipa-inline.h (inline_summary): Add ctor.
20079 (create_ggc): Do not use ggc_cleared_alloc.
20081 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
20082 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20084 * gcc.c (handle_braces): Support escaping in switch matching
20086 * doc/invoke.texi (Spec Files): Document it.
20087 Remove superfluous @code markup in items.
20089 2017-05-03 David Malcolm <dmalcolm@redhat.com>
20091 * diagnostic-show-locus.c (struct column_range): New struct.
20092 (get_affected_columns): New function.
20093 (get_printed_columns): New function.
20094 (struct correction): New struct.
20095 (correction::ensure_capacity): New function.
20096 (correction::ensure_terminated): New function.
20097 (struct line_corrections): New struct.
20098 (line_corrections::~line_corrections): New dtor.
20099 (line_corrections::add_hint): New function.
20100 (layout::print_trailing_fixits): Reimplement in terms of the new
20102 (selftest::test_overlapped_fixit_printing): New function.
20103 (selftest::diagnostic_show_locus_c_tests): Call it.
20105 2017-05-03 Nathan Sidwell <nathan@acm.org>
20107 Canonicalize canonical type hashing
20108 * tree.h (type_hash_canon_hash): Declare.
20109 * tree.c (type_hash_list, attribute_hash_list): Move into
20110 type_hash_canon_hash.
20111 (build_type_attribute_qual_variant): Break out hash code calc into
20112 type_hash_canon_hash.
20113 (type_hash_canon_hash): New. Generic type hash computation.
20114 (build_range_type_1, build_array_type_1, build_function_type,
20115 build_method_type_directly, build_offset_type, build_complex_type,
20116 make_vector_type): Call it.
20118 2017-05-03 Richard Biener <rguenther@suse.de>
20120 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20121 When all DRs have unknown misaligned do not always peel
20122 when there is a store but apply the same costing model as if
20123 there were only loads.
20125 2017-05-03 Richard Biener <rguenther@suse.de>
20128 PR tree-optimization/80492
20129 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20130 compare_base_decls returning dont-know properly.
20132 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
20134 * config/arm/iterators.md (CCSI): New mode iterator.
20135 (arch): New mode attribute.
20136 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
20137 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
20138 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
20139 code iterator for success result mode.
20140 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
20141 the corresponding new insn generators.
20143 2017-05-03 Bin Cheng <bin.cheng@arm.com>
20146 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20147 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20149 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
20151 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
20152 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
20153 (DDR_A): Wrap DDR argument in brackets.
20154 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
20155 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
20156 (DDR_REVERSED_P): Likewise.
20158 2017-05-03 Jakub Jelinek <jakub@redhat.com>
20160 PR tree-optimization/79472
20161 * tree-switch-conversion.c (struct switch_conv_info): Add
20162 contiguous_range and default_case_nonstandard fields.
20163 (collect_switch_conv_info): Compute contiguous_range and
20164 default_case_nonstandard fields, don't clear final_bb if
20165 contiguous_range and only the default case doesn't have the required
20167 (check_all_empty_except_final): Set default_case_nonstandard instead
20168 of failing if contiguous_range and the default case doesn't have empty
20170 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
20171 and only the default case doesn't have the required constants. Skip
20173 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
20174 if default_case_nonstandard.
20175 (build_constructors): Build constant 1 just once. Assert that default
20176 values aren't inserted in between cases if contiguous_range. Skip
20178 (build_arrays): Skip virtual phis.
20179 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
20180 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
20181 Handle virtual phis.
20182 (gen_inbound_check): Handle default_case_nonstandard case.
20183 (process_switch): Adjust check_final_bb caller. Call
20184 gather_default_values with the first non-default case instead of
20185 default case if default_case_nonstandard.
20187 2017-05-02 Nathan Sidwell <nathan@acm.org>
20189 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
20190 check. Fix formatting.
20192 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
20194 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
20195 errors when comparing specialized and unspecialized times.
20197 2017-05-02 David Malcolm <dmalcolm@redhat.com>
20199 * diagnostic-show-locus.c
20200 (layout::should_print_annotation_line_p): Make private.
20201 (layout::print_annotation_line): Make private.
20202 (layout::annotation_line_showed_range_p): Make private.
20203 (layout::show_ruler): Make private.
20204 (layout::print_source_line): Make private. Pass in line and
20205 line_width, rather than calling location_get_source_line. Drop
20207 (layout::print_leading_fixits): New method.
20208 (layout::print_any_fixits): Rename to...
20209 (layout::print_trailing_fixits): ...this, and make private.
20210 Don't print newline fixits.
20211 (diagnostic_show_locus): Move logic for printing one row into...
20212 (layout::print_line): ...this new function. Move the
20213 location_get_source_line call and error-handling from
20214 print_source_line to here. Call print_leading_fixits, and rename
20215 print_any_fixits to print_trailing_fixits.
20216 (selftest::test_fixit_insert_containing_newline): Update now that
20217 newlines are partially supported.
20218 (selftest::test_fixit_insert_containing_newline_2): New test.
20219 (selftest::test_fixit_replace_containing_newline): Update comments.
20220 (selftest::diagnostic_show_locus_c_tests): Call the new test.
20221 * edit-context.c (class added_line): New class.
20222 (class edited_line): Describe newline handling in comment.
20223 (edited_line::actually_edited_p): New method.
20224 (edited_line::print_content): Delete redundant decl.
20225 (edited_line::m_predecessors): New field.
20226 (edited_file::print_content): Call edited_line::print_content.
20227 (edited_file::print_diff): Update to support newlines.
20228 (edited_file::print_diff_hunk): Likewise.
20229 (edited_file::print_run_of_changed_lines): New function.
20230 (edited_file::print_diff_line): Convert to...
20231 (print_diff_line): ...this.
20232 (edited_file::get_effective_line_count): New function.
20233 (edited_line::edited_line): Initialize new field m_predecessors.
20234 (edited_line::~edited_line): Clean up m_predecessors.
20235 (edited_line::apply_fixit): Handle newlines.
20236 (edited_line::get_effective_line_count): New function.
20237 (edited_line::print_content): New function.
20238 (edited_line::print_diff_lines): New function.
20239 (selftest::test_applying_fixits_insert_containing_newline): New
20241 (selftest::test_applying_fixits_replace_containing_newline): New
20243 (selftest::insert_line): New function.
20244 (selftest::test_applying_fixits_multiple_lines): Add example of
20246 (selftest::edit_context_c_tests): Call the new tests.
20248 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20250 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
20251 parameter cand. Update dump information.
20252 (get_computation_cost): Update uses.
20254 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20256 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
20257 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
20258 (get_computation_at, rewrite_use_address): Update use of
20259 get_computation_aff.
20261 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20263 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
20264 (get_computation): Delete.
20265 (get_computation_cost): Implement like get_computation_cost_at.
20266 Use get_computation_at.
20267 (get_computation_cost_at): Delete.
20268 (rewrite_use_nonlinear_expr): Use get_computation_at.
20269 (rewrite_use_compare, remove_unused_ivs): Ditto.
20271 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20273 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
20275 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20277 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
20278 (ivopts_global_cost_for_size): Rename parameter and update uses.
20279 (iv_ca_recount_cost): Update uses.
20280 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
20281 candidates seperately in n_invs and n_cands.
20282 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
20284 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20286 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
20287 (find_inv_vars_cb): New.
20288 (find_depends): Renamed to ...
20289 (find_inv_vars): ... this.
20290 (add_candidate_1, force_var_cost): Call find_inv_vars.
20291 (split_address_cost, determine_group_iv_cost_cond): Ditto.
20293 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20295 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
20296 inv_vars. Add inv_exprs.
20297 (struct iv_cand): Rename depends_on to inv_vars.
20298 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
20299 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
20300 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
20301 (dump_cand): Dump inv_vars.
20302 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
20303 (record_invariant, find_depends, add_candidate_1): Ditto.
20304 (set_group_iv_cost, force_var_cost): Ditto.
20305 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
20306 (get_computation_cost_at, get_computation_cost): Ditto.
20307 (determine_group_iv_cost_generic): Ditto.
20308 (determine_group_iv_cost_address): Ditto.
20309 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
20310 (determine_group_iv_costs): Ditto.
20311 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
20312 (iv_ca_set_remove_invariants): Renamed to ...
20313 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
20314 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
20315 (iv_ca_set_add_invariants): Renamed to ...
20316 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
20317 (iv_ca_set_cp): Use iv_ca_set_add_invs.
20318 (iv_ca_has_deps): Support inv_vars and inv_exprs.
20319 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
20320 (create_new_ivs): Remove useless dump.
20322 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20324 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
20326 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
20327 (iv_ca_set_no_cp, create_new_iv): Ditto.
20329 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20331 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20333 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20335 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
20336 function tree_check2.
20338 2017-05-02 Martin Liska <mliska@suse.cz>
20340 * doc/gcov.texi: Add missing preposition.
20341 * gcov.c (function_info::function_info): Properly fill up
20342 all member variables.
20344 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20346 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
20348 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20350 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
20352 2017-05-02 Martin Liska <mliska@suse.cz>
20355 * lto-streamer-in.c (lto_read_tree_1): Remove
20356 LTO_STREAMER_DEBUG.
20357 * lto-streamer.c (struct tree_hash_entry): Likewise.
20358 (struct tree_entry_hasher): Likewise.
20359 (tree_entry_hasher::hash): Likewise.
20360 (tree_entry_hasher::equal): Likewise.
20361 (lto_streamer_init): Likewise.
20362 (lto_orig_address_map): Likewise.
20363 (lto_orig_address_get): Likewise.
20364 (lto_orig_address_remove): Likewise.
20365 * lto-streamer.h: Likewise.
20366 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
20367 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
20369 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
20371 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
20372 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
20373 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
20374 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
20375 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
20376 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
20377 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
20378 (mm_maskz_sub_ss): New intrinsics.
20379 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20380 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20381 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
20382 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
20383 (__builtin_ia32_subss_mask_round): New builtins.
20384 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20385 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20386 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
20388 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
20389 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20391 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20394 2017-05-02 Martin Jambor <mjambor@suse.cz>
20396 PR tree-optimization/78687
20397 * tree-sra.c (access): New field parent.
20398 (process_subtree_disqualification): New function.
20399 (disqualify_candidate): Call it.
20400 (build_accesses_from_assign): Reset write flag if creating an
20402 (build_access_subtree): Fill in parent field and also prpagate
20403 down grp_write flag.
20404 (create_artificial_child_access): New parameter set_grp_write, set
20405 grp_write to its value.
20406 (propagate_subaccesses_across_link): Also propagate grp_write flag
20408 (propagate_all_subaccesses): Push the closest parent back to work
20409 queue if add_access_to_work_queue returned true.
20411 2017-05-02 Richard Biener <rguenther@suse.de>
20413 * common.opt (fstrict-overflow): Alias negative to fwrapv.
20414 * doc/invoke.texi (fstrict-overflow): Remove all traces of
20415 -fstrict-overflow documentation.
20416 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
20417 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
20418 flag_strict_overflow.
20419 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
20420 * lto-opts.c (lto_write_options): Do not stream it.
20421 * lto-wrapper.c (merge_and_complain): Do not handle it.
20422 * opts.c (default_options_table): Do not set -fstrict-overflow.
20423 (finish_options): Likewise do not clear it when sanitizing.
20424 * simplify-rtx.c (simplify_const_relational_operation): Do not
20425 test flag_strict_overflow.
20427 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20429 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
20430 using enabled attribute.
20431 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
20432 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
20433 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
20434 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
20435 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
20436 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
20437 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
20438 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
20439 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
20440 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
20442 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20444 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
20446 2017-05-02 Richard Biener <rguenther@suse.de>
20448 PR tree-optimization/80591
20450 2017-04-10 Richard Biener <rguenther@suse.de>
20452 * tree-ssa-structalias.c (find_func_aliases): Properly handle
20455 2017-05-02 Richard Biener <rguenther@suse.de>
20457 PR tree-optimization/80549
20458 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
20459 (cleanup_tree_cfg_noloop): Create forwarders to known loop
20460 headers if they do not have a preheader.
20462 2017-05-02 Martin Liska <mliska@suse.cz>
20465 * common.opt: Fix typo.
20466 * doc/invoke.texi: Likewise.
20468 2017-05-01 Jan Beulich <jbeulich@suse.com>
20470 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
20471 swapping, add (x,x,m,x,n) alternative.
20473 2017-05-01 Nathan Sidwell <nathan@acm.org>
20475 * calls.c (combine_pending_stack_adjustment_and_call): Remove
20476 unnecessary unadjusted_alignment check.
20478 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20481 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
20483 * gimplify.c (gimplify_cilk_detach): New function.
20484 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
20485 * tree-core.h: Document EXPR_CILK_SPAWN.
20486 * tree.h (EXPR_CILK_SPAWN): Define.
20488 2017-05-01 David Malcolm <dmalcolm@redhat.com>
20490 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
20491 to use new fixit_hint representation, using the "replace" logic.
20492 (get_line_span_for_fixit_hint): Likewise.
20493 (layout::print_any_fixits): Likewise.
20494 (selftest::test_one_liner_many_fixits): Rename to...
20495 (selftest::test_one_liner_many_fixits_1): ...this, and update
20496 comment and expected output to reflect that the multiple fix-it
20497 hints are now consolidated into one insertion.
20498 (selftest::test_one_liner_many_fixits_2): New test.
20499 (selftest::test_diagnostic_show_locus_one_liner): Update for
20501 (selftest::test_fixit_consolidation): Update for fix-it API
20503 * diagnostic.c (print_parseable_fixits): Likewise.
20504 * edit-context.c (edited_line::m_line_events): Convert from
20505 auto_vec <line_event *> to auto_vec <line_event>.
20506 (class line_event): Convert from abstract base class to a concrete
20507 class, taking over the role of replace_event.
20508 (class insert_event): Delete.
20509 (class replace_event): Rename to class line_event. Convert to
20511 (edit_context::add_fixits): Reimplement.
20512 (edit_context::apply_insert): Delete.
20513 (edit_context::apply_replace): Rename to...
20514 (edit_context::apply_fixit): ...this. Convert to half-open range.
20515 (edited_file::apply_insert): Delete.
20516 (edited_file::apply_replace): Rename to...
20517 (edited_file::apply_fixit): ...this.
20518 (edited_line::~edited_line): Drop deletion of events.
20519 (edited_line::apply_insert): Delete.
20520 (edited_line::apply_replace): Rename to...
20521 (edited_line::apply_fixit): ...this. Convert to half-open range.
20522 Update for change to type of m_line_events.
20523 * edit-context.h (edit_context::apply_insert): Delete.
20524 (edit_context::apply_replace): Rename to...
20525 (edit_context::apply_fixit): ...this.
20527 2017-05-01 Martin Sebor <msebor@redhat.com>
20529 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
20532 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
20535 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
20536 __get_cpuid_max returns 0.
20537 (__get_cpuid_count): Ditto.
20539 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
20541 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
20542 replacement expression is another instance of one of its arguments.
20544 2017-05-01 Jakub Jelinek <jakub@redhat.com>
20547 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
20548 check for stack push/pop autoinc.
20549 * config/i386/i386.c (ix86_agi_dependent): Return false
20550 if the only reason why modified_in_p returned true is that
20551 addr is SP based and set_insn is a push or pop.
20553 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20555 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
20558 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20561 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
20562 (account_size_time): Use two predicates - exec_pred and
20564 (evaluate_conditions_for_known_args): Compute both clause and
20566 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
20567 (inline_summary_t::duplicate): Update.
20568 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
20570 (compute_inline_parameters): Likewise.
20571 (estimate_edge_size_and_time): Update caluclation of time.
20572 (estimate_node_size_and_time): Compute both time and nonspecialized
20574 (estimate_ipcp_clone_size_and_time): Update.
20575 (inline_merge_summary): Update.
20576 (do_estimate_edge_time): Update.
20577 (do_estimate_edge_size): Update.
20578 (do_estimate_edge_hints): Update.
20579 (inline_read_section, inline_write_summary): Stream both new predicates.
20580 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
20582 (compute_inlined_call_time): Cleanup.
20583 (big_speedup_p): Update.
20584 (edge_badness): Update.
20585 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
20586 (size_time_entry): Replace predicate by exec_predicate and
20587 nonconst_predicate.
20588 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
20589 (estimate_edge_time): Return also nonspec_time.
20590 (reset_edge_growth_cache): Update.
20592 2017-04-29 Jakub Jelinek <jakub@redhat.com>
20594 PR rtl-optimization/80491
20595 * ifcvt.c (noce_process_if_block): When looking for x setter
20596 with missing else_bb, don't check only the insn right before
20597 cond_earliest, but look for the last insn that x is modified in
20598 within the same bb.
20600 PR rtl-optimization/80491
20601 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
20603 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
20605 PR tree-optimization/80487
20606 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
20608 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20610 PR tree-optimization/79697
20611 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
20612 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
20613 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
20615 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
20616 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
20618 2017-04-28 Martin Sebor <msebor@redhat.com>
20620 PR tree-optimization/80523
20621 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
20622 (init_target_to_host_charmap, target_to_host, target_strtol10): New
20624 (maybe_warn, format_directive, parse_directive): Use new functions.
20625 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
20627 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
20629 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
20631 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
20633 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
20634 target_header_dir): Set correctly.
20635 * configure: Regenerated.
20636 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20637 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20638 instead of SYSTEM_HEADER_DIR.
20640 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
20642 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
20643 (estimate_local_effects): Likewise.
20644 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
20645 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
20646 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
20647 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
20648 do_estimate_edge_time, estimate_edge_time): Likewise.
20649 * ipa-inline-analysis.c (estimate_node_size_and_time,
20650 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
20651 (estimate_time_after_inlining): Remove.
20653 2017-04-28 Martin Liska <mliska@suse.cz>
20655 * doc/gcov.texi: Enhance documentation of gcov.
20657 2017-04-28 Martin Liska <mliska@suse.cz>
20659 * doc/gcov.texi: Sort options in alphabetic order.
20660 * doc/gcov-dump.texi: Likewise.
20661 * doc/gcov-tool.texi: Likewise.
20662 * gcov.c (print_usage): Likewise.
20663 * gcov-dump.c (print_usage): Likewise.
20664 * gcov-tool.c (print_merge_usage_message): Likewise.
20665 (print_rewrite_usage_message): Likewise.
20666 (print_overlap_usage_message): Likewise.
20668 2017-04-28 Martin Liska <mliska@suse.cz>
20670 PR gcov-profile/53915
20671 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
20673 2017-04-28 Martin Liska <mliska@suse.cz>
20675 PR gcov-profile/79891
20676 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
20677 is marked by compiler as living on a line.
20678 (get_cycles_count): Remove usage of the union.
20679 (output_intermediate_file): Likewise.
20680 (find_source): Fix GNU coding style.
20681 (accumulate_line_counts): Remove old non-all block mode.
20682 (output_lines): Remove usage of the union.
20683 * profile.c (output_location): Include all BBs, even if
20684 belonging to a same line (and file) as a previous BB.
20686 2017-04-28 Martin Liska <mliska@suse.cz>
20688 * gcov.c (process_args): Handle new argument 'w'.
20689 (read_graph_file): Assign ID to BBs.
20690 (output_branch_count): Display BB # if verbose flag is set.
20691 (output_lines): Likewise for arcs.
20692 (print_usage): Add '--verbose' option help.
20693 * doc/gcov.texi: Document --verbose (-w) option.
20695 2017-04-28 Martin Liska <mliska@suse.cz>
20697 * gcov.c (struct block_location_info): New struct.
20698 (process_file): Fill up the new structure.
20699 (read_graph_file): Replace usage of encoding by the newly added
20701 (add_line_counts): Likewise.
20702 (accumulate_line_counts): Remove usage of the union.
20703 (function_info::function_info): New function.
20704 (function_info::~function_info): Likewise.
20705 (process_file): Call delete instead of release_function.
20706 (release_function): Release the function.
20707 (release_structures): Call delete instead of release_function.
20708 (solve_flow_graph): Replace usage of num_blocks.
20709 (find_exception_blocks): Likewise.
20710 (output_lines): Fix GNU coding style.
20712 2017-04-28 Martin Liska <mliska@suse.cz>
20715 * coverage.c (coverage_remove_note_file): New function.
20716 * coverage.h: Declare the function.
20717 * toplev.c (finalize): Clean if an error has been seen.
20719 2017-04-28 Martin Liska <mliska@suse.cz>
20721 PR gcov-profile/80031
20722 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
20723 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
20724 * gcov.c (read_graph_file): Read just number of blocks.
20725 * profile.c (branch_prob): Do not stream 0 flags per a basic
20728 2017-04-28 Martin Liska <mliska@suse.cz>
20730 * gcov-dump.c (tag_*): Add new argument to declarations.
20731 (dump_gcov_file): Likewise.
20732 (tag_blocks): Add and use new argument depth.
20733 (tag_arcs): Likewise.
20734 (tag_lines): Likewise.
20735 (tag_counters): Likewise.
20736 (tag_summary): Likewise.
20737 (dump_working_sets): Use depth to do a proper indentation.
20739 2017-04-28 Jakub Jelinek <jakub@redhat.com>
20742 * cgraph.h (symtab_node::debug_symtab): No longer inline.
20743 * symtab.c (symtab_node::debug_symtab): Move definition here.
20745 2017-04-28 Richard Biener <rguenther@suse.de>
20747 * lto-streamer.h (LTO_major_version): Bump to 7.
20749 2017-04-28 Richard Biener <rguenther@suse.de>
20751 * tree-vrp.c (assert_info): New struct.
20752 (add_assert_info): New helper.
20753 (register_edge_assert_for_2): Refactor to add asserts to a vector
20755 (register_edge_assert_for_1): Likewise.
20756 (register_edge_assert_for): Likewise.
20757 (finish_register_edge_assert_for): New helper actually registering
20758 asserts where live on edge.
20759 (find_conditional_asserts): Adjust.
20760 (find_switch_asserts): Likewise.
20761 (evrp_dom_walker::try_find_new_range): Generalize.
20762 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
20764 2017-04-27 Marek Polacek <polacek@redhat.com>
20767 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
20768 arg10 and arg11 to itype.
20770 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
20772 * doc/extend.texi (Object Size Checking): Improve grammar.
20774 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
20777 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
20778 that the logic for permitting reciprocal estimates matches that
20781 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20784 * tree.c (type_cache_hasher::equal): Only compare
20785 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
20786 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
20787 non-aggregate element types.
20788 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
20789 about the flag on ARRAY_TYPEs in the comment, formatting fix.
20791 2017-04-27 Richard Biener <rguenther@suse.de>
20793 PR middle-end/80533
20794 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
20795 stripping ARRAY_REFs from MEM_EXPR make sure we're not
20796 keeping a reference to a trailing array.
20798 2017-04-27 Richard Biener <rguenther@suse.de>
20800 PR middle-end/80539
20801 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
20802 being in loop-closed SSA form conservatively.
20803 (chrec_fold_multiply_poly_poly): Likewise.
20805 2017-04-27 Tamar Christina <tamar.christina@arm.com>
20807 PR middle-end/79665
20808 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
20809 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
20811 2017-04-27 Jakub Jelinek <jakub@redhat.com>
20814 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
20815 (aarch64_function_arg_alignment): Return unsigned int again, but still
20816 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
20817 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20818 Don't emit -Wpsabi note.
20819 (aarch64_function_arg_boundary): Likewise.
20820 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20823 2017-04-26 Nathan Sidwell <nathan@acm.org>
20825 * tree.h (crc32_unsigned_n): Declare.
20826 (crc32_unsigned, crc32_unsigned): Make inline.
20827 * tree.c (crc32_unsigned_bits): Replace with ...
20828 (crc32_unsigned_n): ... this.
20829 (crc32_unsigned, crc32_byte): Remove.
20830 (crc32_string): Remove unnecessary braces.
20832 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20834 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
20835 * ipa-inline-analysis.c (MAX_TIME): Remove.
20836 (account_size_time): Use sreal for time.
20837 (dump_inline_summary): Update.
20838 (estimate_function_body_sizes): Update.
20839 (estimate_edge_size_and_time): Update.
20840 (estimate_calls_size_and_time): Update.
20841 (estimate_node_size_and_time): Update.
20842 (inline_merge_summary): Update.
20843 (inline_update_overall_summary): Update.
20844 (estimate_time_after_inlining): Update.
20845 (inline_read_section): Update.
20846 (inline_write_summary): Update.
20847 * ipa-inline.c (compute_uninlined_call_time): Update.
20848 (compute_inlined_call_time): Update.
20849 (recursive_inlining): Update.
20850 (inline_small_functions): Update.
20851 (dump_overall_stats): Update.
20852 * ipa-inline.h: Include sreal.h.
20853 (size_time_entry): Turn time to sreal.
20854 (inline_summary): Turn self_time nad time to sreal.
20856 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
20858 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
20860 (sreal::stream_out, sreal::stream_in): New.
20861 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
20863 2017-04-25 Jakub Jelinek <jakub@redhat.com>
20865 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
20868 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
20871 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
20873 (dimode_scalar_chain::compute_convert_gain): Ditto.
20874 (dimode_scalar_chain::make_vector_copies): Ditto.
20875 (dimode_scalar_chain::convert_reg): Ditto.
20876 (dimode_scalar_chain::convert_insn): Ditto.
20877 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
20878 (VI248_AVX512BW_1): New mode iterator.
20879 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
20880 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
20883 2017-04-25 Martin Sebor <msebor@redhat.com>
20885 PR tree-optimization/80497
20886 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
20887 constants are representable in HOST_WIDE_INT.
20888 (parse_directive): Ditto.
20890 2017-04-25 Martin Sebor <msebor@redhat.com>
20893 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
20894 (new_zero_array): Adjust signature.
20895 (dom_info::dom_init): Used unsigned rather that size_t.
20896 (dom_info::dom_info): Same.
20898 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20899 Jakub Jelinek <jakub@redhat.com>
20902 * config/arm/arm.c: Include gimple.h.
20903 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20904 returns negative, increment ncrn only if it returned positive.
20905 (arm_needs_doubleword_align): Return int instead of bool,
20906 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
20907 members, but if there is any such non-FIELD_DECL
20908 > PARM_BOUNDARY aligned decl, return -1 instead of false.
20909 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
20910 returns negative, increment nregs only if it returned positive.
20911 (arm_setup_incoming_varargs): Likewise.
20912 (arm_function_arg_boundary): Emit -Wpsabi note if
20913 arm_needs_doubleword_align returns negative, return
20914 DOUBLEWORD_ALIGNMENT only if it returned positive.
20916 2017-04-25 Marek Polacek <polacek@redhat.com>
20919 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
20920 first argument to type.
20922 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
20925 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
20926 type checks to test for compatibility instead of equality.
20928 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20929 Jakub Jelinek <jakub@redhat.com>
20932 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
20934 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
20935 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
20936 the alignment computation, but return their maximum in warn_alignment.
20937 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
20938 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
20940 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
20941 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
20944 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20946 * config/arc/simdext.md (dmpyh): Fix typo.
20948 2017-04-25 Richard Biener <rguenther@suse.de>
20950 PR tree-optimization/80492
20951 * alias.c (compare_base_decls): Handle registers with asm
20952 specification conservatively.
20953 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20954 compare_base_decls returning dont-know properly.
20956 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20958 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
20959 (legitimate_offset_address_p): New function.
20960 (arc_legitimate_address_p): Use above function.
20962 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20964 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
20966 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20968 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
20969 ACCH registers whenever they are available.
20971 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20973 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
20974 double regs fix when not used.
20976 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20978 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
20980 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
20981 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
20983 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20985 * config/arc/arc.c (arc_output_addsi): Check for h-register class
20986 when emitting short ADD instructions.
20988 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
20990 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
20992 (cmpsi_cc_c_insn): Likewise.
20993 (cbranchsi4_scratch): Compute proper instruction length using
20994 compact_hreg_operand.
20995 * config/arc/predicates.md (compact_hreg_operand): New predicate.
20997 2017-04-25 Richard Biener <rguenther@suse.de>
20999 PR middle-end/80509
21000 * passes.c (pass_manager::pass_manager): Initialize
21001 m_name_to_pass_map.
21003 2017-04-25 Richard Biener <rguenther@suse.de>
21005 PR tree-optimization/79201
21006 * tree-ssa-sink.c (statement_sink_location): Handle calls.
21008 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21011 * config/s390/vector.md: Split MEM->GPR vector moves for
21012 non-s_operand addresses.
21014 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21017 * config/s390/predicates.md (reload_const_wide_int_operand): New
21019 * config/s390/s390.md ("movti"): Remove d/P alternative.
21020 ("movti_bigconst"): New pattern definition.
21022 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21025 * s390-protos.h (s390_expand_cs_hqi): Removed.
21026 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
21027 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
21028 modes as well as CCZ1mode and CCZmode.
21029 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
21030 signature of s390_emit_compare_and_swap.
21031 (s390_expand_cs_hqi): Likewise, make static.
21032 (s390_expand_cs_tdsi): Generate an explicit compare before trying
21033 compare-and-swap, in some cases.
21034 (s390_expand_cs): Wrapper function.
21035 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
21037 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
21038 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
21039 patterns for small and large integers. Forbid symref memory operands.
21040 Move expander to s390.c. Require cc register.
21041 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
21042 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
21043 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
21044 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
21045 symref memory operands. Remove CC mode and call s390_match_ccmode
21047 ("atomic_exchange<mode>"): Allow and implement all integer modes.
21049 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21051 * config/s390/s390.md (define_peephole2): New peephole to help
21052 combining the load-and-test pattern with volatile memory.
21054 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21056 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
21057 with CCZmode for TARGET_Z196.
21059 2017-04-25 Jakub Jelinek <jakub@redhat.com>
21061 PR rtl-optimization/80501
21062 * combine.c (make_compound_operation_int): Set subreg_code to SET
21063 even for AND with mask of the sign bit of mode.
21065 PR rtl-optimization/80500
21066 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
21067 sum's initial value.
21069 2017-04-25 Julian Brown <julian@codesourcery.com>
21070 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21072 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
21074 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
21076 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
21078 2017-04-25 Julian Brown <julian@codesourcery.com>
21079 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21081 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
21082 (thunderx2t99_sha): New Reservation.
21084 2017-04-25 Julian Brown <julian@codesourcery.com>
21085 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21087 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
21088 type for 1-element load.
21090 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
21092 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
21094 2017-04-24 Martin Jambor <mjambor@suse.cz>
21096 PR tree-optimization/80293
21097 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
21098 char arrays not totally scalarizable if it is false.
21099 (analyze_all_variable_accesses): Pass correct value in the new
21100 parameter. Add a statistics counter.
21102 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
21104 PR middle-end/79931
21105 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
21107 2017-04-24 Richard Biener <rguenther@suse.de>
21109 PR tree-optimization/80494
21110 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
21111 out for complex types.
21113 2017-04-24 Richard Biener <rguenther@suse.de>
21115 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
21116 * tree-ssa-sccvn.c (print_scc): Print SCC size.
21117 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
21118 (DFS): Adjust and never fail.
21119 (sccvn_dom_walker::fail): Remove.
21120 (sccvn_dom_walker::before_dom_children): Adjust.
21121 (run_scc_vn): Likewise and never fail.
21122 * tree-ssa-pre.c (pass_pre::execute): Adjust.
21123 (pass_fre::execute): Likewise.
21125 2017-04-24 Richard Biener <rguenther@suse.de>
21127 PR tree-optimization/79725
21128 * tree-ssa-sink.c (statement_sink_location): Return whether
21129 failure reason was zero uses. Move that check later.
21130 (sink_code_in_bb): Deal with zero uses by removing the stmt
21133 2017-04-24 Richard Biener <rguenther@suse.de>
21136 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
21137 pointer-based references.
21139 2017-04-24 Richard Biener <rguenther@suse.de>
21142 * pass_manager.h (pass_manager::operator new): Remove.
21143 (pass_manager::operator delete): Likewise.
21144 * passes.c (pass_manager::operator new): Remove.
21145 (pass_manager::operator delete): Likewise.
21146 (pass_manager::pass_manager): Zero individual pass members.
21148 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
21151 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
21152 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
21153 Check "XEXP (src, 1)" operand here.
21154 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
21155 Check "XEXP (src, 1)" operand here.
21156 (dimode_scalar_chain::make_vector_copies): Detect count register
21157 of a shift instruction. Zero extend count register from QImode
21158 to DImode to satisfy vector shift pattern count operand predicate.
21159 Substitute vector shift count operand with a DImode copy.
21160 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
21163 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
21165 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
21166 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21167 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
21168 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21169 (UNSPEC_NOREX_MEM): Remove definition.
21171 2017-04-21 Richard Biener <rguenther@suse.de>
21173 PR tree-optimization/79547
21174 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
21175 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
21176 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
21177 without any constraints.
21179 2017-04-21 Richard Biener <rguenther@suse.de>
21181 PR tree-optimization/78847
21182 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
21184 2017-04-21 Richard Biener <rguenther@suse.de>
21186 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
21187 (build_distinct_type_copy): Likewise.
21188 (build_variant_type_copy): Likewise.
21189 * tree.c (build_qualified_type): Pass down mem-stat info.
21190 (build_distinct_type_copy): Likewise.
21191 (build_variant_type_copy): Likewise.
21193 2017-04-21 Richard Biener <rguenther@suse.de>
21195 PR tree-optimization/80237
21196 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
21198 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
21199 for a simplified result.
21201 2016-04-21 Richard Biener <rguenther@suse.de>
21203 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
21204 sth as strict as a simple_iv but a chrec without symbols and an
21205 operand defined in the loop we are peeling (and not some subloop).
21206 (propagate_constants_for_unrolling): Propagate all constants.
21208 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21211 * config/i386/i386.c (print_reg): Remove assert for disalowed
21212 regno values, call output_operand_lossage instead.
21214 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21217 * config/i386/constraints.md (Yc): New register constraint.
21218 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
21219 Use Yc constraint for alternative 2 of operand 0. Remove
21220 preferred_for_speed attribute.
21222 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
21224 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
21225 lastprivate clauses in SIMT case.
21227 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
21229 * doc/invoke.texi (-Wextra-semi): Document new warning option.
21231 2017-04-20 Richard Biener <rguenther@suse.de>
21233 PR tree-optimization/57796
21234 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
21235 as N scalar stores.
21236 (vect_model_load_cost): Cost gathers as N scalar loads.
21238 2017-04-20 Richard Biener <rguenther@suse.de>
21240 * ggc-page.c (ggc_allocated_p): Rename to ...
21241 (safe_lookup_page_table_entry): ... this and return the lookup
21243 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
21245 2017-04-20 Richard Biener <rguenther@suse.de>
21247 PR tree-optimization/80453
21248 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
21249 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
21250 from the conditions.
21251 (vn_phi_eq): Pass them down.
21252 (vn_phi_lookup): Record them.
21253 (vn_phi_insert): Likewise.
21255 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
21257 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
21258 uninitialized variable warning to avoid buffer overrun.
21260 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
21263 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
21264 is suppressed for '{ 0 }' in C.
21266 2017-04-20 Jakub Jelinek <jakub@redhat.com>
21268 * BASE-VER: Set to 8.0.0.
21270 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
21272 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
21273 priority .init_array and .fini_array section with SECTION_NOTYPE
21276 2017-04-20 Jakub Jelinek <jakub@redhat.com>
21278 PR middle-end/80423
21279 * tree.h (build_array_type): Add typeless_storage default argument.
21280 * tree.c (type_cache_hasher::equal): Also compare
21281 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
21282 (build_array_type): Add typeless_storage argument, set
21283 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
21285 (build_nonshared_array_type): Adjust build_array_type_1 caller.
21286 (build_array_type): Likewise. Add typeless_storage argument.
21288 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21289 Jakub Jelinek <jakub@redhat.com>
21291 PR tree-optimization/80426
21292 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
21293 operation on symbolic operands, also compute the overflow for the
21294 invariant part when the operation degenerates into a negation.
21296 2017-04-19 Jakub Jelinek <jakub@redhat.com>
21299 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
21300 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
21303 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
21305 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
21308 * config/avr/avr.c (tree.h): Include it.
21309 (cgraph.h): Include it.
21310 (avr_encode_section_info): Don't warn for uninitialized progmem
21311 variable if it's just an alias.
21313 2017-04-19 Richard Biener <rguenther@suse.de>
21316 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
21317 when needed by AutoPGO.
21319 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
21322 * doc/lto.texi: Remove an extra 'that'.
21324 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
21326 PR rtl-optimization/80429
21327 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
21328 are only used in debug insns.
21330 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21331 Vladimir Makarov <vmakarov@redhat.com>
21333 * config/sparc/predicates.md (input_operand): Add comment. Return
21334 true for any memory operand when LRA is in progress.
21335 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
21337 2017-04-18 Jeff Law <law@redhat.com>
21340 * mips.md ({return,simple_return}_internal): Do not overwrite
21343 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21345 PR tree-optimization/80443
21346 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
21347 instead of adding 1, subtract -1 and similarly instead of subtracting
21350 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
21352 PR rtl-optimization/80357
21353 * haifa-sched.c (tmp_bitmap): New variable.
21354 (model_recompute): Handle duplicate use records.
21355 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
21356 (free_global_sched_pressure_data): Free it.
21358 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
21361 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21362 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21363 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21364 instead of SYSTEM_HEADER_DIR.
21366 2017-04-18 Jeff Law <law@redhat.com>
21368 PR middle-end/80422
21369 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
21370 predecessors after walking up the insn chain.
21372 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21375 * dwarf2out.c (modified_type_die): Try harder not to emit internal
21376 sizetype type into debug info.
21378 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21381 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
21382 unneeded test for TARGET_UPPER_REGS_SF.
21383 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
21385 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21388 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
21389 instead of gsi_after_labels.
21391 2017-04-18 Jeff Law <law@redhat.com>
21393 * regcprop.c (maybe_mode_change): Avoid creating copies of the
21397 2017-04-13 Jeff Law <law@redhat.com>
21398 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21399 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21401 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
21404 * config/avr/avr.c (intl.h): Include it.
21405 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
21407 2017-04-18 Martin Liska <mliska@suse.cz>
21409 PR gcov-profile/78783
21410 * gcov-tool.c (gcov_output_files): Validate that destination
21411 file is either removed by the tool or by a user.
21413 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21414 Guy Benyei <guybe@mellanox.com>
21416 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
21417 block, and do not negate it, the stored id is already negative.
21419 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21421 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
21423 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21426 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
21427 masks of options that should be turned off if the VSX vector
21428 options are turned off.
21429 (OTHER_P8_VECTOR_MASKS): Likewise.
21430 (OTHER_VSX_VECTOR_MASKS): Likewise.
21431 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
21432 rs6000_disable_incompatible_switches to validate no type switches
21434 (rs6000_incompatible_switch): New function to disallow turning on
21435 other vector options if -mno-vsx, -mno-power8-vector, or
21436 -mno-power9-vector are specified.
21438 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21440 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
21442 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21444 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
21445 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
21446 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
21447 (ARG_POINTER_CFA_OFFSET): Likewise.
21449 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21451 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
21452 conditions to take advantage of various optimizations.
21454 2017-04-13 Jeff Law <law@redhat.com>
21456 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21457 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21458 (zero_extendsidi2_dext): Likewise.
21460 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21463 * fold-const.c (fold_ternary_loc): Revert
21464 use op0 instead of fold_convert_loc (loc, type, arg0) part of
21467 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
21469 PR rtl-optimization/80343
21470 * lra-remat.c (update_scratch_ops): Assign original hard reg to
21471 new scratch pseudo.
21473 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
21476 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
21477 to ubsan_encode_value.
21479 2017-04-13 Jeff Law <law@redhat.com>
21481 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
21482 appearing in DEBUG_INSNs.
21484 2017-04-13 Martin Liska <mliska@suse.cz>
21486 PR gcov-profile/80413
21487 * gcov-io.c (gcov_write_string): Copy to buffer just when
21488 allocated size is greater than zero.
21490 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21493 * dwarf2out.c (decls_for_scope): Ignore declarations of
21494 current_function_decl in BLOCK_NONLOCALIZED_VARS.
21496 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
21499 * ipa-visibility.c (non_local_p): Fix typos.
21500 (localize_node): When localizing symbol in same comdat group,
21501 dissolve the group only when we know external symbols are going
21503 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
21505 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21507 PR tree-optimization/79390
21508 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
21509 order does not result in usable sequence, retry with reversed operand
21515 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
21516 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
21517 op0 instead of fold_convert_loc (loc, type, arg0).
21519 2017-04-12 Jeff Law <law@redhat.com>
21521 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
21522 has a delay slot in the generated code.
21524 * config/cris/cris.md (cris_preferred_reload_class): Return
21525 GENNONACR_REGS rather than GENERAL_REGS.
21527 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21530 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
21531 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
21532 signedness of the result type.
21534 2017-04-12 Richard Biener <rguenther@suse.de>
21535 Jeff Law <law@redhat.com>
21537 PR tree-optimization/80359
21538 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
21539 trim stores to TARGET_MEM_REFs.
21541 2017-04-12 Richard Biener <rguenther@suse.de>
21543 PR tree-optimization/79390
21544 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
21545 threading case even more.
21547 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
21550 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
21551 for quad_address_p for TImode, instead of just not indexed_address.
21553 2017-04-12 Richard Biener <rguenther@suse.de>
21554 Bernd Edlinger <bernd.edlinger@hotmail.de>
21556 PR middle-end/79671
21557 * alias.c (component_uses_parent_alias_set_from): Handle
21558 TYPE_TYPELESS_STORAGE.
21559 (get_alias_set): Likewise.
21560 * tree-core.h (tree_type_common): Add typeless_storage flag.
21561 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
21562 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
21563 for types containing members with TYPE_TYPELESS_STORAGE.
21564 (place_field): Likewise.
21565 (layout_type): Likewise for ARRAY_TYPE.
21566 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
21567 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
21568 TYPE_TYPELESS_STORAGE.
21569 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
21571 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21574 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21575 first argument to type.
21577 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21581 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
21582 CONST0_RTX (mode) rather than const0_rtx where appropriate.
21583 (rs6000_expand_binop_builtin): Likewise.
21584 (rs6000_expand_ternop_builtin): Likewise; also add missing
21585 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
21586 vshasigma built-ins.
21587 * doc/extend.texi: Document that vec_xxpermdi's third argument
21588 must be a constant.
21590 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
21592 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
21593 Use shift_const cost parameter when calculating gain of STV shifts.
21595 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
21597 PR rtl-optimization/70478
21598 * lra-constraints.c (process_alt_operands): Check memory for
21599 disfavoring memory insn operand.
21601 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21603 PR middle-end/80100
21604 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
21605 left shift in unsigned HOST_WIDE_INT type.
21607 PR rtl-optimization/80385
21608 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
21609 (not (neg X)) into (plus X -1) for complex or non-integral modes.
21612 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
21613 if they have any depend clauses.
21615 2017-04-11 Martin Liska <mliska@suse.cz>
21618 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
21619 * ipa-split.c (split_function): Create a local comdat symbol
21620 if caller is in a comdat group.
21622 2017-04-11 Martin Liska <mliska@suse.cz>
21625 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
21628 2017-04-11 Martin Sebor <msebor@redhat.com>
21630 PR middle-end/80364
21631 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
21632 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
21634 (directive::set_width, directive::set_precision, format_character):
21636 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
21639 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
21642 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
21643 conflict, set target->arch_name instead of target->cpu_name.
21645 2017-04-11 Richard Biener <rguenther@suse.de>
21647 PR tree-optimization/80374
21648 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
21649 build_zero_cst, remove fold_convertible_p check again.
21651 2017-04-11 Martin Liska <mliska@suse.cz>
21654 * ubsan.c (instrument_object_size): Do not instrument register
21657 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21660 * config/i386/i386-builtin-types.def
21661 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
21662 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
21663 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
21664 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
21665 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
21666 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
21667 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
21668 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
21669 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
21670 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
21671 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
21672 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
21673 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
21674 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
21675 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
21676 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
21677 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
21678 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
21679 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
21680 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
21681 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
21682 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
21683 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
21684 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
21685 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
21686 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
21687 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
21688 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
21689 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
21690 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
21691 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
21692 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
21693 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
21694 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
21695 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
21696 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
21697 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
21698 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
21699 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
21700 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
21701 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
21702 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
21703 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
21704 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
21705 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
21706 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
21708 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
21709 flag to second_arg_count, handle 4 argument function type _COUNT
21710 aliases, handle second_arg_count on second argument rather than last.
21712 2017-04-10 Jeff Law <law@redhat.com>
21714 PR tree-optimization/80374
21715 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
21716 record anything if we can not convert integer_zero_node to the
21719 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21722 * config/rs6000/rs6000.c (rs6000_option_override_internal):
21723 Enhance special handling given to the TARGET_P9_MINMAX option in
21724 relation to certain other options.
21726 2017-04-10 Bin Cheng <bin.cheng@arm.com>
21728 PR tree-optimization/80153
21729 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
21730 remove POINTER_PLUS_EXPR's base part directly, rather than through
21733 2017-04-10 Richard Biener <rguenther@suse.de>
21734 Bin Cheng <bin.cheng@arm.com>
21736 PR tree-optimization/80153
21737 * tree-affine.c (aff_combination_to_tree): Get base pointer from
21738 the first element of pointer type aff_tree. Build result expr in
21740 (add_elt_to_tree): Convert to type unconditionally. Remove other
21741 fold_convert calls.
21742 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
21743 (rewrite_use_nonlinear_expr): Check invariant using iv information.
21745 2017-04-10 Richard Biener <rguenther@suse.de>
21747 * tree-ssa-structalias.c (find_func_aliases): Properly handle
21750 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
21752 PR rtl-optimization/70478
21753 * lra-constraints.c (curr_small_class_check): New.
21754 (update_and_check_small_class_inputs): New.
21755 (process_alt_operands): Update curr_small_class_check. Disfavor
21756 alternative insn memory operands. Check available regs for small
21759 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
21762 * config/mips/mips.opt (-mvirt): Update description.
21763 * doc/invoke.texi (-mvirt): Likewise.
21765 2017-04-10 Richard Biener <rguenther@suse.de>
21767 PR middle-end/80362
21768 * fold-const.c (fold_binary_loc): Look at unstripped ops when
21769 looking for NEGATE_EXPR in -A / -B to A / B folding.
21771 2017-04-10 Martin Liska <mliska@suse.cz>
21773 PR gcov-profile/80224
21774 * gcov.c (print_usage): Fix usage string.
21775 (get_gcov_intermediate_filename): Remove.
21776 (output_gcov_file): Use both for normal and intermediate format.
21777 (generate_results): Do not initialize special file for
21778 intermediate format.
21780 2017-04-10 Richard Biener <rguenther@suse.de>
21782 PR tree-optimization/80304
21783 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
21786 2017-04-10 Nathan Sidwell <nathan@acm.org>
21789 * config/rs6000/rs6000.c (rs6000_vector_type): New.
21790 (rs6000_init_builtins): Use it.
21792 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21794 * config/arm/arm.md (<mrc>): Add mode to SET source.
21795 (<mrrc>): Likewise.
21797 2017-04-10 Richard Biener <rguenther@suse.de>
21799 PR middle-end/80344
21800 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
21802 2017-04-10 Jakub Jelinek <jakub@redhat.com>
21805 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
21806 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
21807 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
21808 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
21809 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
21810 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
21811 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
21812 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
21813 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
21814 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
21815 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
21816 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
21817 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
21818 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
21819 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
21820 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
21821 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
21822 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
21823 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
21824 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
21825 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
21826 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
21827 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
21829 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
21831 PR rtl-optimization/70478
21832 * lra-constraints.c: Reverse the last patch.
21834 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
21836 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
21837 Add comment for WCHAR_T.
21839 2017-04-08 Martin Liska <mliska@suse.cz>
21842 2017-04-07 Martin Liska <mliska@suse.cz>
21845 * ipa-split.c (split_function): Add function part to a same comdat
21848 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21851 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
21853 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
21855 * rs6000/rs6000.c (vec_load_pendulum): Rename...
21856 (vec_pairing): ...to this.
21857 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
21858 (rs6000_sched_init): Adjust for name change.
21859 (struct rs6000_sched_context): Likewise.
21860 (rs6000_init_sched_context): Likewise.
21861 (rs6000_set_sched_context): Likewise.
21863 2017-04-07 Jakub Jelinek <jakub@redhat.com>
21869 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
21871 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
21872 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
21873 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
21875 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
21877 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
21879 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21881 PR rtl-optimization/70703
21882 * ira-color.c (update_conflict_hard_regno_costs): Use
21883 int64_t instead of HOST_WIDE_INT.
21885 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
21887 PR rtl-optimization/70478
21888 * lra-constraints.c (process_alt_operands): Disfavor alternative
21889 insn memory operands.
21891 2017-04-07 Jeff Law <law@redhat.com>
21893 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
21894 CALL and NOTE_INSN_CALL_ARG_LOCATION.
21896 2017-04-07 Martin Liska <mliska@suse.cz>
21899 * config/aarch64/aarch64.c (aarch64_process_target_attr):
21900 Show error message instead of an ICE.
21902 2017-04-07 Martin Liska <mliska@suse.cz>
21905 * ipa-split.c (split_function): Add function part to a same comdat
21908 2017-04-07 Richard Biener <rguenther@suse.de>
21910 PR middle-end/80341
21911 * tree.c (get_unwidened): Also handle ! for_type case for
21913 * convert.c (do_narrow): Split out from ...
21914 (convert_to_integer_1): ... here. Do not pass final truncation
21915 type to get_unwidened for TRUNC_DIV_EXPR.
21917 2017-04-07 Richard Biener <rguenther@suse.de>
21919 * tree-affine.c (wide_int_ext_for_comb): Take type rather
21921 (aff_combination_const): Adjust.
21922 (aff_combination_scale): Likewise.
21923 (aff_combination_add_elt): Likewise.
21924 (aff_combination_add_cst): Likewise.
21925 (aff_combination_convert): Likewise.
21926 (add_elt_to_tree): Likewise. Remove unused argument.
21927 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
21929 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
21931 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
21933 * config/arm/arm.c (arm_default_short_enums): Use
21934 ARM_DEFAULT_SHORT_ENUMS.
21935 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
21937 2017-04-06 Jakub Jelinek <jakub@redhat.com>
21940 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
21941 members with redundant out-of-class redeclaration.
21943 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21946 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
21947 * config/i386/i386.md (*zero_extendsidi2):
21948 Add (?*x,*x) and (?*v,*v) alternatives.
21950 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
21953 * config/i386/i386.c (ix86_expand_builtin)
21954 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
21955 mode from insn data. Convert operands to insn operand mode.
21956 Copy operands that don't satisfy insn predicate to a register.
21958 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
21960 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
21963 2017-04-06 Richard Biener <rguenther@suse.de>
21965 PR tree-optimization/80334
21966 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
21967 preserve alignment of accesses.
21969 2017-04-06 Richard Biener <rguenther@suse.de>
21971 PR tree-optimization/80262
21972 * tree-sra.c (build_ref_for_offset): Preserve address-space
21974 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
21975 Drop useless address-space information on MEM_REF offsets.
21977 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
21979 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
21981 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21983 PR rtl-optimization/70703
21984 * ira-color.c (update_conflict_hard_regno_costs): Use
21985 HOST_WIDE_INT instead of long.
21987 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
21990 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
21991 not defined for x86_64 target. Add -mmmx target option when __SSE2__
21993 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
21994 for x86_64 target. Handle -m3dnowa option.
21996 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
21998 PR rtl-optimization/70703
21999 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
22000 (update_conflict_hard_regno_costs): Use long instead of unsigned
22001 arithmetic for cost calculation.
22003 2017-04-05 Jakub Jelinek <jakub@redhat.com>
22004 Bernd Edlinger <bernd.edlinger@hotmail.de>
22007 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
22010 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
22013 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
22014 ptr_mode with Pmode throughout.
22015 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
22016 into probe_stack_range and use DImode.
22018 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
22021 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
22022 call_eh_return is true.
22024 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22026 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
22027 Initialize last_match_fntype_index.
22029 2017-04-05 Jakub Jelinek <jakub@redhat.com>
22032 * tree-nvr.c: Include internal-fn.h.
22033 (pass_return_slot::execute): Ignore internal calls without
22036 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22037 Richard Biener <rguenther@suse.de>
22040 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
22041 captures used multiple times, except for the last use.
22042 * generic-match-head.c: Include gimplify.h.
22044 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22046 PR tree-optimization/79390
22047 * target.h (struct noce_if_info): Declare.
22048 * targhooks.h (default_noce_conversion_profitable_p): Declare.
22049 * target.def (noce_conversion_profitable_p): New target hook.
22050 * ifcvt.h (struct noce_if_info): New type, moved from ...
22051 * ifcvt.c (struct noce_if_info): ... here.
22052 (noce_conversion_profitable_p): Renamed to ...
22053 (default_noce_conversion_profitable_p): ... this. No longer
22055 (noce_try_store_flag_constants, noce_try_addcc,
22056 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
22057 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
22058 instead of noce_conversion_profitable_p.
22059 * config/i386/i386.c: Include ifcvt.h.
22060 (ix86_option_override_internal): Don't override
22061 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
22062 (ix86_noce_conversion_profitable_p): New function.
22063 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
22064 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
22065 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
22066 * doc/tm.texi: Regenerated.
22068 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22070 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
22073 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22076 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
22077 instructions for small multiply cores.
22079 2017-04-04 Jeff Law <law@redhat.com>
22081 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
22083 (mips_expand_vec_perm_const): Initialize elements in orig_perm
22084 that are not set by the loop over the elements.
22086 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22089 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
22090 int mode, convert_modes it to mode as unsigned, otherwise use
22091 lowpart_subreg to mode rather than SImode.
22092 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
22093 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
22094 Use DImode instead of SImode for the shift count operand.
22095 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
22098 2017-04-04 Richard Biener <rguenther@suse.de>
22100 PR middle-end/80281
22101 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
22102 arithmetic done for the negate or the plus. Simplify.
22103 (A - (-B) -> A + B): Likewise.
22104 * fold-const.c (split_tree): Make sure to not negate pointers.
22106 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
22108 PR rtl-optimization/60818
22109 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
22110 a compare of comparisons with the thing compared if this results
22111 in a different machine mode.
22113 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
22115 * alias.c (base_alias_check): Fix typo in comment.
22116 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
22117 * cgraphunit.c (symbol_table::compile): Likewise.
22118 * collect2.c (maybe_run_lto_and_relink): Likewise.
22119 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
22120 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
22121 * config/avr/avr.c (avr_map_op_t): Likewise.
22122 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
22123 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
22124 * config/epiphany/epiphany.md (movcc): Likewise.
22125 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
22126 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
22128 * config/mips/mips.c (mips_save_restore_reg): Likewise.
22129 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
22130 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
22131 * config/sh/sh.c (sh_rtx_costs): Likewise.
22132 * fold-const.c (fold_truth_andor): Likewise.
22133 * genautomata.c (collapse_flag): Likewise.
22134 * gengtype.h (struct type::u::s): Likewise.
22135 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
22136 * input.c (FORMAT_AMOUNT): Likewise.
22137 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
22138 (known_aggs_to_agg_replacement_list): Likewise.
22139 * ipa-inline-analysis.c: Likewise.
22140 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
22141 * ipa-polymorphic-call.c
22142 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
22143 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
22144 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
22146 * modulo-sched.c (apply_reg_moves): Likewise.
22147 * omp-expand.c (build_omp_regions_1): Likewise.
22148 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
22149 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
22150 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
22151 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
22152 * value-prof.c: Likewise.
22153 * var-tracking.c (val_reset): Likewise.
22155 2017-04-03 Richard Biener <rguenther@suse.de>
22157 PR tree-optimization/80275
22158 * fold-const.c (split_address_to_core_and_offset): Handle
22161 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
22163 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
22164 descriptors is at least equal to that of functions.
22166 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22168 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
22170 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22173 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
22174 (mov<IMOD4:mode>): New expander.
22175 (*mov<IMOD4:mode>_internal): New insn and split pattern.
22177 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
22179 PR rtl-optimization/79405
22180 * fwprop.c (propagations_left): New variable.
22181 (forward_propagate_into): Decrement it.
22182 (fwprop_init): Initialize it.
22183 (fw_prop): If the variable has reached zero, stop propagating.
22184 (fwprop_addr): Ditto.
22186 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22189 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
22190 a FUNCTION_DECL, pass it as decl instead of origin to
22193 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
22195 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
22198 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
22201 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
22202 TARGET_VSX_SMALL_INTEGER.
22204 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22206 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22207 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
22209 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
22211 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
22212 extraction from odd-numbered MSA register.
22214 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22216 PR middle-end/80173
22217 * expmed.c (store_bit_field_1): Don't attempt to create
22218 a word subreg out of hard registers wider than word if they
22219 have HARD_REGNO_NREGS of 1 for their mode.
22221 PR middle-end/80163
22222 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
22223 conversions to integer types wider than word and pointer.
22226 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
22227 (rtx_equal_for_cselib_p): Pass 0 to it.
22228 * cselib.c (cselib_hasher::equal): Likewise.
22229 (rtx_equal_for_cselib_1): Add depth argument. If depth
22230 is 128, don't look up VALUE locs and punt. Increment
22231 depth in recursive calls when walking VALUE locs.
22233 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
22235 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
22236 (make_gcov_file_name): Use the canonical path name for generating
22238 (read_line): Fix handling of files with ascii null bytes.
22240 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
22242 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
22243 to initialise a vector register instead
22244 of using a const_int.
22246 2017-03-30 Jakub Jelinek <jakub@redhat.com>
22248 PR translation/80189
22249 * gimplify.c (omp_default_clause): Use %qs instead of %s in
22250 diagnostic messages.
22252 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
22255 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
22256 (dfp_diex_<mode>): Update mode of operand 1.
22257 * doc/extend.texi (dxex, dxexq): Document change to return type.
22258 (diex, diexq): Document change to argument type.
22260 2017-03-30 Martin Jambor <mjambor@suse.cz>
22263 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
22264 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
22265 it reflects the signature changes performed at the callee side.
22266 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
22267 to cgraph_build_function_type_skip_args.
22268 (build_function_decl_skip_args): Adjust call to the above function.
22270 2017-03-30 Jakub Jelinek <jakub@redhat.com>
22273 * config/i386/sse.md
22274 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
22275 register as dest whenever it is a MEM not rtx_equal_p to the
22276 corresponding dup operand, and when forcing into reg move the
22277 reg into the memory afterwards.
22278 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
22279 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
22280 for the force_reg mode.
22281 (avx512vl_vextractf128<mode>): Use register as dest either
22282 always when a MEM, or when it is a MEM not rtx_equal_p to the
22283 corresponding dup operand, or even not when it is a CONST_VECTOR
22284 depending on the mode and lo vs. hi.
22285 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
22287 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
22288 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
22289 Likewise. Require that operands[2] is even.
22290 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
22291 Remove extraneous parens. Require that operands[2] is a multiple
22293 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
22294 operands[0] is a MEM if <mask_applied>, the predicates/constraints
22295 disallow memory then.
22297 2017-03-30 Richard Biener <rguenther@suse.de>
22299 PR tree-optimization/77498
22300 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
22301 to non-constants over backedges.
22303 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
22305 PR rtl-optimization/80233
22306 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
22307 as last_combined_insn. Do not test for BARRIER_P separately.
22309 2017-03-29 Andreas Schwab <schwab@suse.de>
22312 * calls.c (prepare_call_address): Convert funexp to Pmode before
22313 copying to temp reg.
22315 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22317 PR tree-optimization/80158
22318 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
22319 Handle possible future case of more than one alternate
22321 (replace_rhs_if_not_dup): Likewise.
22322 (replace_one_candidate): Likewise.
22324 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
22326 PR rtl-optimization/80193
22327 * ira.c (ira): Do not check allocation for LRA.
22329 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
22331 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
22332 (nvptx_output_simt_exit): Declare.
22333 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
22334 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
22335 (init_softstack_frame): Move initialization of crtl->is_leaf to...
22336 (nvptx_declare_function_name): ...here. Emit declaration of local
22337 memory space buffer for omp_simt_enter insn.
22338 (nvptx_output_unisimt_switch): New.
22339 (nvptx_output_softstack_switch): New.
22340 (nvptx_output_simt_enter): New.
22341 (nvptx_output_simt_exit): New.
22342 * config/nvptx/nvptx.h (struct machine_function): New fields
22343 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
22344 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
22345 (UNSPECV_SIMT_EXIT): Ditto.
22346 (omp_simt_enter_insn): New insn.
22347 (omp_simt_enter): New expansion.
22348 (omp_simt_exit): New insn.
22349 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
22351 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
22352 (expand_GOMP_SIMT_ENTER_ALLOC): New.
22353 (expand_GOMP_SIMT_EXIT): New.
22354 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
22355 (GOMP_SIMT_ENTER_ALLOC): Ditto.
22356 (GOMP_SIMT_EXIT): Ditto.
22357 * target-insns.def (omp_simt_enter): New insn.
22358 (omp_simt_exit): Ditto.
22359 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
22361 (lower_rec_simd_input_clauses): Implement SIMT privatization.
22362 (lower_rec_input_clauses): Likewise.
22363 (lower_lastprivate_clauses): Handle SIMT privatization.
22365 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
22366 (ompdevlow_adjust_simt_enter): New.
22367 (find_simtpriv_var_op): New.
22368 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
22369 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
22371 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
22372 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
22373 (copy_decl_for_dup_finish): Ditto.
22375 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
22377 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
22380 * config/i386/i386.c (ix86_option_override_internal): Always
22381 allow -mpreferred-stack-boundary=3 for 64-bit targets.
22383 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22385 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
22387 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22389 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
22390 mark new edge's irreducible flag accordign to it.
22391 (vect_do_peeling): Check loop preheader edge's irreducible flag
22392 and pass it to function slpeel_add_loop_guard.
22394 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
22396 PR tree-optimization/80218
22397 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
22398 Update block frequencies and counts.
22400 2017-03-28 Richard Biener <rguenther@suse.de>
22402 PR tree-optimization/78644
22403 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
22404 of a simplification result we may not use it at all.
22406 2017-03-28 Richard Biener <rguenther@suse.de>
22409 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
22410 without arguments, generate default definition of a SSA name.
22412 2017-03-28 Richard Biener <rguenther@suse.de>
22414 PR middle-end/80222
22415 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
22416 TYPE_REF_CAN_ALIAS_ALL references.
22417 * fold-const.c (fold_indirect_ref_1): Likewise.
22419 2017-03-28 Martin Liska <mliska@suse.cz>
22422 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
22423 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
22425 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22426 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
22428 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
22429 (EXTRA_SPECS): Define.
22430 (SUBTARGET_EXTRA_SPECS): Likewise.
22431 (SUBTARGET_CPP_SPEC): Likewise.
22432 * config/arc/elf.h (EXTRA_SPECS): Renamed to
22433 SUBTARGET_EXTRA_SPECS.
22434 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
22436 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22438 * config/arc/simdext.md (vst64_insn): Update pattern.
22439 (vld32wh_insn): Likewise.
22440 (vld32wl_insn): Likewise.
22441 (vld64_insn): Likewise.
22442 (vld32_insn): Likewise.
22444 2017-03-28 Marek Polacek <polacek@redhat.com>
22447 * fold-const.c (fold_comparison): Use protected_set_expr_location
22448 instead of SET_EXPR_LOCATION.
22450 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
22452 * tree.c (add_expr): Avoid name lookup warning.
22454 2017-03-27 Jeff Law <law@redhat.com>
22456 PR tree-optimization/80216
22457 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
22458 function name. Limit recursion depth.
22459 (record_temporary_equivalences): Corresponding changes.
22461 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
22463 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
22466 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22469 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
22471 * cfgcleanup.c (reg_note_cfa_p): New array.
22472 (insns_have_identical_cfa_notes): New function.
22473 (old_insns_match_p): Don't cross-jump in between /f
22474 and non-/f instructions. If both i1 and i2 are frame related,
22475 verify all CFA notes, their order and content.
22477 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
22480 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
22481 HImode and SImode with zero extend to DImode to one insn.
22482 (bswap<mode>2_extenddi): Likewise.
22483 (bswapsi2_extenddi): Likewise.
22484 (bswaphi2_extendsi): Likewise.
22485 (bswaphi2): Combine bswap HImode and SImode into one insn.
22486 Separate memory insns from swapping register.
22487 (bswapsi2): Likewise.
22488 (bswap<mode>2): Likewise.
22489 (bswaphi2_internal): Delete, no longer used.
22490 (bswapsi2_internal): Likewise.
22491 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
22492 store, and gpr<-gpr swap insns.
22493 (bswap<mode>2_store): Likewise.
22494 (bswaphi2_reg): Register only splitter, combine with the splitter.
22495 (bswaphi2 splitter): Likewise.
22496 (bswapsi2_reg): Likewise.
22497 (bswapsi2 splitter): Likewise.
22498 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
22499 the insns into load, store, and register/register insns.
22500 (bswapdi2_ldbrx): Likewise.
22501 (bswapdi2_load): Likewise.
22502 (bswapdi2_store): Likewise.
22503 (bswapdi2_reg): Likewise.
22505 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
22507 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
22508 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
22510 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
22513 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
22515 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22516 special handling for target option conflicts between dform
22517 options (-mpower9-dform, -mpower9-dform-vector,
22518 -mpower9-dform-scalar) and -mno-direct-move.
22520 2017-03-27 Richard Biener <rguenther@suse.de>
22522 PR tree-optimization/80181
22523 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
22525 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22527 * config/arc/predicates.md (move_double_src_operand): Replace the
22528 call to move_double_src_operand with a call to address_operand.
22530 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22532 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
22533 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
22534 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
22536 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22538 * config/arc/predicates.md (long_immediate_loadstore_operand):
22539 Consider scaled addresses cases.
22541 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22543 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
22544 restored when in interrupt.
22545 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
22546 doesn't have delay slot.
22548 2017-03-27 Richard Biener <rguenther@suse.de>
22551 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
22552 inlined thunk clones.
22554 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22557 * asan.c (instrument_derefs): Copy over last operand from
22558 original COMPONENT_REF to the new COMPONENT_REF with
22559 DECL_BIT_FIELD_REPRESENTATIVE.
22560 * ubsan.c (instrument_object_size): Likewise.
22562 2017-03-27 Richard Biener <rguenther@suse.de>
22564 PR tree-optimization/80170
22565 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
22566 sure DR/SCEV didnt fold in constants we do not see when looking
22567 at the reference base alignment.
22569 2017-03-27 Richard Biener <rguenther@suse.de>
22571 PR middle-end/80171
22572 * gimple-fold.c (fold_ctor_reference): Properly guard against
22573 NULL return value from canonicalize_constructor_val.
22575 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
22578 * config/i386/i386.c (ix86_expand_builtin)
22579 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
22580 flags reg setting and flags reg using instructions.
22581 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
22582 clobbering instructions to zero extend op2.
22584 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
22586 * doc/install.texi (Configuration) <--with-aix-soname>:
22587 Update link to AIX ld.
22589 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
22591 PR rtl-optimization/80160
22592 PR rtl-optimization/80159
22593 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
22594 reg_alternate_class into account.
22596 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
22599 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
22600 to consider in curr_insn_transform.
22602 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22604 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
22605 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
22606 and emit_mode_inner.
22608 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22610 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
22611 argument to the overloaded builtin variants. Use the new flag to
22612 deprecate certain builtin variants.
22613 * config/s390/s390-builtin-types.def: Add new builtin types.
22614 * config/s390/s390-builtins.h: Support new flags field for
22615 overloaded builtins.
22616 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
22617 (s390_macro_to_expand): Enable vector float data type.
22618 (s390_cpu_cpp_builtins_internal): Indicate support of the new
22619 builtins by incrementing the __VEC__ version number.
22620 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
22622 (s390_resolve_overloaded_builtin): Emit error messages depending
22623 on the builtin flags.
22624 * config/s390/s390.c (s390_expand_builtin): Support additional
22625 flags argument. Change error message to match the messages
22626 emitted in s390-c.c.
22627 * config/s390/s390.md: New UNSPEC_* constants.
22628 (op_type): Add new instruction types.
22629 * config/s390/vecintrin.h: Add new builtins and test data class
22631 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
22632 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
22633 (VEC_INEXACT, VEC_NOINEXACT): New constants.
22634 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
22635 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
22636 ("vec_mergel<mode>"): V_HW -> VEC_HW.
22638 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
22639 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
22640 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
22641 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
22643 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
22644 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
22645 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
22646 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
22648 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
22649 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
22650 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
22651 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
22652 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
22653 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
22654 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
22656 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
22657 ("vec_scatter_element<V_HW_4:mode>_DI")
22658 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
22659 ("vec_fpint<mode>", "vflls")
22660 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
22661 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
22662 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
22663 ("*vec_cmphe<mode>_cc"): ... these.
22665 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
22666 mode constant instead of magic value.
22668 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22670 * config/s390/s390.c (s390_expand_vec_compare): Support other
22671 vector floating point modes than just V2DF.
22672 (s390_expand_vcond): Likewise.
22673 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
22674 (s390_cannot_change_mode_class): Prevent mode changes between TF
22675 and V1TF in vector registers.
22676 * config/s390/s390.md (DF, SF): New mode attributes.
22677 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
22678 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
22679 SFmode support for VRs.
22680 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
22682 (VFT, VF_HW): New mode iterators.
22683 (vw, sdx): New mode attributes.
22684 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
22685 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
22686 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
22687 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
22688 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
22689 also the new vector floating point modes. Renaming to ...
22691 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
22692 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
22693 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
22694 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
22695 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
22696 ("vec_unordered<mode>"): ... these.
22698 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
22699 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
22700 ("*vec_extendv2df"): New insn definitions.
22702 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22704 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
22705 ("mulditi3_2", "*muldi3_sign"): New patterns.
22706 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
22707 rename the pattern definition.
22709 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22711 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
22713 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
22715 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22717 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
22718 instruction if possible.
22719 * config/s390/vector.md (vec_halfnumelts): New mode
22721 ("*vec_vllezlf<mode>"): New pattern.
22723 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22725 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
22726 ("popcountv4si2", "popcountv2di2"): Rename to ...
22727 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
22728 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
22730 ("popcount<mode>2_vxe"): New pattern.
22732 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22734 * common/config/s390/s390-common.c (processor_flags_table): Add
22736 * config.gcc: Add arch12.
22737 * config/s390/driver-native.c (s390_host_detect_local_cpu):
22738 Default to arch12 for unknown CPU model numbers.
22739 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
22740 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
22741 PROCESSOR_max sanity check.
22742 * config/s390/s390-opts.h (enum processor_type): Add
22744 * config/s390/s390.c (processor_table): Add arch12.
22745 (s390_expand_builtin): Add check for B_VXE flag.
22746 (s390_issue_rate): Add PROCESSOR_ARCH12.
22747 (s390_get_sched_attrmask): Likewise.
22748 (s390_get_unit_mask): Likewise.
22749 (s390_sched_score): Enable z13 scheduling for arch12.
22750 (s390_sched_reorder): Likewise.
22751 (s390_sched_variable_issue): Likewise.
22752 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
22754 (s390_tune_attr): Use z13 scheduling also for arch12.
22755 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
22756 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
22757 (TARGET_VXE_P): New macros.
22758 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
22759 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
22760 * config/s390/s390.opt: Add arch12 as processor_type.
22762 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22764 * config/s390/s390.md
22765 ("fixuns_truncdddi2", "fixuns_trunctddi2")
22766 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
22767 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
22769 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
22770 Rename expanders to ...
22772 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
22773 ("fixuns_truncdddi2_emu"): ... these.
22775 ("fixuns_trunc<mode>si2_emu"): New expander.
22777 ("*fixuns_truncdfdi2_z13"): Rename to ...
22778 ("*fixuns_truncdfdi2_vx"): ... this.
22780 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22782 * config/s390/2964.md: Remove the single element vector compare
22783 instructions which are no longer used.
22784 * config/s390/s390.c (s390_select_ccmode): Remove handling of
22786 (s390_canonicalize_comparison): Remove handling of DFmode
22788 (s390_expand_vec_compare_scalar): Remove function.
22789 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
22790 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
22792 ("*cmp<mode>_ccs"): Add wfcdb instruction.
22794 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22796 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
22798 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
22799 will anyway by matched by mov<mode>_64dfp.
22801 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22803 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
22804 vlef/vstef. Add missing operand to vleif.
22806 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22808 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
22809 pair for all vector types with 64 bit elements.
22810 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
22811 * config/s390/vector.md (V_HW_64): ... here.
22812 (V_128_NOSINGLE): New mode iterator.
22813 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
22814 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
22815 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
22816 ("*vec_load_pairv2di"): Change to ...
22817 ("*vec_load_pair<mode>"): ... this one.
22819 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22821 * config/s390/constraints.md: Add comments.
22822 (jKK): Reject element sizes > 8 bytes.
22823 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
22825 * config/s390/s390.md: Add the s_operand checks formerly in
22826 s390_split_ok_p to various splitters where they are still
22828 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
22829 for 128 bit vectors. Plus two splitters.
22831 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22833 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
22836 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22839 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
22840 error if the boundary argument is not constant.
22842 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22844 PR rtl-optimization/80112
22845 * loop-doloop.c (doloop_condition_get): Don't check condition
22846 if cmp isn't SET with IF_THEN_ELSE src.
22848 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22850 PR tree-optimization/80158
22851 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
22852 replacing a candidate statement, also replace it for the
22853 candidate's alternate interpretation.
22854 (replace_rhs_if_not_dup): Likewise.
22855 (replace_one_candidate): Likewise.
22857 2017-03-24 Richard Biener <rguenther@suse.de>
22859 PR tree-optimization/80167
22860 * graphite-isl-ast-to-gimple.c
22861 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
22863 (translate_isl_ast_to_gimple::get_rename): Likewise.
22865 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
22867 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
22868 handling of certain combinations of target options, including the
22869 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
22870 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
22872 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22875 * config/arm/arm.md (*load_multiple): Add reload_completed to
22876 matching condition.
22878 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22879 Richard Biener <rguenth@suse.de>
22881 PR tree-optimization/79908
22882 PR tree-optimization/80136
22883 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
22884 been cast away, gimplify_and_add suffices.
22886 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
22888 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
22890 2017-03-23 Richard Biener <rguenther@suse.de>
22892 PR tree-optimization/80032
22893 * gimplify.c (gimple_push_cleanup): Forced unconditional
22894 cleanups still have to go to the conditional_cleanups
22897 2017-03-22 Jakub Jelinek <jakub@redhat.com>
22899 PR tree-optimization/80072
22900 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
22902 (next_operand_entry_id): Change type to unsigned int.
22903 (sort_by_operand_rank): Make sure to return the right return value
22904 even if unsigned fields are bigger than INT_MAX.
22905 (struct oecount): Change cnt and id type to unsigned int.
22906 (oecount_hasher::equal): Formatting fix.
22907 (oecount_cmp): Make sure to return the right return value
22908 even if unsigned fields are bigger than INT_MAX.
22909 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
22912 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
22913 TREE_READONLY on result if writing it more than once.
22916 * doc/invoke.texi (-fsanitize=thread): Document that with
22917 -fnon-call-exceptions atomics are not able to throw
22921 * tsan.c: Include tree-eh.h.
22922 (instrument_builtin_call): Call maybe_clean_eh_stmt or
22923 maybe_clean_or_replace_eh_stmt where needed.
22924 (instrument_memory_accesses): Add cfg_changed argument.
22925 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
22926 if it returned true.
22927 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
22929 PR rtl-optimization/63191
22930 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
22931 wrapper function, moved the whole old content into ...
22932 (ix86_delegitimize_address_1): ... this. New inline function.
22933 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
22934 true as last argument instead of ix86_delegitimize_address.
22936 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22938 * config/aarch64/aarch64.c (generic_branch_cost): Copy
22939 cortexa57_branch_cost.
22941 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
22943 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
22945 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22948 * doc/md.texi (Constraints): Document wA constraint.
22949 * config/rs6000/constraints.md (wA): New.
22950 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
22951 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
22952 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
22953 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
22955 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
22958 * gimplify.c (is_oacc_declared): New function.
22959 (oacc_default_clause): Use it to set default flags for acc declared
22960 variables inside parallel regions.
22961 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
22962 declared variables.
22963 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
22964 declare attribute to any decl as necessary.
22966 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
22969 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
22970 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
22971 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
22972 (arm_arch_lpae): This.
22973 * config/arm/arm.c (arm_arch7ve): Rename into ...
22974 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
22975 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
22978 2017-03-22 Martin Liska <mliska@suse.cz>
22981 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
22982 error message instead of an ICE.
22984 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22986 * doc/extend.texi (6.11 Additional Floating Types): Revise.
22988 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
22990 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
22992 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22995 2017-03-21 Martin Sebor <msebor@redhat.com>
22997 * doc/extend.texi: Use "cannot" instead of "can't."
22998 * doc/hostconfig.texi: Same.
22999 * doc/install.texi: Same.
23000 * doc/invoke.texi: Same.
23001 * doc/loop.texi: Same.
23002 * doc/md.texi: Same.
23003 * doc/objc.texi: Same.
23004 * doc/rtl.texi: Same.
23005 * doc/tm.texi: Same.
23006 * doc/tm.texi.in: Same.
23007 * doc/trouble.texi: Same.
23009 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
23012 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
23013 (collect_checksum_attributes): Set it.
23014 (die_checksum_ordered): Use it.
23016 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23018 PR tree-optimization/79908
23019 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
23020 change: For a VA_ARG whose LHS has been cast away, use
23021 force_gimple_operand to construct the side effects.
23023 2017-03-21 David Malcolm <dmalcolm@redhat.com>
23025 PR translation/80001
23026 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
23027 more amenable to translation.
23028 (oacc_loop_auto_partitions): Likewise.
23030 2017-03-21 Marek Polacek <polacek@redhat.com>
23031 Martin Sebor <msebor@redhat.com>
23033 PR tree-optimization/80109
23034 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
23035 on INTEGRAL_TYPE_P.
23037 2017-03-21 Jakub Jelinek <jakub@redhat.com>
23038 Segher Boessenkool <segher@kernel.crashing.org>
23041 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
23042 check reg_used_between_p between insn and one of succ or succ2
23043 depending on if succ is artificial insn not inserted into insn
23046 2017-03-21 Martin Liska <mliska@suse.cz>
23048 PR gcov-profile/80081
23049 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
23050 * doc/gcc.texi: Include gcov-dump stuff.
23051 * doc/gcov-dump.texi: New file.
23053 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
23055 PR rtl-optimization/79150
23056 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
23057 conditional jump, if the jump is the last insn of the loop.
23059 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23060 Richard Biener <rguenth@suse.de>
23062 PR tree-optimization/79908
23063 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
23064 been cast away, use force_gimple_operand to construct the side
23067 2017-03-21 Martin Liska <mliska@suse.cz>
23069 PR libfortran/79956
23070 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
23073 2017-03-21 Brad Spengler <spender@grsecurity.net>
23076 * plugin.c (htab_hash_plugin): New function.
23077 (add_new_plugin): Use it and adjust.
23078 (parse_plugin_arg_opt): Adjust.
23079 (init_one_plugin): Likewise.
23081 2017-03-21 Richard Biener <rguenther@suse.de>
23083 PR tree-optimization/80032
23084 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
23085 if set force the cleanup to happen unconditionally.
23086 (gimplify_target_expr): Push inserted clobbers with force_uncond
23087 to avoid them being removed by control-dependent DCE.
23089 2017-03-21 Richard Biener <rguenther@suse.de>
23091 PR tree-optimization/80122
23092 * tree-inline.c (copy_bb): Do not expans va-arg packs or
23093 va_arg_pack_len when the inlined call stmt requires pack
23095 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
23097 2017-03-21 Jakub Jelinek <jakub@redhat.com>
23100 * tsan.c (instrument_builtin_call): If the memory model argument
23101 is not a constant, assume it is valid.
23104 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
23107 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
23109 PR rtl-optimization/79910
23110 * combine.c (can_combine_p): Do not allow combining an I0 or I1
23111 if its dest is used by an insn before I2 (other than the combined
23112 insns themselves, which are properly handled already).
23114 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
23117 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
23119 * combine.c (record_used_regs): New static function.
23120 (try_combine): Handle situations where there is an additional
23121 instruction between I2 and I3 which needs to have a LOG_LINK
23125 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
23127 * combine.c (try_combine): Delete redundant i1 test. Call
23128 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
23130 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23133 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
23134 alternatives 13/14.
23136 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23138 PR tree-optimization/80054
23139 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
23140 the optimization if a PHI or any of its arguments is not dominated
23141 by the candidate's basis. Use gphi* rather than gimple* as
23143 (replace_profitable_candidates): Clean up a gimple* variable that
23144 should be a gphi* variable.
23146 2017-03-20 Martin Sebor <msebor@redhat.com>
23149 * doc/extend.texi (attribute constructor): Document present limitation.
23151 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
23154 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
23155 __POWER9_VECTOR__ #ifdef control, change template definition to
23156 use Power9-specific built-in function.
23157 (vec_any_eq): Likewise.
23158 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
23159 to control outcomes from this test.
23160 (vector_ae_<mode>p): For VEC_F modes, likewise.
23162 2017-03-20 Ian Lance Taylor <iant@google.com>
23164 * config/i386/i386.c (ix86_function_regparm): Save an extra
23165 register for -fsplit-stack with DECL_STATIC_CHAIN.
23167 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23170 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
23171 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
23173 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23175 * config/riscv/riscv.c (riscv_print_operand): Use "fence
23177 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
23180 2017-03-20 Marek Polacek <polacek@redhat.com>
23183 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
23185 2017-03-20 Richard Biener <rguenther@suse.de>
23187 PR tree-optimization/80113
23188 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
23189 allocate extra SSA name for PHI def.
23190 (add_close_phis_to_outer_loops): Likewise.
23191 (add_close_phis_to_merge_points): Likewise.
23192 (copy_loop_close_phi_args): Likewise.
23193 (copy_cond_phi_nodes): Likewise.
23195 2017-03-20 Martin Liska <mliska@suse.cz>
23197 PR middle-end/79753
23198 * tree-chkp.c (chkp_build_returned_bound): Do not build
23199 returned bounds for a LHS that's not a BOUNDED_P type.
23201 2017-03-20 Martin Liska <mliska@suse.cz>
23205 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
23206 COMPLEX_CST and VECTOR_CST.
23208 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23211 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
23212 target operand. A new splitter adds the clobber statement in case
23213 the target operand is dead anyway.
23215 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
23217 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
23218 to age-old versions of binutils and glibc.
23220 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
23222 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
23224 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
23226 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
23228 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
23230 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
23231 requirement for binutils 2.13.
23233 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
23235 * combine.c (try_combine): Delete redundant i1 test. Call
23236 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
23238 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
23240 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
23241 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
23243 <riscv64-*-elf>: Re-arrange section
23244 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
23245 <riscv32-*-linux>: Likewise.
23246 <riscv64-*-elf>: Likewise
23247 <riscv64-*-linux>: Likewise.
23249 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
23252 * aarch64.opt(verbose-cost-dump): Fix typo.
23254 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
23257 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
23258 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
23260 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
23262 * reload.c (find_reloads): When reloading a nonoffsettable address,
23263 use RELOAD_OTHER for it and its address reloads.
23265 PR rtl-optimization/79910
23266 * combine.c (record_used_regs): New static function.
23267 (try_combine): Handle situations where there is an additional
23268 instruction between I2 and I3 which needs to have a LOG_LINK
23271 2017-03-17 Jeff Law <law@redhat.com>
23273 PR tree-optimization/71437
23274 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
23275 conditional in the hash table first.
23276 (vrp_dom_walker::before_dom_children): Extract condition from
23277 ASSERT_EXPR. Record condition, its inverion and any implied
23278 conditions as well.
23280 2017-03-17 Marek Polacek <polacek@redhat.com>
23281 Markus Trippelsdorf <markus@trippelsdorf.de>
23283 PR tree-optimization/80079
23284 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
23287 2017-03-17 Richard Biener <rguenther@suse.de>
23289 PR middle-end/80075
23290 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
23291 Properly verify the LHS before the RHS possibly claims to be
23293 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
23296 2017-03-17 Martin Jambor <mjambor@suse.cz>
23298 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
23299 (List of -O2 options): Likewise.
23300 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
23303 2017-03-17 Tom de Vries <tom@codesourcery.com>
23305 * gcov-dump.c (print_usage): Print bug_report_url.
23307 2017-03-17 Richard Biener <rguenther@suse.de>
23309 PR middle-end/80050
23310 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
23311 (parser::peek): Likewise.
23313 2017-03-17 Richard Biener <rguenther@suse.de>
23315 PR tree-optimization/80048
23316 * sese.c (free_sese_info): Properly release rename_map and
23317 copied_bb_map elements.
23319 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
23321 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
23322 Add linked-list forward and backlinks. Insert on
23323 construction, remove on destruction.
23324 (class pass_store_merging): Add m_stores_head field.
23325 (pass_store_merging::terminate_and_process_all_chains):
23326 Iterate over m_stores_head list.
23327 (pass_store_merging::terminate_all_aliasing_chains):
23329 (pass_store_merging::execute): Check for debug stmts first.
23330 Push new chains onto the m_stores_head stack.
23332 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
23335 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
23336 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
23337 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
23339 2017-03-16 Jeff Law <law@redhat.com>
23341 PR tree-optimization/71437
23342 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
23343 member function. Implementation moved into after_dom_children
23344 member function and into the threader's thread_outgoing_edges
23346 (dom_opt_dom_walker::after_dom_children): Simplify by moving
23347 some code into new thread_outgoing_edges.
23348 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
23349 definition. Simplify marker handling (do it here). Assume we always
23350 have the available expression and the const/copies tables.
23351 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
23353 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
23354 * tree-vrp.c (equiv_stack): No longer file scoped.
23355 (vrp_dom_walker): New class.
23356 (vrp_dom_walker::before_dom_children): New member function.
23357 (vrp_dom_walker::after_dom_children): Likewise.
23358 (identify_jump_threads): Setup domwalker. Use it rather than
23359 walking edges in a random order by hand. Simplify setup/finalization.
23360 (finalize_jump_threads): Remove.
23361 (vrp_finalize): Do not call identify_jump_threads here.
23362 (execute_vrp): Do it here instead and call thread_through_all_blocks
23365 PR tree-optimization/71437
23366 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
23368 (simplify_stmt_for_jump_threading): Add basic_block argument. All
23370 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
23371 (dom_opt_dom_walker::thread_across_edge): Remove
23372 handle_dominating_asserts argument. All callers changed.
23373 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
23374 changes. Remove calls to lhs_of_dominating_assert. Other
23375 uses of handle_dominating_asserts turn into unconditional code
23376 (simplify_control_stmt_condition_1): Likewise.
23377 (simplify_control_stmt_condition): Likewise.
23378 (thread_through_normal_block, thread_across_edge): Likewise.
23379 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
23380 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
23381 object if it is not an SSA_NAME.
23382 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
23383 before calling into the VRP specific simplifiers.
23384 (identify_jump_threads): Remove handle_dominating_asserts
23387 2017-03-16 Jakub Jelinek <jakub@redhat.com>
23390 * tree-diagnostic.c (default_tree_printer): No longer static.
23391 * tree-diagnostic.h (default_tree_printer): New prototype.
23393 2017-03-16 Tamar Christina <tamar.christina@arm.com>
23395 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
23396 Change ins into fmov.
23398 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23400 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
23401 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
23402 Use h_con constraint for operand 1.
23403 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
23404 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
23406 2017-03-15 Jeff Law <law@redhat.com>
23408 PR tree-optimization/71437
23409 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
23410 (record_temporary_equivalences): Use it.
23412 PR tree-optimization/71437
23413 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
23414 tree-ssa-scopedtables.
23415 (lookup_avail_expr, build_and_record_new_cond): Likewise.
23416 (record_conditions, record_cond, vuse_eq): Likewise.
23417 (record_edge_info): Adjust to API tweak of record_conditions.
23418 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
23419 (record_temporary_equivalences, optimize_stmt): Likewise.
23420 (eliminate_redundant_computations): Likewise.
23421 (record_equivalences_from_stmt): Likewise.
23422 * tree-ssa-scopedtables.c: Include options.h and params.h.
23423 (vuse_eq): New function, moved from tree-ssa-dom.c
23424 (build_and_record_new_cond): Likewise.
23425 (record_conditions): Likewise. Accept vector of conditions rather
23426 than edge_equivalence structure for first argument.
23427 for the first argument.
23428 (avail_exprs_stack::lookup_avail_expr): New member function, moved
23429 from tree-ssa-dom.c.
23430 (avail_exprs_stack::record_cond): Likewise.
23431 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
23432 from tree-ssa-dom.c.
23433 (avail_exprs_stack): Add new member functions lookup_avail_expr
23435 (record_conditions): Declare.
23437 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
23440 * lra-constraints.c (process_alt_operands): Increase reject for
23441 reloading an input/output operand.
23443 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
23446 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
23447 insns to convert from signed/unsigned char/short to IEEE 128-bit
23449 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
23451 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
23454 * config/i386/i386.c (ix86_vector_duplicate_value): Create
23455 subreg of inner mode for values already in registers.
23457 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
23459 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
23460 iteration reg is used after the loop.
23462 2017-03-14 Martin Sebor <msebor@redhat.com>
23464 PR tree-optimization/79800
23465 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
23466 precision in negative-positive range.
23467 (format_floating): Call non-const overload with adjusted precision.
23469 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23472 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
23475 2017-03-14 Martin Sebor <msebor@redhat.com>
23477 PR middle-end/80020
23478 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
23479 * builtins.def (aligned_alloc): Use it.
23482 * Makefile.in (GTFILES): Add calls.c.
23483 * calls.c: Include "gt-calls.h".
23485 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
23487 PR rtl-optimization/79728
23488 * regs.h (struct target_regs): New field
23489 x_contains_allocatable_regs_of_mode.
23490 (contains_allocatable_regs_of_mode): New macro.
23491 * reginfo.c (init_reg_sets_1): Initialize it, and change
23492 contains_reg_of_mode so it includes global regs as well.
23493 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
23494 rather than contains_regs_of_mode.
23496 2017-03-14 Martin Liska <mliska@suse.cz>
23498 * doc/invoke.texi: Document options that can't be combined with
23499 -fcheck-pointer-bounds.
23501 2017-03-14 Martin Liska <mliska@suse.cz>
23503 PR middle-end/79831
23504 * doc/invoke.texi (-Wchkp): Document the option.
23506 2017-03-14 Martin Liska <mliska@suse.cz>
23508 * Makefile.in: Install gcov-dump.
23510 2017-03-14 Martin Liska <mliska@suse.cz>
23512 * multiple_target.c (expand_target_clones): Bail out for
23513 an invalid attribute.
23515 2017-03-14 Richard Biener <rguenther@suse.de>
23517 * alias.c (struct alias_set_entry): Pack properly.
23518 * cfgloop.h (struct loop): Likewise.
23519 * cse.c (struct set): Likewise.
23520 * ipa-utils.c (struct searchc_env): Likewise.
23521 * loop-invariant.c (struct invariant): Likewise.
23522 * lra-remat.c (struct cand): Likewise.
23523 * recog.c (struct change_t): Likewise.
23524 * rtl.h (struct address_info): Likewise.
23525 * symbol-summary.h (function_summary): Likewise.
23526 * tree-loop-distribution.c (struct partition): Likewise.
23527 * tree-object-size.c (struct object_size_info): Likewise.
23528 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
23529 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
23530 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
23531 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
23532 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
23533 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
23534 (struct _stmt_vec_info): Likewise.
23536 2017-03-14 Martin Liska <mliska@suse.cz>
23539 * multiple_target.c (create_dispatcher_calls): Check that
23540 a target can create a function dispatcher.
23542 2017-03-14 Martin Liska <mliska@suse.cz>
23545 * multiple_target.c (expand_target_clones): Drop local.local
23546 flag for default implementation.
23548 2017-03-14 Richard Biener <rguenther@suse.de>
23550 PR tree-optimization/80030
23551 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
23553 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
23555 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
23556 gcc_fallthrough() instead of __attribute__((fallthrough));
23558 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23560 * doc/gcc.texi: Remove "up" link to (DIR).
23561 * doc/gccint.texi: Ditto.
23563 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23565 * doc/install.texi (Specific) <avr>: Remove reference to
23568 2017-03-13 Jeff Law <law@redhat.com>
23570 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
23571 attribute rather than comments.
23573 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
23574 match_scratch operand is highest.
23576 2017-03-13 Martin Liska <mliska@suse.cz>
23578 PR middle-end/78339
23579 * ipa-pure-const.c (warn_function_noreturn): If the declarations
23580 is a CHKP clone, use original declaration.
23582 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23584 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
23585 (arc_conditional_register_usage): Use a different allocation order
23586 when optimizing for size.
23587 * common/config/arc/arc-common.c (arc_option_optimization_table):
23588 Section anchors default on when optimizing for size.
23590 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23592 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
23594 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23596 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
23597 * config/arc/arc.md (cpu_facility): Add cd variant.
23598 (*movqi_insn): Add code density variant.
23599 (*movhi_insn): Likewise.
23600 (*movqi_insn): Likewise.
23601 (*addsi3_mixed): Likewise.
23602 (subsi3_insn): Likewise.
23604 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23606 * config/arc/arc.md (movsi_cond_exec): Update constraint.
23608 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23610 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
23611 expressions with MINUS and UNARY ops.
23613 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23616 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
23618 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
23619 between vec_select and vector argument.
23620 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
23621 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
23622 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
23623 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
23624 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
23625 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
23627 2017-03-13 Richard Biener <rguenther@suse.de>
23630 * params.def (vect-max-peeling-for-alignment): Fix typo.
23632 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23634 * doc/install.texi (Specific) <mips-*-*>: Remove description of
23635 issue that only occurred with binutils below 2.18.
23637 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23639 * doc/install.texi (Specific) <cris-axis-elf>: No longer
23640 refer to binutils 2.11/2.12 minimum.
23642 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23644 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
23645 ftp.kernel.org and simplify binutils requirement.
23647 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
23649 * doc/invoke.texi (Warning Options): Fix spelling of link-time
23651 (Optimize Options): Ditto. Also remove redundancy.
23653 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23655 PR translation/79848
23656 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
23658 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
23659 to G_ to avoid double translation.
23661 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23663 PR translation/79923
23664 * auto-profile.c (get_combined_location): Convert leading
23665 character of diagnostics to lower case and remove trailing period.
23666 (read_profile): Likewise for various diagnostics.
23667 * config/arm/arm.c (arm_option_override): Remove trailing period
23668 from various diagnostics.
23669 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
23670 (msp430_expand_delay_cycles): Likewise.
23672 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23675 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
23676 full command-line argument, rather than just "str".
23677 (aarch64_validate_march): Likewise.
23678 (aarch64_validate_mtune): Likewise.
23680 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
23682 PR rtl-optimization/78911
23683 * lra-assigns.c (must_not_spill_p): New function.
23684 (spill_for): Use it.
23686 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23688 PR tree-optimization/79981
23689 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
23690 ATOMIC_COMPARE_EXCHANGE ifn result.
23691 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
23692 IFN_ATOMIC_COMPARE_EXCHANGE.
23694 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23697 * opts.c (parse_sanitizer_options): Add missing question mark to
23698 "did you mean" message.
23700 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23702 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
23704 (VMULEUH_UNS): Likewise.
23705 (VMULOUB_UNS): Likewise.
23706 (VMULOUH_UNS): Likewise.
23707 * config/rs6000/rs6000.c (builtin_function_type): Remove
23708 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
23710 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23713 * read-rtl-function.c (function_reader::read_rtx_operand): Update
23714 x with result of extra_parsing_for_operand_code_0.
23715 (function_reader::extra_parsing_for_operand_code_0): Convert
23716 return type from void to rtx, returning x. When reading
23717 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
23718 larger size containing struct block_symbol.
23720 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
23722 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
23723 -mfloat128-hardware without -m64.
23725 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
23728 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
23729 entries to the case statement that marks unsigned arguments to
23730 overloaded functions.
23732 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23734 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
23735 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
23737 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
23740 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
23741 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
23743 2017-03-10 Martin Liska <mliska@suse.cz>
23747 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
23748 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
23751 2017-03-10 Olivier Hainque <hainque@adacore.com>
23753 * tree-switch-conversion (array_value_type): Start by resetting
23754 candidate type to it's main variant.
23756 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23758 PR rtl-optimization/79909
23759 * combine.c (try_combine): Use simplify_replace_rtx on individual
23760 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
23761 of the whole CALL_INSN_FUNCTION_USAGE.
23763 PR tree-optimization/79972
23764 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
23765 get_range_info on SSA_NAMEs. Formatting fixes.
23767 2017-03-10 Richard Biener <rguenther@suse.de>
23768 Jakub Jelinek <jakub@redhat.com>
23770 PR tree-optimization/77975
23771 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
23772 edge to be constant.
23773 (get_val_for): For constant x return it. Formatting fix.
23774 (loop_niter_by_eval): Avoid pointless looping if the next iteration
23775 would use the same bases as the current one.
23777 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23779 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
23780 instead of vec_select for V1TImode.
23781 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
23783 (VSX_LE_128): Add V1TI to this mode iterator.
23784 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
23785 (*vsx_le_perm_store_<mode>): Likewise.
23786 (pre-reload splitter for VSX stores): Likewise.
23787 (post-reload splitter for VSX stores): Likewise.
23788 (*vsx_xxpermdi2_le_<mode>): Likewise.
23789 (*vsx_lxvd2x2_le_<mode>): Likewise.
23790 (*vsx_stxvd2x2_le_<mode>): Likewise.
23792 2017-03-09 Michael Eager <eager@eagercon.com>
23794 Correct failures with --enable-checking=yes,rtl.
23796 * config/microblaze/microblaze.c (microblaze_expand_shift):
23797 Replace GET_CODE test with CONST_INT_P and INTVAL test with
23798 test for const0_rtx.
23799 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
23800 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
23802 2017-03-09 Richard Biener <rguenther@suse.de>
23804 PR tree-optimization/79977
23805 * graphite-scop-detection.c (scop_detection::merge_sese):
23806 Handle the case of extra exits to blocks dominating the entry.
23808 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
23810 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
23813 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
23815 PR rtl-optimization/79949
23816 * lra-constraints.c (process_alt_operands): Check memory when
23817 trying to predict a cycle. Print about the overall increase.
23819 2017-03-09 Richard Biener <rguenther@suse.de>
23821 PR middle-end/79971
23822 * gimple-expr.c (useless_type_conversion_p): Preserve
23823 TYPE_SATURATING for fixed-point types.
23825 2017-03-09 Richard Biener <rguenther@suse.de>
23828 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
23829 alignment of BLKmode params.
23831 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23834 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
23835 (VALL_NO_V2Q): Likewise.
23837 * config/aarch64/aarch64-simd.md
23838 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
23840 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
23841 VALL_NO_V2Q mode iterator.
23842 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
23844 2017-03-09 Martin Liska <mliska@suse.cz>
23846 PR tree-optimization/79631
23847 * tree-chkp-opt.c (chkp_is_constant_addr): Call
23848 tree_int_cst_sign_bit just for INTEGER constants.
23850 2017-03-09 Martin Liska <mliska@suse.cz>
23854 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
23857 2017-03-09 Marek Polacek <polacek@redhat.com>
23860 * tree.c (inchash::add_expr): Handle TREE_VEC.
23862 2017-03-09 Martin Liska <mliska@suse.cz>
23865 (chkp_narrow_size_and_offset): New function.
23866 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
23867 (void chkp_parse_bit_field_ref): New function.
23868 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
23869 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
23871 2017-03-09 Martin Liska <mliska@suse.cz>
23874 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
23875 (chkp_find_bounds_1): Remove gcc_unreachable.
23877 2017-03-09 Jakub Jelinek <jakub@redhat.com>
23880 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
23881 BUILT_IN_SYNC*, determine the access type from the size suffix and
23882 always build a MEM_REF with that type. Handle forgotten
23883 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
23886 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
23887 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
23888 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
23889 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
23890 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
23891 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
23892 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
23893 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
23894 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
23895 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
23896 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
23897 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
23898 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
23899 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
23900 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
23901 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
23902 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
23903 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
23904 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
23905 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
23906 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
23907 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
23908 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
23909 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
23910 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
23911 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
23912 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
23913 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
23914 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
23915 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
23916 definitions outside of __OPTIMIZE__ guarded section.
23919 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
23920 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
23921 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
23922 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
23925 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23927 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
23928 ("vfenez<mode>"): Add missing constraints.
23930 2017-03-08 Martin Sebor <msebor@redhat.com>
23933 * config/nds32/nds32.c (nds32_option_override):
23934 Fix misspelled diagnostic.
23936 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23939 * gimplify.c (gimplify_omp_for): Replace index var in outer
23940 taskloop statement with an artificial variable and add
23941 OMP_CLAUSE_PRIVATE clause for it.
23943 2017-03-08 Richard Biener <rguenther@suse.de>
23945 PR tree-optimization/79955
23946 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
23947 for accesses that are completely outside of the variable.
23949 2017-03-08 Andrew Haley <aph@redhat.com>
23951 PR tree-optimization/79943
23952 * tree-ssa-loop-split.c (compute_new_first_bound): When
23953 calculating the new upper bound, (END-BEG) should be added, not
23956 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23958 * config/avr/avr.md (setmemhi): Make sure match_dup
23959 operand number comes before match_scratch.
23961 2017-03-08 Richard Biener <rguenther@suse.de>
23963 PR tree-optimization/79920
23964 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
23965 with ncopies == 1 to ...
23966 (vect_transform_slp_perm_load): ... here. Properly compute
23967 all element loads by iterating VF times over the group. Do
23968 not handle ncopies (computed in a broken way) in
23969 vect_create_mask_and_perm.
23971 2017-03-08 Jakub Jelinek <jakub@redhat.com>
23974 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
23975 is a uniform vector, use uniform_vector_p return value instead of
23976 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
23978 2017-03-07 Marek Polacek <polacek@redhat.com>
23980 PR middle-end/79809
23981 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
23982 (alloca_call_type): Likewise.
23984 2017-03-07 Martin Liska <mliska@suse.cz>
23986 * gcov.c (process_args): Put comment to correct location.
23988 2017-03-07 Martin Liska <mliska@suse.cz>
23990 PR middle-end/68270
23991 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
23992 Use array_at_struct_end_p instead of DECL_CHAIN (field).
23993 (chkp_narrow_bounds_for_field): Likewise.
23994 (chkp_parse_array_and_component_ref): Pass one more argument to
23997 2017-03-07 Richard Biener <rguenther@suse.de>
23999 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
24002 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
24004 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
24005 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
24007 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24010 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
24011 to end of description.
24012 (PARAM_MAX_STORES_TO_MERGE): Likewise.
24014 2017-03-07 Jakub Jelinek <jakub@redhat.com>
24016 PR rtl-optimization/79901
24017 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
24019 (*avx512f_<code><mode>3<mask_name>): ... this.
24020 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
24021 iterator instead of VI8_AVX2_AVX512BW.
24023 PR rtl-optimization/79901
24024 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
24025 min/max expander, expand it using expand_vec_cond_expr.
24028 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
24031 2017-03-06 Jakub Jelinek <jakub@redhat.com>
24034 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
24035 to void * for PCH reasons.
24036 * dwarf2out.c (output_loc_operands, output_die): Cast
24037 v.val_vec.array to unsigned char *.
24039 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
24042 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
24045 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
24047 PR rtl-optimization/79571
24048 * lra-constraints.c (process_alt_operands): Calculate static
24049 reject and subtract it from overall when only addresses will be
24052 2017-03-06 Julia Koval <julia.koval@intel.com>
24055 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
24056 incoming stack boundary to 128 for 64-bit targets.
24058 2017-03-06 Richard Biener <rguenther@suse.de>
24060 PR tree-optimization/79894
24061 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
24062 to NULL after folding it.
24064 2017-03-06 Richard Biener <rguenther@suse.de>
24066 PR tree-optimization/79824
24067 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
24068 check disabling peeling for gaps.
24070 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
24072 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24073 attributes): Document gettimeofday.
24075 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
24077 * config/s390/s390.c (s390_option_override_internal): Set
24078 PARAM_MIN_VECT_LOOP_BOUND
24080 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
24082 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
24083 * config/s390/s390.md: Likewise.
24085 2017-03-06 Jakub Jelinek <jakub@redhat.com>
24088 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
24089 (<avx2_avx512>_perm<mode>): Rename to ...
24090 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
24092 (<avx512>_perm<mode>_mask): Rename to ...
24093 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
24095 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
24096 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
24097 instead of VI8F_256_512.
24098 (avx512f_perm<mode>): New define_expand.
24099 (avx512f_perm<mode>_mask): Likewise.
24100 (avx512f_perm<mode>_1<mask_name>): New define_insn.
24101 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
24103 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24105 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
24106 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
24108 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
24110 2017-03-06 Martin Liska <mliska@suse.cz>
24113 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
24114 when having a SSA NAME w/o VAR_DECL assigned to it.
24116 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24118 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
24119 msa_dpsub_<su>_d): Fix MODE for vec_select.
24121 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24123 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
24125 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
24127 2017-03-06 Richard Biener <rguenther@suse.de>
24129 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
24130 * plugin.c (register_plugin_info): Likewise.
24131 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
24133 2017-03-05 Jakub Jelinek <jakub@redhat.com>
24135 * config/i386/sse.md (sse_storehps, sse_storelps,
24136 avx_<castmode><avxsizesuffix>_<castmode>,
24137 avx512f_<castmode><avxsizesuffix>_<castmode>,
24138 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
24139 in condition that at least one operand is not a MEM.
24141 2017-03-03 Jakub Jelinek <jakub@redhat.com>
24143 PR middle-end/79805
24144 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
24145 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
24147 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
24148 gimple_call_nothrow_p flag based on whether original builtin can throw.
24149 If it can, emit following stmts on the fallthrough edge.
24150 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
24151 don't create new bb if inserting just debug stmts on the edge, try to
24152 insert them on the fallthru bb or just reset debug stmts.
24154 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
24157 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
24158 restore recog_data (including the operand rtxes inside it) around
24159 the call to get_insn_template.
24161 2017-03-03 Martin Sebor <msebor@redhat.com>
24163 PR tree-optimization/79699
24164 * context.c (context::~context): Free MPFR caches to avoid
24165 a memory leak on program exit.
24167 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24169 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
24170 Use wide_int::ulow () instead of .elt (0).
24172 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24174 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
24175 (*pushxf): Limit oF constraint to 32bit targets and add oC
24176 constraint for 64bit targets.
24177 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
24178 (*pushdf): Change rmF constraint to rmC.
24180 2017-03-03 Martin Liska <mliska@suse.cz>
24182 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
24183 Remove unused variable.
24185 2017-03-03 Jakub Jelinek <jakub@redhat.com>
24188 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
24189 is a memory operand, increase num_memory.
24190 (ix86_expand_args_builtin): Likewise.
24192 2017-03-03 Jan Hubicka <jh@suse.cz>
24195 * ipa-devirt.c (maybe_record_node): Properly handle
24196 __cxa_pure_virtual visibility.
24198 2017-03-03 Martin Liska <mliska@suse.cz>
24200 PR tree-optimization/79803
24201 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
24203 (pass_loop_prefetch::execute): Disabled optimization if an
24204 assumption about L1 cache size is not met.
24206 2017-03-03 Martin Liska <mliska@suse.cz>
24208 PR rtl-optimization/79574
24209 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
24210 (hash_scan_set): Likewise.
24211 (dump_hash_table): Likewise.
24212 (hoist_code): Likewise.
24214 2017-03-03 Richard Biener <rguenther@suse.de>
24216 * fixed-value.c (fixed_from_string): Restore use of elt (1)
24217 in place of uhigh ().
24218 (fixed_convert_from_real): Likewise.
24220 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24223 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
24225 2017-03-03 Richard Biener <rguenther@suse.de>
24227 PR middle-end/79818
24228 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
24229 TYPE_OVERFLOW_UNDEFINED check.
24231 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24233 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
24235 (vector_ae_<mode>_p): Likewise.
24236 (vector_nez_<mode>_p): Likewise.
24237 (vector_ne_v2di_p): Likewise.
24238 (vector_ae_v2di_p): Likewise.
24239 (vector_ne_<mode>_p): Likewise.
24240 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
24242 (vsx_tsqrt<mode>2_fe): Likewise.
24244 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
24247 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
24249 2017-03-02 Jakub Jelinek <jakub@redhat.com>
24251 PR rtl-optimization/79780
24252 * cprop.c (one_cprop_pass): When second and further conditional trap
24253 in a single basic block is turned into an unconditional trap, turn it
24254 into a deleted note to avoid RTL verification failures.
24256 2017-03-02 Richard Biener <rguenther@suse.de>
24258 * fold-const.c (const_binop): Use ulow () instead of elt (0).
24260 2017-03-02 Richard Biener <rguenther@suse.de>
24262 PR tree-optimization/79345
24264 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
24265 param and abort the walk, returning -1 if it is hit.
24266 (walk_aliased_vdefs): Take a limit param and pass it on.
24267 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
24268 defaulting to 0 and return a signed int.
24269 * tree-ssa-uninit.c (struct check_defs_data): New struct.
24270 (check_defs): New helper.
24271 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
24272 about uninitialized memory.
24273 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
24274 bogus uninitialized warning.
24275 (fixed_convert_from_real): Likewise.
24277 2017-03-02 Bin Cheng <bin.cheng@arm.com>
24279 PR tree-optimization/66768
24280 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
24281 iv_use if base object can't be determined.
24283 2017-03-02 Jakub Jelinek <jakub@redhat.com>
24285 PR tree-optimization/79345
24286 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
24287 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
24288 (get_pattern_stats): Initialize it.
24289 * genemit.c (gen_expand): Verify match_scratch numbers come after
24290 match_operand/match_dup numbers.
24291 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
24292 match_scratch numbers.
24293 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
24295 * config/s390/s390.md (trunctdsd2): Likewise.
24297 2017-03-02 Richard Biener <rguenther@suse.de>
24299 * wide-int.h (wide_int_storage::operator=): Implement in terms
24302 2017-03-02 Richard Biener <rguenther@suse.de>
24304 PR tree-optimization/79777
24305 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
24306 the to insert expression to sth existing.
24308 2017-03-01 Martin Sebor <msebor@redhat.com>
24310 PR middle-end/79692
24311 * gimple-ssa-sprintf.c
24312 (directive::known_width_and_precision): New function.
24313 (format_integer): Use it.
24314 (get_mpfr_format_length): Consider the full range of precision
24315 when computing %g output with the # flag. Set the likely byte
24316 count to 3 rather than 1 when precision is indeterminate.
24317 (format_floating): Correct the lower bound of precision.
24319 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24321 * doc/invoke.texi: Document default code model for 64-bit Linux.
24323 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24326 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
24327 udiv rather than div since input pattern is unsigned.
24329 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
24331 * config/i386/i386.c (print_reg): Warn for values of
24332 unsupported size in integer register.
24334 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
24337 * config/rs6000/predicates.md (current_file_function_operand): Do
24338 not allow self calls to be local if the function is replaceable.
24340 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
24343 * config/rs6000/altivec.h (vec_ctz and others): Change the
24344 preprocessor macro that controls conditional compilation from
24345 _ARCH_PWR9 to __POWER9_VECTOR__.
24346 (vec_all_ne): Change parameterization of __altivec_scalar_pred
24347 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
24348 control (instead of _ARCH_PWR9 control) so that template
24349 definition uses power9-specific function.
24350 (vec_any_eq): Likewise.
24351 (vec_all_ne): Change macro definition to use a power9-specific
24352 expansion under #ifdef __POWER9_VECTOR__ control (instead of
24353 _ARCH_PWR9 control).
24354 (vec_any_eq) Likewise.
24355 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
24356 expansion for CMPNEF to remove support for xvcmpnesp instruction.
24357 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
24358 support for xvcmpnedp instruction.
24359 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
24360 macro expansion so that Power9 implementation of vec_all_ne does
24361 not use the AltiVec predicate framework.
24362 (VCMPNEH_P): Likewise.
24363 (VCMPNEW_P): Likewise.
24364 (VCMPNED_P): Likewise.
24365 (VCMPNEFP_P): Likewise.
24366 (VCMPNEDP_P): Likewise.
24367 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
24368 implementation of vec_any_eq to not use AltiVec predicate
24370 (VCMPAEH_P): Likewise.
24371 (VCMPAEW_P): Likewise.
24372 (VCMPAED_P): Likewise.
24373 (VCMPAEFP_P): Likewise.
24374 (VCMPAEDP_P): Likewise.
24375 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
24376 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
24377 not use the AltiVec predicate framework.
24378 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
24379 of vec_any_eq to not use AltiVec predicate framework.
24380 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24381 support for predefined __POWER9_VECTOR__ macro to indicate that
24382 Power9 instruction selection is enabled.
24383 (altivec_overloaded_builtins): Remove extraneous
24384 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
24385 function argument types RS6000_BTI_bool_V16QI and
24386 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
24387 entry for overloaded function argument types RS6000_BTI_bool_V4SI
24388 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
24389 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
24390 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
24391 Power9 for implementations of vec_cmpne. Change the signature for
24392 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
24393 (representing vec_all_ne) to remove the previously described first
24394 argument of type RS6000_BTI_INTSI, as this was an artifact of
24395 reliance on the AltiVec predicate framework, which is no longer
24396 used in the implementation of these functions. Add
24397 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
24398 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
24399 since, unlike the AltiVec predicate framework implementation, we
24400 do not share function descriptors between vec_alle and vec_anyeq.
24401 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
24402 set of modes that receive special treatment even when
24403 TARGET_P9_VECTOR is true. The special treatment emits code that
24404 does not depend on Power9 instructions.
24405 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
24406 define_expand to not rely on AltiVec predicate framework.
24407 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24409 (vector_ne_v2di_p): Change this define_expand to not rely on
24410 AltiVec predicate framework.
24411 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
24413 (vector_ne_<mode>_p): Change this define_expand to not rely on
24414 AltiVec predicate framework.
24415 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24417 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
24418 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
24419 define_insn pattern.
24420 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
24421 define_insn pattern because the xvcmpne<VSs>. instruction is not
24423 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
24424 instruction is not supported.
24426 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24428 * config/nvptx/nvptx.c: Include intl.h.
24430 2017-03-01 Martin Jambor <mjambor@suse.cz>
24433 * ipa-prop.h (ipa_bits): Removed field known.
24434 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
24435 to pointers. Adjusted their comments to warn about their sharing.
24436 (ipcp_transformation_summary): Change bits to a vector of pointers.
24437 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
24438 (ipa_get_ipa_bits_for_value): Declare.
24439 * tree-vrp.h (value_range): Mark as GTY((for_user)).
24440 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
24441 (ipa_bits_hash_table): Likewise.
24442 (ipa_vr_ggc_hash_traits): Likewise.
24443 (ipa_vr_hash_table): Likewise.
24444 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
24445 being pointers and vr_known being removed.
24446 (ipa_set_jf_unknown): Likewise.
24447 (ipa_get_ipa_bits_for_value): New function.
24448 (ipa_set_jfunc_bits): Likewise.
24449 (ipa_get_value_range): New overloaded functions.
24450 (ipa_set_jfunc_vr): Likewise.
24451 (ipa_compute_jump_functions_for_edge): Use the above functions to
24452 construct bits and vr parts of jump functions.
24453 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
24454 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24456 (ipcp_grow_transformations_if_necessary): Also allocate
24457 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24459 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
24460 them. Fix too long lines.
24461 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
24462 vr_known being removed.
24463 (ipa_read_jump_function): Use new setter functions to construct bits
24464 and vr parts of jump functions or set them to NULL.
24465 (write_ipcp_transformation_info): Adjust for bits being pointers.
24466 (read_ipcp_transformation_info): Likewise.
24467 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
24469 Include gt-ipa-prop.h.
24470 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
24472 (ipcp_store_bits_results): Likewise.
24473 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
24474 Do not write to existing jump functions but use a temporary instead.
24476 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24479 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
24480 attempt to use its first operand as BIT_FIELD_REF base.
24482 2017-03-01 Richard Biener <rguenther@suse.de>
24484 PR middle-end/79721
24485 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
24486 interpolating formula in wrapping arithmetic.
24487 (chrec_apply): Convert chrec_evaluate return value to wanted type.
24489 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24491 PR tree-optimization/79734
24492 * tree-vect-generic.c (expand_vector_condition): Optimize
24493 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
24494 Handle VEC_COND_EXPR where comparison has different inner width from
24495 type's inner width.
24497 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
24499 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
24500 markup, and similar issues. Remove @opindex entries for things
24501 that aren't options. Add missing -mmpy-option entries.
24503 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24505 PR tree-optimization/79737
24506 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
24507 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
24508 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
24509 instead of byte_size. Formatting fix.
24510 (shift_bytes_in_array_right): Formatting fix.
24512 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
24515 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
24516 condition on optimize for the leaf function test.
24518 2017-02-28 Martin Liska <mliska@suse.cz>
24521 * read-rtl-function.c (function_reader::handle_unknown_directive):
24522 Bail out when one uses -flto.
24524 2017-02-28 Martin Liska <mliska@suse.cz>
24526 * common.opt: Replace space with tabular for options of <number>
24528 * config/i386/i386.opt: Show <number> value for
24529 -mlarge-data-threshold.
24530 * opts.c (print_filtered_help): Do not display number in hexadecimal
24533 2017-02-28 Martin Liska <mliska@suse.cz>
24535 * common.opt: Fix --help=option -Q for options which are of
24538 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
24540 * config/i386/i386.c (print_reg): Error out for values
24541 of 8-bit size in invalid integer register.
24543 2017-02-28 Martin Sebor <msebor@redhat.com>
24545 PR tree-optimization/79691
24546 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
24548 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24551 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
24552 gcc_unreachable with output_operand_lossage.
24554 2017-02-28 Richard Biener <rguenther@suse.de>
24556 PR tree-optimization/79740
24557 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
24559 (visit_nary_op): Insert the nary into the hashtable if we
24560 pattern-matched sth.
24561 * tree-ssa-pre.c (eliminate_insert): Robustify.
24563 2017-02-28 Richard Biener <rguenther@suse.de>
24565 PR middle-end/79731
24566 * fold-const.c (decode_field_reference): Reject out-of-bound
24569 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24571 * config/i386/i386.c: Include intl.h.
24572 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
24573 instead of just cond ? "..." : "...".
24574 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
24575 * coverage.c (read_counts_file): Likewise.
24576 * omp-offload.c: Include intl.h.
24577 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
24578 of just cond ? "..." : "...".
24579 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
24580 of just cond ? "..." : "...".
24582 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
24585 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
24587 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
24588 'tune for' CPU name.
24589 * config/arm/arm-cpu-data.h: Regenerated.
24591 2017-02-28 Richard Biener <rguenther@suse.de>
24593 PR tree-optimization/79732
24594 * tree-inline.c (expand_call_inline): Do not shadow var.
24596 2017-02-28 Richard Biener <rguenther@suse.de>
24598 PR tree-optimization/79723
24599 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
24600 address-space properly.
24602 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
24604 * doc/optinfo.texi (Optimization groups): Fix option used for
24606 * doc/invoke.texi (-fopt-info): Document "omp".
24607 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
24608 (OPTGROUP_ALL): Add OPTGROUP_OMP.
24609 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
24610 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
24611 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
24613 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
24615 * dumpfile.c (optgroup_options): Instead of "openmp", associate
24616 OPTGROUP_OMP with "omp".
24618 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
24621 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
24622 for arithmetic shift of unsigned V2DI.
24624 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
24626 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
24627 arc/linux.h headers.
24628 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
24629 (LINK_SPEC): Likewise.
24630 (ARC_TLS_EXTRA_START_SPEC): Likewise.
24631 (EXTRA_SPECS): Likewise.
24632 (STARTFILE_SPEC): Likewise.
24633 (ENDFILE_SPEC): Likewise.
24634 (LIB_SPEC): Likewise.
24635 (TARGET_SDATA_DEFAULT): Likewise.
24636 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
24637 (MULTILIB_DEFAULTS): Likewise.
24638 (DWARF2_UNWIND_INFO): Likewise.
24639 * config/arc/big.h: New file.
24640 * config/arc/elf.h: Likewise.
24641 * config/arc/linux.h: Likewise.
24642 * config/arc/t-uClibc: Remove.
24644 2017-02-27 Bin Cheng <bin.cheng@arm.com>
24646 PR tree-optimization/77536
24647 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
24648 (tree_transform_and_unroll_loop): Use above function to compute the
24649 estimated niter of unrolled loop and use it when scaling profile.
24650 Also use count info rather than frequency if it's non-zero.
24651 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
24652 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
24653 (vect_transform_loop): Call above function.
24655 2017-02-27 Richard Biener <rguenther@suse.de>
24657 PR tree-optimization/45397
24658 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
24659 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
24660 (visit_nary_op): Add pattern matching for CSEing sign-changed
24661 or truncated operations with wider ones.
24663 2017-02-27 Richard Biener <rguenther@suse.de>
24665 PR tree-optimization/79690
24666 * tree-vect-stmts.c (vectorizable_store): Use vector type
24667 built from the DR with address-space.
24669 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
24671 * doc/invoke.texi (Optimize Options): Refine the description
24672 of asan-use-after-return.
24674 2017-02-25 Alan Modra <amodra@gmail.com>
24676 PR rtl-optimization/79584
24677 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
24678 base, not ad->base_term, the reg within base. Remove assertion
24679 that ad->base == ad->base_term. Replace gen_int_mode using
24680 bogus mode with const0_rtx.
24682 2017-02-25 Jakub Jelinek <jakub@redhat.com>
24684 PR middle-end/79396
24685 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
24686 FMA_EXPR like tcc_binary or tcc_unary.
24688 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
24691 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
24693 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
24694 (output_loc_operands): Handle DW_OP_call_ref and
24695 DW_OP_GNU_variable_value.
24696 (struct variable_value_struct): New type.
24697 (struct variable_value_hasher): Likewise.
24698 (variable_value_hash): New variable.
24699 (string_types): Remove.
24700 (copy_loc_descr): New function.
24701 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
24702 (prepend_loc_descr_to_each): New function.
24703 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
24704 instead of add_loc_descr_to_each if the first argument is single
24705 location list and the second has multiple.
24706 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
24707 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
24708 when looking for variable value which doesn't have other location info.
24709 (loc_list_from_tree): Formatting fix.
24710 (gen_array_type_die): Simplify DW_AT_string_length handling.
24711 (adjust_string_types): Remove.
24712 (gen_subprogram_die): Don't call adjust_string_types nor test/set
24713 string_types. Call resolve_variable_values.
24714 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
24715 (resolve_addr_in_expr): Likewise. Add A argument.
24716 (copy_deref_exprloc): Remove deref argument. Adjust for the
24717 original expression being DW_OP_GNU_variable_value with optionally
24718 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
24719 optionally after it.
24720 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
24721 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
24722 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
24723 (variable_value_hasher::hash, variable_value_hasher::equal): New
24725 (resolve_variable_value_in_expr, resolve_variable_value,
24726 resolve_variable_values, note_variable_value_in_expr,
24727 note_variable_value): New functions.
24728 (dwarf2out_early_finish): Call note_variable_value on all toplevel
24731 2017-02-24 Jakub Jelinek <jakub@redhat.com>
24734 * opts.h (handle_generated_option): Add GENERATED_P argument.
24735 * opts-common.c (handle_option): Adjust function comment.
24736 (handle_generated_option): Add GENERATED_P argument, pass it to
24738 (control_warning_option): Pass false to handle_generated_option
24740 * opts.c (maybe_default_option): Pass true to handle_generated_option
24742 * optc-gen.awk: Likewise.
24744 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24746 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
24747 a REG, look at the REG it is a SUBREG of.
24748 (splitter for cmpeqsi_t): Ditto.
24750 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24752 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
24753 the special USEs with the pattern of the insn, not the insn itself.
24755 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
24758 * doc/invoke.texi: Document -mload-store-pairs.
24760 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24761 Sandra Loosemore <sandra@codesourcery.com>
24763 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
24764 argument isn't a CONST_INT.
24765 (nios2_alternate_compare_const): Assert op is a CONST_INT.
24766 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
24767 (nios2_validate_compare): Bypass alternate compare logic if *op2
24768 is not a CONST_INT.
24769 (ldstwm_operation_p): Return false if first_base is not a REG or
24770 if first_offset is not a CONST_INT.
24772 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24774 * config/cris/cris.md: Use correct operand in a define_peephole2.
24776 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24778 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
24780 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
24782 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
24783 this_insn if it is an INSN or JUMP_INSN.
24784 (force_offsettable): Look at base, not at addr.
24785 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
24786 on things that aren't necessarily CONST_INTs.
24788 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
24790 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
24791 -mfpmath=sse is the default also for x86-32 targets with SSE2
24792 instruction set when @option{-ffast-math} is enabled
24794 2017-02-24 Jeff Law <law@redhat.com>
24796 PR rtl-optimizatoin/79286
24797 * ira.c (update_equiv_regs): Drop may_trap_p exception to
24800 2017-02-24 Richard Biener <rguenther@suse.de>
24802 PR tree-optimization/79389
24803 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
24806 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
24808 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
24809 function comment to reflect reality.
24810 (loop_exits_before_overflow): Fix typo in function description.
24812 2017-02-24 Richard Biener <rguenther@suse.de>
24814 PR tree-optimization/79389
24815 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
24816 properly that a threading opportunity exists. Detect conditional
24817 copy/constant propagation opportunities.
24819 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
24821 * config/visium/visium.md (type): Add trap.
24822 (b): New mode attribute.
24823 (*btst): Rename into...
24824 (*btst<mode>): ...this and adjust.
24825 (*cbranchsi4_btst_insn): Rename into...
24826 (*cbranch<mode>4_btst_insn): ...this and adjust.
24827 (trap): New define_insn.
24829 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24831 PR tree-optimization/79389
24832 * ifcvt.c (struct noce_if_info): Add rev_cond field.
24833 (noce_reversed_cond_code): New function.
24834 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
24835 reversed_comparison_code. Formatting fix.
24836 (noce_try_store_flag): Test rev_cond != NULL in addition to
24837 reversed_comparison_code.
24838 (noce_try_store_flag_constants): Likewise.
24839 (noce_try_store_flag_mask): Likewise.
24840 (noce_try_addcc): Use rev_cond if non-NULL instead of
24841 reversed_comparison_code.
24842 (noce_try_cmove_arith): Likewise. Formatting fixes.
24843 (noce_try_minmax, noce_try_abs): Clear rev_cond.
24844 (noce_find_if_block): Initialize rev_cond.
24845 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
24846 instead of false as last argument never attempt to reverse it
24849 2017-02-23 Bin Cheng <bin.cheng@arm.com>
24851 PR tree-optimization/79663
24852 * tree-predcom.c (combine_chains): Process refs in reverse order
24853 only for ZERO length chains, and add explaining comment.
24855 2017-02-23 Jeff Law <law@redhat.com>
24857 PR tree-optimization/79578
24858 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
24859 in call to operand_equal_p.
24861 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
24864 * config/i386/cpuid.h: Fix another undefined behavior.
24866 2017-02-23 Richard Biener <rguenther@suse.de>
24868 PR tree-optimization/79683
24869 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
24870 vector types for data-refs.
24872 2017-02-23 Martin Liska <mliska@suse.cz>
24874 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
24876 2017-02-23 Jakub Jelinek <jakub@redhat.com>
24878 PR middle-end/79665
24879 * internal-fn.c (get_range_pos_neg): Moved to ...
24880 * tree.c (get_range_pos_neg): ... here. No longer static.
24881 * tree.h (get_range_pos_neg): New prototype.
24882 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
24883 are known to be in between 0 and signed maximum inclusive, try to
24884 expand both unsigned and signed divmod and use the cheaper one from
24887 2017-02-22 Jeff Law <law@redhat.com>
24889 PR tree-optimization/79578
24890 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
24891 to compare base operands.
24893 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
24896 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
24897 gpc_reg_operand instead of fpr_reg_operand.
24899 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
24901 * config/mips/mips.c (mips_return_in_memory): Force FP
24902 vector types to be returned in memory for o32 ABI.
24904 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24906 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
24907 instead of DW_TAG_member for static data member declarations and don't
24908 set no_linkage_name for static inline data members.
24909 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
24912 2017-02-22 Martin Liska <mliska@suse.cz>
24914 * doc/invoke.texi: Replace inequality signs with square brackets
24917 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24919 PR tree-optimization/68644
24920 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
24922 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
24925 * lra-constraints.c (simplify_operand_subreg): Handle
24926 WORD_REGISTER_OPERATIONS targets.
24928 2017-02-22 Jakub Jelinek <jakub@redhat.com>
24931 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
24932 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
24933 elimination by swapping fld*.
24935 2017-02-22 Richard Biener <rguenther@suse.de>
24937 PR tree-optimization/79673
24938 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
24939 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
24940 irrelevant address-space qualifiers and avoiding a
24941 ADDR_SPACE_CONVERT_EXPR from fold_convert.
24943 2017-02-22 Richard Biener <rguenther@suse.de>
24945 PR tree-optimization/79666
24946 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
24947 to not symbolically negate if that may introduce undefined
24950 2017-02-22 Martin Liska <mliska@suse.cz>
24953 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
24954 * data-streamer-out.c (streamer_write_gcov_count_stream):
24956 * value-prof.c (stream_out_histogram_value): Make assert more
24957 precise based on type of counter.
24959 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
24962 * config/i386/i386.md (standard_x87sse_constant_load splitter):
24963 Use nonimmediate_operand instead of memory_operand for operand 1.
24964 (float-extend standard_x87sse_constant_load splitter): Ditto.
24966 2017-02-21 Jeff Law <law@redhat.com>
24968 PR tree-optimization/79621
24969 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
24970 blocks with edges to themselves.
24972 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24975 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
24976 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
24977 Use gimple_call_builtin_p.
24980 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
24981 on temporarily removed DEBUG_INSNs.
24983 PR tree-optimization/79649
24984 * tree-loop-distribution.c (classify_partition): Give up on
24985 non-generic address space loads/stores.
24987 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
24989 * doc/loop.texi (Loop manipulation): Remove nonexistent
24990 tree_ssa_loop_version from the documentation.
24991 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
24993 2017-02-21 Jakub Jelinek <jakub@redhat.com>
24996 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
24997 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
24998 * config/rs6000/rs6000.c: Include except.h.
24999 (rs6000_expand_split_stack_prologue): Call
25000 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
25002 2017-02-21 Martin Jambor <mjambor@suse.cz>
25005 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
25006 have been analyzed.
25008 2017-02-21 Martin Jambor <mjambor@suse.cz>
25010 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
25011 for backward compatibility only.
25012 * doc/invoke.texi (Option Summary): Remove all references to
25013 -fipa-cp-alignment.
25015 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
25019 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25021 * lra-constraints.c (curr_insn_transform): Handle
25022 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
25024 2017-02-21 Martin Liska <mliska@suse.cz>
25026 * config/i386/i386.opt: Replace -masm-dialect with -masm.
25028 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
25030 PR translation/79638
25031 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
25033 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
25036 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
25037 (arm_function_ok_for_sibcall): Return false for an indirect call by
25038 descriptor if all the argument registers are used.
25039 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
25040 alignment of the function.
25042 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25044 PR tree-optimization/61441
25045 * simplify-rtx.c (simplify_const_unary_operation): For
25046 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
25047 the sNaN unmodified.
25049 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25051 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25052 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25053 instead of SYSTEM_HEADER_DIR.
25055 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
25056 Martin Liška <mliska@suse.cz>
25058 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
25059 Fix typos and grammar, use active voice, and clarify.
25061 2017-02-20 Marek Polacek <polacek@redhat.com>
25063 PR middle-end/79537
25064 * gimplify.c (gimplify_expr): Handle unused *&&L;.
25067 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
25069 2017-02-20 Jakub Jelinek <jakub@redhat.com>
25072 * config/i386/i386.c (ix86_expand_builtin): Handle
25073 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
25074 ix86_builtins_isa[fcode].isa as a requirement of those
25075 flags and any other flag in the bitmask.
25076 (ix86_init_mmx_sse_builtins): Use 0 instead of
25077 ~OPTION_MASK_ISA_64BIT as mask.
25078 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
25079 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
25080 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
25081 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
25083 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25086 * lra-constraints.c (split_reg): Check requested split mode
25087 is supported by the register.
25089 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25091 * lra-constraints.c (simplify_operand_subreg): Remove early
25094 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25097 * lra-constraints.c (curr_insn_transform): Tighten condition
25098 for converting SUBREG reloads from OP_OUT to OP_INOUT.
25100 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25103 * lra-constraints.c (curr_insn_transform): Handle
25104 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
25106 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
25109 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
25111 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
25113 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
25116 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
25119 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25121 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
25124 2017-02-18 Jakub Jelinek <jakub@redhat.com>
25126 * final.c (last_columnnum, override_columnnum): New variables.
25127 (final_start_function): Set last_columnnum, pass it to begin_prologue
25128 hook and pass 0 to dwarf2out_begin_prologue.
25129 (final_scan_insn): Update override_columnnum. Pass last_columnnum
25130 to source_line debug hook.
25131 (notice_source_line): Compute last_columnnum and for debug_column_info
25132 return true on column changes.
25133 * debug.h (struct gcc_debug_hooks): Add column argument to
25134 source_line and begin_prologue hooks.
25135 (debug_nothing_int_charstar_int_bool): Remove prototype.
25136 (debug_nothing_int_int_charstar,
25137 debug_nothing_int_int_charstar_int_bool): New prototypes.
25138 (dwarf2out_begin_prologue): Add column argument.
25139 * debug.c (do_nothing_debug_hooks): Adjust source_line and
25140 begin_prologue hooks.
25141 (debug_nothing_int_charstar_int_bool): Remove.
25142 (debug_nothing_int_int_charstar,
25143 debug_nothing_int_int_charstar_int_bool): New functions.
25144 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
25145 through to dwarf2out_source_line.
25146 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
25147 (dwarf2out_source_line): Add column argument, emit it if requested.
25148 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
25150 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
25151 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
25152 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
25153 through to dwarf2out_begin_prologue.
25154 (vmsdbgout_source_line): Add column argument, pass it through to
25155 dwarf2out_source_line.
25156 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
25157 dbxout_source_line caller.
25158 (dbxout_source_line): Add column argument.
25160 * common.opt (gno-column-info, gcolumn-info): New options.
25161 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
25162 (check_die): Also test for multiple DW_AT_decl_column attributes.
25163 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
25164 DW_AT_decl_column if requested.
25165 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
25167 (gen_variable_die): Likewise.
25168 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
25169 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
25172 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
25173 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
25174 (ix86_handle_option): Handle OPT_m3dnowa.
25175 * doc/invoke.texi (-m3dnowa): Document.
25176 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
25177 -m3dnowa instead of -m3dnow -march=athlon.
25180 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
25181 instead of gcc_assert for K, r and R code checks. Formatting fixes.
25183 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25186 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
25187 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
25188 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
25189 generator for vsx_xxpermdi_<mode>_be.
25190 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
25191 force big-endian semantics.
25192 (vsx_xxpermdi_<mode>_be): New define_expand with same
25193 implementation as previous version of vsx_xxpermdi_<mode>.
25195 2017-02-17 Jakub Jelinek <jakub@redhat.com>
25197 PR tree-optimization/79327
25198 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
25199 variable, its initialization and use.
25201 2017-02-17 Julia Koval <julia.koval@intel.com>
25203 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
25204 (OPTION_MASK_ISA_PKU_UNSET): New.
25205 (ix86_handle_option): Handle -mrdpid.
25206 * config/i386/cpuid.h (bit_RDPID): New.
25207 * config/i386/driver-i386.c (host_detect_local_cpu):
25208 Detect RDPID feature.
25209 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
25210 * config/i386/i386-c.c (ix86_target_macros_internal):
25212 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
25213 (ix86_valid_target_attribute_inner_p): Add "rdpid".
25214 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
25215 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
25216 * config/i386/i386.md (define_insn "rdpid"): New.
25217 * config/i386/i386.opt Add -mrdpid.
25218 * config/i386/immintrin.h (_rdpid_u32): New.
25220 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
25222 PR rtl-optimization/79541
25223 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
25224 instead of transforming it into USE.
25226 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
25228 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
25229 If HONOR_SNANS (SFmode) force the input to a register.
25230 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
25231 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
25232 an frsp or similar insn.
25234 2017-02-17 Martin Liska <mliska@suse.cz>
25236 PR rtl-optimization/79577
25237 * params.def (selsched-max-sched-times): Increase minimum to 1.
25239 2017-02-17 Martin Liska <mliska@suse.cz>
25241 PR rtl-optimization/79574
25242 * gcse.c (want_to_gcse_p): Prevent integer overflow.
25244 2017-02-17 Martin Liska <mliska@suse.cz>
25246 PR tree-optimization/79529
25247 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
25248 ssa_defined_default_def_p to handle cases which are implicitly
25250 * tree-ssa.c (ssa_defined_default_def_p): New function.
25251 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
25252 which are implicitly defined.
25253 * tree-ssa.h (ssa_defined_default_def_p): Declare.
25255 2017-02-17 Richard Biener <rguenther@suse.de>
25257 PR middle-end/79576
25258 * params.def (max-ssa-name-query-depth): Limit to 10.
25260 2017-02-17 Richard Biener <rguenther@suse.de>
25262 PR tree-optimization/79552
25263 * tree-ssa-structalias.c (visit_loadstore): Properly verify
25266 2017-02-17 Richard Biener <rguenther@suse.de>
25269 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
25271 2017-02-17 Marek Polacek <polacek@redhat.com>
25273 PR middle-end/79536
25274 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
25275 (fold_negate_expr): New wrapper.
25277 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
25279 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
25280 Correct terminology and de-emphasize pre-standard behavior.
25282 2017-02-16 Alan Modra <amodra@gmail.com>
25284 PR rtl-optimization/79286
25285 * ira.c (def_dominates_uses): New function.
25286 (update_equiv_regs): Don't create an equivalence for insns that
25287 may trap where the register def does not dominate the use.
25289 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
25291 PR rtl-optimization/78127
25292 * lra.c (lra): Call lra_eliminate before finish the loop after
25295 2017-02-16 Richard Biener <rguenther@suse.de>
25297 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
25299 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
25300 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
25301 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
25302 (isl_val_int_from_wi): New function.
25303 (extract_affine_gmp): Rename to ...
25304 (extract_affine_wi): ... this, take a widest_int.
25305 (extract_affine_int): Just wrap extract_affine_wi.
25306 (add_param_constraints): Use isl_val_int_from_wi.
25307 (add_loop_constraints): Likewise, and extract_affine_wi.
25309 2017-02-15 Jeff Law <law@redhat.com>
25311 PR middle-end/79521
25312 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
25313 ira_init_register_move_cost_if_necessary.
25315 2017-02-15 Martin Sebor <msebor@redhat.com>
25317 PR middle-end/32003
25318 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
25319 removed in a prior commit.
25321 2017-02-15 Bin Cheng <bin.cheng@arm.com>
25323 PR tree-optimization/79347
25324 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
25325 counters during peeling.
25327 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
25329 * Makefile.in (site.exp): Remove "set ISLVER".
25331 2017-02-15 Jakub Jelinek <jakub@redhat.com>
25334 * real.c (real_from_integer): Call real_convert even for decimal.
25336 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
25339 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
25341 2017-02-14 Andrew Pinski <apinski@cavium.com>
25343 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
25344 cores and change the partno/implementer to be correct.
25345 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
25346 the 'B" as the implementer.
25347 * config/aarch64/aarch64-tune.md: Regenerate.
25349 2017-02-14 Carl Love <cel@us.ibm.com>
25351 * config/rs6000/rs6000.c: Add case statement entry to make the
25352 xvcvuxdsp built-in argument unsigned.
25353 * config/rs6000/vsx.md: Fix the source and return operand types so they
25354 match the instruction definitions from the ISA document. Fix typo
25355 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
25358 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
25361 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
25362 member early_clobber_alts.
25363 * lra-lives.c (reg_early_clobber_p): New.
25364 (process_bb_lives): Use it.
25365 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
25366 (debug_operand_data): Initialize early_clobber_alts.
25367 (setup_operand_alternative): Set up early_clobber_alts.
25368 (collect_non_operand_hard_regs): Ditto. Pass early clobber
25369 alternatives to new_insn_reg.
25370 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
25372 (lra_update_insn_regno_info): Pass the new arg.
25374 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25376 PR middle-end/79505
25377 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
25378 (new_oacc_loop_raw): Don't clear already cleared fields.
25381 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
25382 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
25383 _mm512_prefetch_i64gather_ps): New inline functions and macros.
25385 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
25388 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
25390 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
25393 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
25394 the extra instruction to the right place to store 128-bit constant
25397 2017-02-14 Martin Sebor <msebor@redhat.com>
25399 PR middle-end/79448
25400 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
25401 warning for strings of unknown length.
25403 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
25405 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
25407 2017-02-14 Jeff Law <law@redhat.com>
25410 * ira-costs.c (scan_one_insn): Initialize register move costs
25411 for pseudos seen in USE/CLOBBER insns.
25413 PR tree-optimization/79095
25414 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
25415 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
25416 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
25417 if the operands are known to be not equal, then the resulting range
25419 (intersect_ranges): If the new range is ~[0,0] and the old range is
25420 wide, then prefer ~[0,0].
25421 * tree-vrp.c (overflow_comparison_p_1): New function.
25422 (overflow_comparison_p): New function.
25423 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
25424 if NAME is used in an overflow test.
25425 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
25426 overflow check that can be expressed as an equality test, then adjust
25427 ops to be that equality test.
25429 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25431 * config/s390/s390-builtin-types.def: Remove flags argument.
25432 * config/s390/s390.c (s390_init_builtins): Likewise.
25434 2017-02-14 Martin Liska <mliska@suse.cz>
25436 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
25437 vector. Fix trailing white spaces.
25439 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
25441 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
25444 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25446 PR rtl-optimization/68664
25447 * config/arm/arm.c (arm_sched_can_speculate_insn):
25448 New function. Declare prototype.
25449 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25451 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25453 PR rtl-optimization/68664
25454 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
25456 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25458 2017-02-14 Amit Pawar <amit.pawar@amd.com>
25460 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
25461 max skip bytes for function, loop and jump.
25463 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25465 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
25466 ABS_EXPR for gimple dump.
25468 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25471 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
25473 PR tree-optimization/79408
25474 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
25475 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
25476 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
25477 also if rhs1 is INTEGER_CST.
25479 2017-02-14 Richard Biener <rguenther@suse.de>
25481 PR middle-end/79432
25482 * tree-into-ssa.c (insert_phi_nodes): When the function can
25483 have abnormal edges rewrite SSA names with broken use-def
25484 dominance out of SSA and register them for PHI insertion.
25486 2017-02-13 Martin Sebor <msebor@redhat.com>
25488 PR middle-end/79496
25489 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
25490 clearing info.nowrite flag when snprintf size argument is a range.
25492 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25494 * cprop.c (cprop_jump): Add missing space in string literal.
25495 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
25496 (get_constraint_for_component_ref): Likewise.
25497 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
25498 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
25499 * lra-constraints.c (process_alt_operands): Likewise.
25500 * ipa-inline.c (inline_small_functions): Likewise.
25501 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
25502 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25503 * trans-mem.c (diagnose_tm_1_op): Likewise.
25504 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
25505 (grid_parallel_clauses_gridifiable): Likewise.
25507 * config/nvptx/mkoffload.c (process): Add space in between
25510 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
25511 "MOD4_SSE_REGS" and "ALL_REGS".
25513 * spellcheck.c (test_data): Add , in between "foo" and "food".
25515 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25518 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
25519 boundary crossing check and subsequent code generation agree.
25521 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25523 * config/aarch64/aarch64.c (has_memory_op): Delete.
25524 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
25527 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25529 PR rtl-optimization/79388
25530 PR rtl-optimization/79450
25531 * combine.c (distribute_notes): When removing TEM_INSN for which
25532 corresponding dest has last value recorded, invalidate that last
25535 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25537 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
25538 of explicit '@'. Add missing assembly comment marker on branch costs
25541 2017-02-13 Nathan Sidwell <nathan@acm.org>
25543 * gengtype-lex.l (<in_struct>): Add '/'.
25545 2017-02-13 Martin Liska <mliska@suse.cz>
25548 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
25550 2017-02-13 Richard Biener <rguenther@suse.de>
25552 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
25554 * configure: Re-generate.
25555 * config.in: Likewise.
25556 * graphite-dependences.c: Simplify as if
25557 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
25558 * graphite-isl-ast-to-gimple.c: Likewise.
25559 * graphite-optimize-isl.c: Likewise.
25560 * graphite-poly.c: Likewise.
25561 * graphite-sese-to-poly.c: Likewise.
25562 * graphite.h: Likewise.
25563 * toplev.c: Include isl/version.h and use isl_version () for
25564 printing the ISL version.
25565 * doc/install.texi: Update ISL requirement.
25567 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25569 * doc/standards.texi (Standards): Update reference to
25572 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25574 * doc/extend.texi (Named Address Spaces): sourceware.org now
25576 * doc/install.texi (Binaries): Ditto.
25579 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25581 * doc/cpp.texi: Replace "stringify"/"stringification" with C
25582 standard terminology "stringize"/"stringizing" throughout.
25583 * doc/cppinternals.texi: Likewise.
25585 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25587 * doc/extend.texi: Fix some spelling mistakes and typos.
25588 * doc/invoke.texi: Likewise.
25590 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25593 * params.def (inline-min-speedup) Change from 10 to 8.
25595 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25597 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
25600 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25603 * ipa-inline-analysis.c (get_minimal_bb): New function.
25604 (record_modified): Use it.
25605 (remap_edge_change_prob): Handle also ancestor functions.
25607 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
25609 * doc/contrib.texi (Contributors): Remove broken link into
25610 the Mauve CVS repository.
25612 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25614 PR middle-end/79454
25615 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
25616 result computation whenever lhs doesn't have vector mode, not
25617 just when it has BLKmode.
25619 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25621 * doc/makefile.texi (profiledbootstrap): Refer to the
25622 installation instructions only in textual form.
25624 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25627 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
25629 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25631 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
25632 (Specific): Update mingw-w64 reference.
25634 (Specific): Remove broken link to Renesas RX processor.
25636 2017-02-10 Richard Biener <rguenther@suse.de>
25638 * toplev.c (process_options): Do not mention obsolete graphite
25639 options when printing sorry message about missing graphite support.
25640 Mention -floop-nest-optimize.
25642 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
25644 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
25645 (vtst_p16): Likewise.
25646 (vtstq_p8): Likewise.
25647 (vtstq_p16): Likewise.
25649 (vtstq_p64): Likewise.
25650 * config/arm/arm_neon.h (vgetq_lane_p64): New.
25651 (vset_lane_p64): New.
25652 (vsetq_lane_p64): New.
25654 2017-02-10 Jakub Jelinek <jakub@redhat.com>
25656 PR tree-optimization/79411
25657 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
25658 stmt operands are SSA_NAMEs used in abnormal phis.
25659 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
25662 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25665 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
25668 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25670 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
25672 2017-02-09 Jakub Jelinek <jakub@redhat.com>
25674 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
25678 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
25679 not arbitrary TREE_CONSTANT.
25682 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
25683 "omp declare target link" attribute unless is_global_var.
25684 * omp-offload.c (find_link_var_op): Likewise.
25686 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
25687 Chung-Lin Tang <cltang@codesourcery.com>
25689 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
25691 (gimplify_adjust_omp_clauses): Don't delete TILE.
25692 (gimplify_omp_for): Deal with TILE.
25693 * internal-fn.c (expand_GOACC_TILE): New function.
25694 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
25696 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
25697 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
25699 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
25700 avoid DIV for outermost collapse var.
25701 (expand_oacc_for): Insert tile element loop as needed. Adjust.
25702 Remove out of date comments, fix whitespace.
25703 * omp-general.c (omp_extract_for_data): Deal with tiling.
25704 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
25705 adjust OLF_DIM_BASE value.
25706 (struct omp_for_data): Add tiling field.
25707 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
25708 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
25709 for auto loops. Remove default auto determining, moved to
25710 oacc_loop_fixed_partitions.
25711 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
25712 stmts, add e_mask field.
25713 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
25714 (oacc_thread_numbers): Use oacc_dim_call.
25715 (oacc_xform_tile): New.
25716 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
25717 (finish_oacc_loop): Adjust for ifns vector.
25718 (oacc_loop_discover_walk): Append loop abstraction sites to list,
25719 add case for GOACC_TILE fns.
25720 (oacc_loop_xform_loop): Delete.
25721 (oacc_loop_process): Iterate over call list directly, and add
25722 handling for GOACC_TILE fns.
25723 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
25725 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
25726 vector partitioning to outer loops. Assign 2 partitions to loops
25727 when available. Add TILE handling.
25728 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
25729 (execite_oacc_device_lower): Process GOACC_TILE fns,
25730 ignore unknown specs.
25731 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
25732 * tree.c (omp_clause_num_ops): Adjust TILE ops.
25733 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
25735 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
25737 * configure.ac (ACX_BUGURL): Update.
25738 * configure: Regenerate.
25740 2017-02-09 Richard Biener <rguenther@suse.de>
25742 PR tree-optimization/69823
25743 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
25744 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
25746 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25748 * config/arc/arc-c.def: Add __NPS400__ definition.
25749 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
25750 (TARGET_NPS400): Define.
25752 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
25754 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
25756 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
25757 pointer, arch_info.
25758 (arc_cpu_types): Fill the arch_info field with a pointer into the
25759 arc_arch_types table.
25760 (arc_selected_cpu): Declare.
25761 * config/arc/arc.c (arc_selected_cpu): Make global.
25762 (arc_selected_arch): Delete.
25763 (arc_base_cpu): Delete.
25764 (arc_override_options): Remove references to deleted variables,
25765 update access to arch information.
25766 (ARC_OPT): Update access to arch information.
25767 (ARC_OPTX): Likewise.
25768 * config/arc/arc.h (arc_base_cpu): Remove declaration.
25769 (TARGET_ARC600): Update access to arch information.
25770 (TARGET_ARC601): Likewise.
25771 (TARGET_ARC700): Likewise.
25772 (TARGET_EM): Likewise.
25773 (TARGET_HS): Likewise.
25774 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
25777 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
25780 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
25781 condition/operands for integer GE/LE/GEU/LEU operations.
25783 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
25785 PR translation/79397
25786 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
25789 2017-02-08 Martin Jambor <mjambor@suse.cz>
25792 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
25793 whether allocation happened.
25794 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
25795 nothing was allocated.
25797 2017-02-08 Jakub Jelinek <jakub@redhat.com>
25799 PR tree-optimization/79408
25800 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
25801 constant, but SSA_NAME with a known integer range, use the minimum
25802 of that range instead of op1 to determine if modulo can be replaced
25803 with its first operand.
25805 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25807 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
25809 2017-02-08 Richard Biener <rguenther@suse.de>
25811 PR tree-optimization/71824
25812 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
25813 Check all loops contained in the merged region.
25815 2017-02-07 Andrew Pinski <apinski@cavium.com>
25817 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
25819 2017-02-07 Andrew Pinski <apinski@cavium.com>
25821 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
25822 (thunderxt88): Likewise.
25823 (thunderxt81): Disable LSE and change v8.1 to v8.
25824 (thunderxt83): Likewise.
25826 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25827 Richard Biener <rguenther@suse.de>
25829 PR middle-end/79399
25830 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
25831 type from int to size_t.
25832 * ira-costs.c (struct_costs_size): Change type from int to size_t.
25834 2017-02-07 Jakub Jelinek <jakub@redhat.com>
25836 PR rtl-optimization/79386
25837 * cprop.c (bypass_conditional_jumps): Initialize
25838 bypass_last_basic_block already before splitting bbs after
25839 unconditional traps...
25840 (bypass_conditional_jumps): ... rather than here.
25843 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
25844 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
25845 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
25846 fix -masm=intel patterns.
25848 2017-02-07 Richard Biener <rguenther@suse.de>
25850 PR tree-optimization/79256
25851 PR middle-end/79278
25852 * builtins.c (get_object_alignment_2): Use min_align_of_type
25853 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
25854 and ADJUST_FIELD_ALIGN.
25856 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
25858 * doc/tm.texi: Regenerate.
25859 * stor-layout.c (layout_decl): Adjust.
25860 (update_alignment_for_field): Likewise.
25861 (place_field): Likewise.
25862 (min_align_of_type): Likewise.
25863 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
25864 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
25865 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
25866 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
25867 * config/frv/frv.c (frv_adjust_field_align): Likewise.
25868 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
25869 * config/i386/i386.c (x86_field_alignment): Likewise.
25870 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
25871 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
25872 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
25873 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
25874 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
25875 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
25879 2017-01-30 Richard Biener <rguenther@suse.de>
25881 PR tree-optimization/79256
25882 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
25883 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
25886 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
25888 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
25889 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
25890 builtins to SImode and emit a zero-extend, if necessary.
25892 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25894 * docs/invoke.texi (RISC-V Options): Alphabetize.
25896 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25898 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
25901 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
25903 * config/riscv/riscv.c: New file.
25904 * common/config/riscv/riscv-common.c: Likewise.
25905 * config.gcc: Likewise.
25906 * config/riscv/constraints.md: Likewise.
25907 * config/riscv/elf.h: Likewise.
25908 * config/riscv/generic.md: Likewise.
25909 * config/riscv/linux.h: Likewise.
25910 * config/riscv/multilib-generator: Likewise.
25911 * config/riscv/peephole.md: Likewise.
25912 * config/riscv/pic.md: Likewise.
25913 * config/riscv/predicates.md: Likewise.
25914 * config/riscv/riscv-builtins.c: Likewise.
25915 * config/riscv/riscv-c.c: Likewise.
25916 * config/riscv/riscv-ftypes.def: Likewise.
25917 * config/riscv/riscv-modes.def: Likewise.
25918 * config/riscv/riscv-opts.h: Likewise.
25919 * config/riscv/riscv-protos.h: Likewise.
25920 * config/riscv/riscv.h: Likewise.
25921 * config/riscv/riscv.md: Likewise.
25922 * config/riscv/riscv.opt: Likewise.
25923 * config/riscv/sync.md: Likewise.
25924 * config/riscv/t-elf-multilib: Likewise.
25925 * config/riscv/t-linux: Likewise.
25926 * config/riscv/t-linux-multilib: Likewise.
25927 * config/riscv/t-riscv: Likewise.
25928 * configure.ac: Likewise.
25929 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
25930 Waterman as RISC-V maintainers.
25931 * doc/install.texi: Add RISC-V entries.
25932 * doc/invoke.texi: Add RISC-V options section.
25933 * doc/md.texi: Add RISC-V constraints section.
25934 * configure: Regenerated.
25936 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
25939 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
25940 false values to be constant vectors with all 0 or all 1 bits set.
25941 (vcondu<mode><mode>): Likewise.
25942 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
25944 (fpmask_comparison_operator): Update comment.
25945 (vecint_comparison_operator): New predicate.
25946 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
25947 vector conditionals when the true and false values are constant
25948 vectors with all 0 bits or all 1 bits set.
25950 2017-02-06 Martin Sebor <msebor@redhat.com>
25952 PR tree-optimization/79376
25953 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
25955 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
25957 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
25958 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
25959 to simplify split condition.
25961 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25963 * omp-expand.c (oxpand_omp_atomic_fetch_op,
25964 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
25967 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
25969 PR rtl-optimization/68664
25970 * target.def (can_speculate_insn): New hook.
25971 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
25972 * doc/tm.texi: Regenerate.
25973 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
25974 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
25975 (rs6000_sched_can_speculate_insn): New function.
25977 2017-02-06 Jakub Jelinek <jakub@redhat.com>
25979 PR tree-optimization/79284
25980 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
25981 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
25982 vectorizable_mask_load_store, vectorizable_operation,
25983 vect_is_simple_cond, get_same_sized_vectype): Use it instead
25984 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
25985 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
25986 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
25987 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
25988 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
25989 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
25990 is_gimple_assign (stmt). Replace another such test with
25991 is_gimple_assign (stmt).
25993 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
25996 * config/avr/avr.c (rtl-iter.h): Include it.
25997 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
25998 (avr_legitimate_combined_insn): ...and implementation.
26000 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
26002 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
26003 * config/s390/s390.c (s390_const_operand_ok)
26004 (s390_canonicalize_comparison, s390_extract_part)
26005 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
26006 (s390_contiguous_bitmask_p, s390_rtx_costs)
26007 (legitimize_pic_address): Likewise.
26008 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
26009 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
26010 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
26011 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
26012 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
26014 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
26016 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
26017 REGNO($0) == REGNO($1).
26019 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26021 * config/s390/linux.h(SIZE_TYPE): Add comment.
26023 2017-02-06 Julian Brown <julian@codesourcery.com>
26024 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26025 Virendra Pathak <virendra.pathak@broadcom.com>
26027 * config/aarch64/aarch64-cores.def: Change the scheduler
26029 * config/aarch64/aarch64.md: Include thunderx2t99.md.
26030 * config/aarch64/thunderx2t99.md: New file.
26032 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26034 * doc/standards.texi (Go Language): Update link to language
26037 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
26039 * tree-eh.c (lower_resx): Sanitize profile.
26040 (cleanup_empty_eh_move_lp): Likewise.
26042 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
26045 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
26047 * cfgloopmanip.h (loop_version): Update prototype.
26048 * modulo-sched.c (sms_schedule): Update call of loop_version.
26049 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
26050 * tree-parloops.c (gen_parallel_loop): Likewise.
26051 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
26052 * tree-ssa-loop-split.c (split_loop): Likewise.
26053 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
26054 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
26056 2017-02-05 Martin Liska <mliska@suse.cz>
26059 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
26061 (print_operand_address): Initialize a struct to zero.
26063 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26065 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
26066 garbage collector only in textual form.
26068 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26070 * doc/extend.texi (x86 specific memory model extensions for
26071 transactional memory): Simplify a phrase.
26073 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
26076 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
26077 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
26078 (atomic_storedi_1): Likewise.
26080 2017-02-04 Jakub Jelinek <jakub@redhat.com>
26082 PR tree-optimization/79338
26083 * tree-parloops.c (gather_scalar_reductions): Don't call
26084 vect_analyze_loop_form for loop->inner before destroying loop's
26087 2017-02-03 Martin Sebor <msebor@redhat.com>
26089 PR tree-optimization/79327
26090 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
26091 when precision has resulted in leading zeros.
26092 (format_integer): Adjust the likely counter to assume an unknown
26093 argument that may be zero is non-zero.
26095 2017-02-03 Jason Merrill <jason@redhat.com>
26098 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
26099 avoid copying non-taken branch.
26101 2017-02-03 Jakub Jelinek <jakub@redhat.com>
26103 PR tree-optimization/79340
26104 * tree-vect-loop.c (vectorizable_reduction): Release
26105 vec_defs elements after safe_splicing them into other vectors.
26108 PR tree-optimization/79327
26109 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
26110 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
26112 (format_integer): Use wide_int_to_tree instead of build_int_cst
26113 + to_?hwi. If argmin is NULL, just set argmin and argmax to
26114 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
26115 of shortest and longest sequence.
26117 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
26119 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
26120 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
26122 2017-02-03 Walter Lee <walt@tilera.com>
26125 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
26126 after initial stackframe link reg save.
26127 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
26129 2017-02-03 Jakub Jelinek <jakub@redhat.com>
26132 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
26133 wu for stxssp alternative.
26135 2017-02-03 Martin Sebor <msebor@redhat.com>
26137 PR tree-optimization/79352
26138 * gimple-fold.c (get_range_strlen): Add argument.
26139 (get_range_strlen): Change return type to bool.
26140 (get_maxval_strlen): Pass in a dummy argument.
26141 * gimple-fold.h (get_range_strlen): Change return type to bool.
26142 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
26143 * tree.h (array_at_struct_end_p): Add argument.
26144 * tree.c (array_at_struct_end_p): Handle it.
26146 2017-02-03 Martin Liska <mliska@suse.cz>
26149 * multiple_target.c (create_dispatcher_calls): Redirect edge
26150 from a caller of a dispatcher.
26151 (expand_target_clones): Make the clones local.
26152 (ipa_target_clone): Do both target clones and resolvers.
26153 (ipa_dispatcher_calls): Remove the pass.
26154 (pass_dispatcher_calls::gate): Likewise.
26155 (make_pass_dispatcher_calls): Likewise.
26156 * passes.def (pass_target_clone): Put as very first IPA early
26159 2017-02-03 Martin Liska <mliska@suse.cz>
26161 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
26162 in case of a function with ifunc attribute.
26164 2017-02-03 Martin Liska <mliska@suse.cz>
26166 * cgraph.c (cgraph_node::dump): Dump function version info.
26167 * symtab.c (symtab_node::dump_base): Add missing new line.
26169 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26171 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
26172 (ifcombine_ifandif): Use it.
26174 2017-02-03 Martin Liska <mliska@suse.cz>
26176 * doc/invoke.texi: Document default value for
26177 use-after-scope-direct-emission-threshold.
26179 2017-02-03 Martin Liska <mliska@suse.cz>
26181 PR tree-optimization/79339
26182 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
26183 (format_floating): Likewise.
26185 2017-02-03 Martin Liska <mliska@suse.cz>
26188 * ipa-prop.c (ipa_node_params_t::insert): Remove current
26190 (ipa_node_params_t::remove): Likewise.
26191 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
26192 initialization from removed ipa_node_params_t::insert.
26193 (ipa_node_params::~ipa_node_params): Move from removed
26194 ipa_node_params_t::release.
26195 * symbol-summary.h (symbol_summary::m_released): New member.
26196 Do not release a summary twice. Do not allow to call finalizer
26197 for types of a summary that live in GGC memory.
26199 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26201 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
26204 2017-02-02 Martin Sebor <msebor@redhat.com>
26206 PR middle-end/79275
26207 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
26208 (format_string): Tighten up the range of output for non-constant
26209 strings and correct the expected range for wide non-constant strings.
26211 2017-02-02 Martin Sebor <msebor@redhat.com>
26213 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
26215 PR middle-end/32003
26216 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
26218 (-fdump-tree-@var): Add to index and document how to come up
26219 with pass-specific option and dump file names.
26220 (-fdump-passes): Clarify where to look for output.
26222 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26224 PR middle-end/77445
26225 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
26226 statistics of the analyzed path; allow threading for speed when
26227 any of BBs along the path are optimized for speed.
26229 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
26231 PR middle-end/78468
26232 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
26233 settings of the virtual registers.
26236 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
26238 * explow.c (get_dynamic_stack_size): Take known alignment of stack
26239 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
26242 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26244 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
26245 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
26247 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26249 * config/s390/s390.md: Add missing comments with the expanded
26251 * config/s390/vector.md: Likewise.
26252 * config/s390/vx-builtins.md: Likewise.
26254 2017-02-02 Jakub Jelinek <jakub@redhat.com>
26257 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
26258 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
26259 conditions on a single line.
26261 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26263 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
26264 __S390_VX__ to __VX__.
26266 2017-02-01 Andrew Pinski <apinski@cavium.com>
26268 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
26269 stmt_info to record_stmt_cost.
26270 (vect_get_known_peeling_cost): Pass stmt_info if known to
26272 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
26273 cpu_vector_cost field into
26274 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
26275 field into vec_int_stmt_cost and vec_fp_stmt_cost.
26276 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
26277 splitting of scalar_stmt_cost and vec_stmt_cost.
26278 (thunderx_vector_cost): Likewise.
26279 (cortexa57_vector_cost): LIkewise.
26280 (exynosm1_vector_cost): Likewise.
26281 (xgene1_vector_cost): Likewise.
26282 (thunderx2t99_vector_cost): Improve after the splitting of the two
26284 (aarch64_builtin_vectorization_cost): Update for the splitting of
26285 scalar_stmt_cost and vec_stmt_cost.
26287 2017-02-01 Torvald Riegel <triegel@redhat.com>
26288 Richard Henderson <rth@redhat.com>
26290 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
26291 conditional on existance of a fast atomic load.
26292 * optabs-query.c (can_atomic_load_p): New function.
26293 * optabs-query.h (can_atomic_load_p): Declare it.
26294 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
26295 no fast atomic load is available for the particular size of access.
26296 (expand_atomic_compare_and_swap): Likewise.
26297 (expand_atomic_load): Likewise.
26298 (expand_atomic_store): Likewise.
26299 (expand_atomic_fetch_op): Likewise.
26300 * testsuite/lib/target-supports.exp
26301 (check_effective_target_sync_int_128): Remove x86 because it provides
26302 no fast atomic load.
26303 (check_effective_target_sync_int_128_runtime): Likewise.
26305 2017-02-01 Richard Biener <rguenther@suse.de>
26307 * graphite.c: Include tree-vectorizer.h for find_loop_location.
26308 (graphite_transform_loops): Provide opt-info for optimized nests.
26309 * tree-parloop.c (parallelize_loops): Provide opt-info for
26310 parallelized loops.
26312 2017-02-01 Richard Biener <rguenther@suse.de>
26314 PR middle-end/79315
26315 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
26316 was not set before.
26318 2017-02-01 Richard Biener <rguenther@suse.de>
26320 PR tree-optimization/71824
26321 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26322 Verify the loops are valid in the merged SESE region.
26323 (scop_detection::can_represent_loop_1): Check analyzing the
26324 evolution of the number of iterations in the region succeeds.
26326 2017-01-31 Ian Lance Taylor <iant@golang.org>
26328 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
26329 REG_ARGS_SIZE note to 32-bit push insns and call insn.
26331 2017-01-31 David Malcolm <dmalcolm@redhat.com>
26333 PR preprocessor/79210
26334 * input.c (get_substring_ranges_for_loc): Replace line_width
26335 assertion with error-handling.
26337 2017-01-31 Richard Biener <rguenther@suse.de>
26339 PR tree-optimization/77318
26340 * graphite-sese-to-poly.c (extract_affine): Fix assert.
26341 (create_pw_aff_from_tree): Take loop parameter.
26342 (add_condition_to_pbb): Pass loop of the condition to
26343 create_pw_aff_from_tree.
26345 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26347 * config/s390/s390.c (s390_asan_shadow_offset): New function.
26348 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
26350 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
26354 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
26356 (convert_int_to_float128): Likewise.
26357 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
26358 (convert_int_to_float128): Likewise.
26359 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
26360 (UNSPEC_IEEE128_CONVERT): Likewise.
26361 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
26362 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
26363 Use local variables for IBM extended format.
26364 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
26365 (fix_trunc<mode>si2_fprs): Likewise.
26366 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
26367 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
26368 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
26369 to know that we can now have integers of all sizes in vector
26371 (fix<uns>_<mode>di2_hw): Likewise.
26372 (float<uns>_<mode>si2_hw): Likewise.
26373 (fix_<mode>si2_hw): Likewise.
26374 (fixuns_<mode>si2_hw): Likewise.
26375 (float<uns>_<mode>di2_hw): Likewise.
26376 (float_<mode>di2_hw): Likewise.
26377 (float_<mode>si2_hw): Likewise.
26378 (floatuns_<mode>di2_hw): Likewise.
26379 (floatuns_<mode>si2_hw): Likewise.
26380 (xscvqp<su>wz_<mode>): Delete, no longer used.
26381 (xscvqp<su>dz_<mode>): Likewise.
26382 (xscv<su>dqp_<mode>): Likewise.
26383 (ieee128_mfvsrd_64bit): Likewise.
26384 (ieee128_mfvsrd_32bit): Likewise.
26385 (ieee128_mfvsrwz): Likewise.
26386 (ieee128_mtvsrw): Likewise.
26387 (ieee128_mtvsrd_64bit): Likewise.
26388 (ieee128_mtvsrd_32bit): Likewise.
26390 2017-01-31 Martin Liska <mliska@suse.cz>
26393 * ipa-prop.c (ipa_free_all_node_params): Call release method
26394 instead of ~sumbol_summary to not to trigger double times
26397 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
26399 PR tree-optimization/71691
26400 * bitmap.h (class auto_bitmap): New.
26401 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
26402 is_maybe_undefined instead of ssa_undefined_value_p.
26404 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26406 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
26407 __S390_ARCH_LEVEL__ to __ARCH__.
26409 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26411 PR tree-optimization/79267
26412 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
26413 if should_remove_lhs_p is true.
26415 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
26418 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
26419 (add_alignment_attribute): New.
26420 (base_type_die): Add alignment attribute.
26421 (subrange_type_die): Likewise.
26422 (modified_type_die): Likewise.
26423 (gen_array_type_die): Likewise.
26424 (gen_descr_array_type_die: Likewise.
26425 (gen_enumeration_type_die): Likewise.
26426 (gen_subprogram_die): Likewise.
26427 (gen_variable_die): Likewise.
26428 (gen_field_die): Likewise.
26429 (gen_ptr_to_mbr_type_die): Likewise.
26430 (gen_struct_or_union_type_die): Likewise.
26431 (gen_subroutine_type_die): Likewise.
26432 (gen_typedef_die): Likewise.
26433 (base_type_cmp): Compare alignment attribute.
26435 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26438 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
26439 (setb_unsigned) New pattern for setb with CCUNS.
26440 * config/rs6000/rs6000.c (expand_block_compare): Use a different
26441 subfc./subfe sequence to avoid overflow problems. Generate a
26442 shorter sequence with cmpld/setb for power9.
26443 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
26444 for generating subfc. instruction.
26445 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
26446 now uses this instruction.
26448 2017-01-30 Ian Lance Taylor <iant@google.com>
26451 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
26452 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
26454 2017-01-30 Martin Sebor <msebor@redhat.com>
26456 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
26457 Move constant to the right of a relational operator.
26458 (get_mpfr_format_length, format_character, format_string): Ditto.
26459 (should_warn_p, maybe_warn): Same.
26461 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
26463 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
26466 * asan.c (get_translation_unit_decl): Remove function.
26467 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
26469 2017-01-30 Martin Liska <mliska@suse.cz>
26471 PR gcov-profile/79259
26472 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
26473 -fprofile-generate.
26475 2017-01-30 Martin Liska <mliska@suse.cz>
26478 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
26479 Initialize variables with NULL value.
26481 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
26484 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
26486 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
26488 2017-01-30 Richard Biener <rguenther@suse.de>
26490 PR tree-optimization/79276
26491 * tree-vrp.c (process_assert_insertions): Properly adjust common
26492 when removing a duplicate.
26494 * gcc.dg/torture/pr79276.c: New testcase.
26496 2017-01-30 Richard Biener <rguenther@suse.de>
26498 PR tree-optimization/79256
26499 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26500 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26502 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
26504 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
26507 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
26508 ("*r<noxa>sbg_<mode>_sll_bitmask")
26509 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
26510 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
26511 Use contiguous_bitmask_nowrap_operand.
26513 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26516 * config/rs6000/altivec.h (vec_xl): Revise #define.
26517 (vec_xst): Likewise.
26519 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
26521 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
26523 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
26525 PR rtl-optimization/79194
26526 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
26527 traps before call to bypass_conditional_jumps.
26529 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26531 PR tree-optimization/71374
26532 * lra-constraints.c (check_conflict_input_operands): New.
26533 (match_reload): Use it.
26535 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26538 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
26539 account to calculate conflict_set.
26541 2017-01-27 Bin Cheng <bin.cheng@arm.com>
26543 PR rtl-optimization/78559
26544 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
26545 other_insn in combine.
26547 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26549 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
26550 uint16_type_node for BT_UINT16.
26552 2017-01-27 David Malcolm <dmalcolm@redhat.com>
26554 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
26555 "RTL Tests" to menu.
26556 (GIMPLE Tests): New node.
26557 (RTL Tests): New node.
26559 2017-01-27 Richard Biener <rguenther@suse.de>
26561 PR tree-optimization/79245
26562 * tree-loop-distribution.c (distribute_loop): Apply cost
26563 modeling also to detected patterns.
26565 2017-01-27 Richard Biener <rguenther@suse.de>
26567 PR tree-optimization/71433
26568 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
26569 (compare_assert_loc): New function.
26570 (process_assert_insertions): Sort and optimize assert locations
26571 to remove duplicates and push down identical assertions on
26572 edges to their destination block.
26574 2017-01-27 Richard Biener <rguenther@suse.de>
26576 PR tree-optimization/79244
26577 * tree-vrp.c (remove_range_assertions): Forcefully propagate
26578 out SSA names even if abnormal.
26580 2017-01-27 Jakub Jelinek <jakub@redhat.com>
26582 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
26583 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
26584 instead of MPFR_RNDN.
26586 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
26589 * arm.c (arm_option_override): Don't call build_target_option_node
26590 until after doing all option overrides.
26591 (arm_valid_target_attribute_tree): Likewise.
26593 2017-01-27 Martin Liska <mliska@suse.cz>
26595 * doc/invoke.texi (-fprofile-arcs): Document profiling support
26596 for {cd}tors and C++ {cd}tors.
26598 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
26600 * config/s390/s390.md ("*setmem_long_and")
26601 ("*setmem_long_and_31z"): Use zero_extend instead of and.
26603 2017-01-26 Martin Sebor <msebor@redhat.com>
26605 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
26608 2017-01-26 Martin Sebor <msebor@redhat.com>
26610 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
26611 HAVE_DFmode before using XFmode or DFmode.
26612 (parse_directive): Avoid using the z length modifier to avoid
26613 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
26615 PR middle-end/78703
26616 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
26617 to accept adjustment as an array.
26618 (get_int_range): New function.
26619 (struct directive): Make width and prec arrays.
26620 (directive::set_width, directive::set_precision): Call get_int_range.
26621 (format_integer, format_floating): Handle width and precision ranges.
26622 (format_string, parse_directive): Same.
26624 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26627 * dwarf2out.c (generate_skeleton_bottom_up): For children with
26628 comdat_type_p set, just clone them, but keep the children in the
26632 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
26633 which have direct callers with -fvar-tracking-assignments enabled
26635 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
26636 inside of type units.
26638 2017-01-26 Martin Sebor <msebor@redhat.com>
26640 PR middle-end/78703
26641 * gimple-ssa-sprintf.c (struct result_range): Add likely and
26643 (struct format_result): Replace number_chars, number_chars_min,
26644 and number_chars_max with a single member of struct result_range.
26646 (format_result::operator+=): Adjust.
26647 (struct fmtresult): Remove bounded. Handle likely and unlikely
26649 (fmtresult::adjust_for_width_or_precision): New function.
26650 (fmtresult:type_max_digits): New function.
26651 (bytes_remaining): Handle likely and unlikely counters.
26652 (min_bytes_remaining): Remove.
26653 (format_percent): Simplify.
26654 (format_integer, format_floating): Set likely and unlikely counters.
26655 (get_string_length, format_character, format_string): Same.
26656 (format_plain, should_warn_p): New function.
26657 (maybe_warn): Call should_warn_p. Update diagnostic messages
26658 and handle those for all directives, including plain strings.
26659 (format_directive): Handle likely and unlikely counters.
26660 Remove unnecessary quoting from diagnostics. Add an informational
26662 (add_bytes): Remove.
26663 (pass_sprintf_length::compute_format_length): Simplify.
26664 (try_substitute_return_value): Handle likely and unlikely counters.
26666 2017-01-26 Carl Love <cel@us.ibm.com>
26668 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
26669 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
26671 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
26674 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
26675 endianess for subregs into account.
26676 * lra-constraints.c (lra_constraints): Do risky transformations
26677 always on the first iteration.
26678 * lra-lives.c (check_pseudos_live_through_calls): Add arg
26679 last_call_used_reg_set.
26680 (process_bb_lives): Define and use last_call_used_reg_set.
26681 * lra.c (lra): Always continue after lra_constraints on the first
26684 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
26686 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
26688 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
26690 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26692 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
26693 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
26694 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
26695 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
26696 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
26697 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
26698 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
26699 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
26700 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
26702 2017-01-26 Marek Polacek <polacek@redhat.com>
26705 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
26706 for the third operand.
26708 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26710 PR middle-end/79236
26711 * omp-low.c (struct omp_context): Add simt_stmt field.
26712 (scan_omp_for): Return omp_context *.
26713 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
26714 context to the _simt_ SIMD stmt.
26715 (lower_omp_for): For combined SIMD with sibling _simt_
26716 SIMD, make sure to use the same decls in _looptemp_
26717 clauses as in the sibling.
26719 2017-01-26 David Sherwood <david.sherwood@arm.com>
26721 PR middle-end/79212
26722 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
26725 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26728 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
26729 emit fld b; fld a; if possible.
26731 * brig-builtins.def: Update copyright years.
26732 * config/arm/arm_acle_builtins.def: Update copyright years.
26734 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
26737 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
26738 constraint instead of o for the stxsd instruction.
26740 2017-01-25 Carl Love <cel@us.ibm.com>
26742 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
26743 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
26745 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
26747 * doc/invoke.texi (C++ Dialect Options): Fix typo.
26749 2017-01-25 Richard Biener <rguenther@suse.de>
26751 PR tree-optimization/69264
26752 * target.def (vector_alignment_reachable): Improve documentation.
26753 * doc/tm.texi: Regenerate.
26754 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
26756 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
26757 earlier changes with respect to TYPE_USER_ALIGN.
26758 (vector_alignment_reachable_p): Likewise. Improve dumping.
26760 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26763 * config/arm/arm.md (xordi3): Force constant operand into a register
26766 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26768 * doc/invoke.texi (-fstore-merging): Correct default optimization
26769 levels at which it is enabled.
26770 (-O): Move -fstore-merging from list to...
26773 2017-01-25 Richard Biener <rguenther@suse.de>
26776 * omp-expand.c: Include debug.h.
26777 (expand_omp_taskreg): Make sure to generate early debug before
26778 outlining anything from a function.
26779 (expand_omp_target): Likewise.
26780 (grid_expand_target_grid_body): Likewise.
26782 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
26785 * asan.c (get_translation_unit_decl): New function.
26786 (asan_add_global): Extract modules file name from globals
26787 TRANSLATION_UNIT_DECL name.
26789 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
26792 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
26793 for long calls with APCS frame and VFP.
26795 2017-01-24 David Malcolm <dmalcolm@redhat.com>
26797 * cfg.c (original_copy_tables_initialized_p): New function.
26798 * cfg.h (original_copy_tables_initialized_p): New decl.
26799 * cfgrtl.c (relink_block_chain): Guard the call to
26800 free_original_copy_tables with a call to
26801 original_copy_tables_initialized_p.
26802 * cgraph.h (symtab_node::native_rtl_p): New decl.
26803 * cgraphunit.c (symtab_node::native_rtl_p): New function.
26804 (symtab_node::needed_p): Don't assert for early assembly output
26805 for __RTL functions.
26806 (cgraph_node::finalize_function): Set "force_output" for __RTL
26808 (cgraph_node::analyze): Bail out early for __RTL functions.
26809 (analyze_functions): Update assertion to support __RTL functions.
26810 (cgraph_node::expand): Bail out early for __RTL functions.
26811 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
26813 * function.h (struct function): Update comment for field
26815 * gimple-expr.c: Include "tree-pass.h".
26816 (gimple_has_body_p): Return false for __RTL functions.
26817 * Makefile.in (OBJS): Add run-rtl-passes.o.
26818 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
26820 (gcc::pass_manager::get_clean_slate): New accessor.
26821 * passes.c: Include "insn-addr.h".
26822 (should_skip_pass_p): Add logging. Update logic for running
26823 "expand" to be compatible with both __GIMPLE and __RTL. Guard
26824 property-provider override so it is only done for gimple passes.
26826 (skip_pass): New function.
26827 (execute_one_pass): Call skip_pass when skipping passes.
26828 * read-md.c (md_reader::read_char): Support filtering
26829 the input to a subset of line numbers.
26830 (md_reader::md_reader): Initialize fields
26831 m_first_line and m_last_line.
26832 (md_reader::read_file_fragment): New function.
26833 * read-md.h (md_reader::read_file_fragment): New decl.
26834 (md_reader::m_first_line): New field.
26835 (md_reader::m_last_line): New field.
26836 * read-rtl-function.c (function_reader::create_function): Only
26837 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
26838 curr_properties. Set DECL_INITIAL to a dummy block.
26839 (read_rtl_function_body_from_file_range): New function.
26840 * read-rtl-function.h (read_rtl_function_body_from_file_range):
26842 * run-rtl-passes.c: New file.
26843 * run-rtl-passes.h: New file.
26845 2017-01-24 Jeff Law <law@redhat.com>
26847 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
26850 2017-01-24 Bin Cheng <bin.cheng@arm.com>
26852 PR tree-optimization/79159
26853 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
26854 (record_nonwrapping_iv): Improve boundary using above function if no
26855 value range information.
26857 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
26858 Martin Jambor <mjambor@suse.cz>
26860 * brig-builtins.def: New file.
26861 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
26862 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
26863 (DEF_HSAIL_SAT_BUILTIN): Likewise.
26864 (DEF_HSAIL_INTR_BUILTIN): Likewise.
26865 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
26866 * builtin-types.def (BT_INT8): New.
26867 (BT_INT16): Likewise.
26868 (BT_UINT8): Likewise.
26869 (BT_UINT16): Likewise.
26870 (BT_FN_ULONG): Likewise.
26871 (BT_FN_UINT_INT): Likewise.
26872 (BT_FN_UINT_ULONG): Likewise.
26873 (BT_FN_UINT_LONG): Likewise.
26874 (BT_FN_UINT_PTR): Likewise.
26875 (BT_FN_ULONG_PTR): Likewise.
26876 (BT_FN_INT8_FLOAT): Likewise.
26877 (BT_FN_INT16_FLOAT): Likewise.
26878 (BT_FN_UINT32_FLOAT): Likewise.
26879 (BT_FN_UINT16_FLOAT): Likewise.
26880 (BT_FN_UINT8_FLOAT): Likewise.
26881 (BT_FN_UINT64_FLOAT): Likewise.
26882 (BT_FN_UINT16_UINT32): Likewise.
26883 (BT_FN_UINT32_UINT16): Likewise.
26884 (BT_FN_UINT16_UINT16_UINT16): Likewise.
26885 (BT_FN_INT_PTR_INT): Likewise.
26886 (BT_FN_UINT_PTR_UINT): Likewise.
26887 (BT_FN_LONG_PTR_LONG): Likewise.
26888 (BT_FN_ULONG_PTR_ULONG): Likewise.
26889 (BT_FN_VOID_UINT64_UINT64): Likewise.
26890 (BT_FN_UINT8_UINT8_UINT8): Likewise.
26891 (BT_FN_INT8_INT8_INT8): Likewise.
26892 (BT_FN_INT16_INT16_INT16): Likewise.
26893 (BT_FN_INT_INT_INT): Likewise.
26894 (BT_FN_UINT_FLOAT_UINT): Likewise.
26895 (BT_FN_FLOAT_UINT_UINT): Likewise.
26896 (BT_FN_ULONG_UINT_UINT): Likewise.
26897 (BT_FN_ULONG_UINT_PTR): Likewise.
26898 (BT_FN_ULONG_ULONG_ULONG): Likewise.
26899 (BT_FN_UINT_UINT_UINT): Likewise.
26900 (BT_FN_VOID_UINT_PTR): Likewise.
26901 (BT_FN_UINT_UINT_PTR: Likewise.
26902 (BT_FN_UINT32_UINT64_PTR): Likewise.
26903 (BT_FN_INT_INT_UINT_UINT): Likewise.
26904 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
26905 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
26906 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
26907 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
26908 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
26909 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
26910 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
26911 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
26912 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
26913 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
26914 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
26915 * doc/frontends.texi: List BRIG FE.
26916 * doc/install.texi (Testing): Add BRIG tesring requirements.
26917 * doc/invoke.texi (Overall Options): Mention BRIG.
26918 * doc/standards.texi (Standards): Doucment BRIG HSA version.
26920 2017-01-24 Richard Biener <rguenther@suse.de>
26922 PR translation/79208
26923 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
26925 2017-01-24 Martin Jambor <mjambor@suse.cz>
26928 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
26929 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
26930 and known_contexts.
26932 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
26934 PR middle-end/79123
26935 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
26936 casts from signed to unsigned really don't have a range.
26938 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
26940 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
26941 GMP_RNDx for compatiblity.
26943 2017-01-24 Martin Liska <mliska@suse.cz>
26946 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
26947 that would prevent us to call alloca with -1 as argument.
26949 2017-01-24 Jakub Jelinek <jakub@redhat.com>
26951 * dwarf2out.c (output_compilation_unit_header, output_file_names):
26952 Avoid -Wformat-security warning.
26954 2017-01-23 Andrew Pinski <apinski@cavium.com>
26956 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
26959 2017-01-23 Martin Sebor <msebor@redhat.com>
26961 PR middle-end/78703
26962 * gimple-ssa-sprintf.c (warn_level): New global.
26963 (format_integer): Use it here and throughout the rest of the file.
26964 Use the same switch to compute sign as base.
26965 (maybe_warn): New function.
26966 (format_directive): Factor out warnings into maybe_warn.
26967 Add debugging output. Use warn_level.
26968 (add_bytes): Use warn_level.
26969 (pass_sprintf_length::compute_format_length): Add debugging output.
26970 (try_substitute_return_value): Same.
26971 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
26973 PR middle-end/78703
26974 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
26975 (struct fmtresult, format_integer, format_floating): Adjust.
26976 (fmtresult::fmtresult): Set max correctly in two argument ctor.
26977 (get_string_length, format_string,format_directive): Same.
26978 (pass_sprintf_length::compute_format_length): Same.
26979 (try_substitute_return_value): Simplify slightly.
26981 PR middle-end/78703
26982 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
26983 (fmtresult::operator+=): Outlined.
26984 (struct fmtresult): Add ctors.
26985 (struct conversion_spec): Rename...
26986 (struct directive): ...to this. Add and remove data members.
26987 (directive::set_width, directive::set_precision): New functions.
26988 (format_percent): Use fmtresult ctor.
26989 (get_width_and_precision): Remove.
26990 (format_integer): Make naming changes. Avoid computing width and
26992 (format_floating): Same. Adjust indentation.
26993 (format_character, format_none): New functions.
26994 (format_string): Moved character handling to format_character.
26995 (format_directive): Remove arguments, change return type.
26996 (parse_directive): New function.
26997 (pass_sprintf_length::compute_format_length): Move directive
26998 parsing to parse_directive.
27000 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27002 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
27003 (assign_assembler_name_if_needed): ... this.
27004 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
27005 (assign_assembler_name_if_needed): ... this.
27006 (free_lang_data_in_cgraph): Adjust callers.
27007 * cgraphunit.c (cgraph_node::analyze): Likewise.
27008 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
27011 2017-01-23 Richard Biener <rguenther@suse.de>
27013 PR tree-optimization/79088
27014 PR tree-optimization/79188
27015 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
27016 resetting loop bounds after last path deletion. Reset loop
27017 bounds of the target loop, make code match the comments.
27018 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
27019 Make sure loops need no fixups.
27021 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27023 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
27024 exponent support with double type for first argument.
27025 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
27026 type returned by __builtin_vec_extract_sig,
27027 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
27028 functions from "vector int" to "vector unsigned int" or from
27029 "vector long long int" to "vector unsigned long long int".
27030 Changed type returned by __builtin_vec_extract_exp,
27031 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
27032 functions from "vector int" to "vector unsigned int" or from
27033 "vector long long int" to "vector unsigned long long int".
27034 Changed return type of __builtin_vec_test_data_class,
27035 __builtin_vec_test_data_class_sp, and
27036 __builtin_vec_test_data_class_dp from "vector int" to
27037 "vector bool int" or from "vector long long int" to "vector bool
27038 long long int" and changed second argument type from "unsigned
27039 int" to "int". Added new overloaded function forms "vector float
27040 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
27041 "vector float __builtin_vec_insert_exp_sp (vector float, vector
27042 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
27043 double, vector unsigned long long int)" and "vector double
27044 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
27045 long int)". Changed return type of
27046 __builtin_scalar_test_data_class and
27047 __builtin_scalar_test_data_class_sp and
27048 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
27049 int" and changed second argument from "unsigned int" to "int".
27050 Changed type returned by __builtin_scalar_test_neg,
27051 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
27052 from "int" to "bool int". Added new overloaded function form
27053 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
27054 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
27055 exponent double-precision with floating point first argument.
27056 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
27057 documentation of scalar_test_data_class, scalar_test_neg,
27058 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
27059 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
27060 vec_test_data_class built-in functions to reflect refinements in
27061 their type signatures.
27063 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
27065 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
27067 (aarch64_elf_asm_destructor): Likewise.
27069 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
27071 PR rtl-optimization/78634
27072 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
27073 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
27074 * ifcvt.c (noce_try_cmove): Add missing cost check.
27076 PR rtl-optimization/71724
27077 * combine.c (if_then_else_cond): Look for situations where it is
27078 beneficial to undo the work of one of the recursive calls.
27080 2017-01-23 Bin Cheng <bin.cheng@arm.com>
27082 PR tree-optimization/70754
27083 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
27084 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
27085 combined stmt before it if not NULL.
27086 (combine_chains): Process refs reversely and compute dominance point
27089 2017-01-23 Martin Liska <mliska@suse.cz>
27091 PR tree-optimization/79196
27092 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
27093 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
27095 (strlen_optimize_stmt): Call the renamed function.
27097 2017-01-23 Michael Matz <matz@suse.de>
27099 PR tree-optimization/78384
27100 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
27102 2017-01-23 Richard Biener <rguenther@suse.de>
27104 PR tree-optimization/79186
27105 * tree-vrp.c (register_new_assert_for): Make sure we've seen
27106 both incoming edges before moving an assert.
27108 2017-01-23 Martin Jambor <mjambor@suse.cz>
27110 * ipa-prop.c (load_from_param_1): Removed.
27111 (load_from_unmodified_param): Bits from load_from_param_1 put back
27113 (load_from_param): Removed.
27114 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
27115 with stmt. Reverted back to use of load_from_unmodified_param.
27117 2017-01-23 Martin Jambor <mjambor@suse.cz>
27120 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
27121 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
27122 field a pointer to garbage collected vector, mark lattices and
27123 ipcp_orig_node with GTY((skip)).
27124 (ipa_get_param_count): Adjust to descriptors being a pointer.
27125 (ipa_get_param): Likewise.
27126 (ipa_get_type): Likewise.
27127 (ipa_get_param_move_cost): Likewise.
27128 (ipa_set_param_used): Likewise.
27129 (ipa_get_controlled_uses): Likewise.
27130 (ipa_set_controlled_uses): Likewise.
27131 (ipa_is_param_used): Likewise.
27132 (ipa_node_params_t): Move into garbage collector. New methods insert
27134 (ipa_node_params_sum): Annotate wth GTY(()).
27135 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
27137 (ipa_load_from_parm_agg): Adjust declaration.
27138 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
27139 * ipa-profile.c (ipa_profile): Likewise.
27140 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
27141 (ipa_populate_param_decls): Make descriptors parameter garbage
27143 (ipa_dump_param): Adjust to descriptors being a pointer.
27144 (ipa_alloc_node_params): Likewise.
27145 (ipa_initialize_node_params): Likewise.
27146 (load_from_param_1): Make descriptors parameter garbage collected.
27147 (load_from_unmodified_param): Likewise.
27148 (load_from_param): Likewise.
27149 (ipa_load_from_parm_agg): Likewise.
27150 (ipa_node_params::~ipa_node_params): Removed.
27151 (ipa_free_all_node_params): Remove call to delete operator.
27152 (ipa_node_params_t::insert): New.
27153 (ipa_node_params_t::remove): Likewise.
27154 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
27155 copy known_csts and known_contexts vectors.
27156 (ipa_read_node_info): Adjust to descriptors being a pointer.
27157 (ipcp_modif_dom_walker): Make m_descriptors field garbage
27159 (ipcp_transform_function): Make descriptors variable garbage
27162 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
27164 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
27165 * config/i386/avx512dqintrin.h: Ditto.
27166 * config/i386/avx512fintrin.h: Ditto.
27167 * config/i386/i386.c: Handle new builtins.
27168 * config/i386/i386-builtin.def: Add new builtins.
27169 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
27170 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
27172 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27173 Martin Liska <mliska@suse.cz>
27175 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
27176 * asan.c (asan_expand_poison_ifn): Support stores and use
27177 appropriate ASAN report function.
27178 * internal-fn.c (expand_ASAN_POISON_USE): New function.
27179 * internal-fn.def (ASAN_POISON_USE): Declare.
27180 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
27181 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
27182 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
27183 ASAN_POISON calls w/o LHS.
27184 * tree-ssa.c (execute_update_addresses_taken): Create clobber
27185 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
27186 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
27187 * gimplify.c (asan_poison_variables): Add attribute
27188 use_after_scope_memory to variables that really needs to live
27190 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
27191 having the attribute.
27193 2017-01-23 Martin Liska <mliska@suse.cz>
27195 * asan.c (create_asan_shadow_var): New function.
27196 (asan_expand_poison_ifn): Likewise.
27197 * asan.h (asan_expand_poison_ifn): New declaration.
27198 * internal-fn.c (expand_ASAN_POISON): Likewise.
27199 * internal-fn.def (ASAN_POISON): New builtin.
27200 * sanopt.c (pass_sanopt::execute): Expand
27201 asan_expand_poison_ifn.
27202 * tree-inline.c (copy_decl_for_dup_finish): Make function
27204 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
27205 * tree-ssa.c (is_asan_mark_p): New function.
27206 (execute_update_addresses_taken): Rewrite local variables
27207 (identified just by use-after-scope as addressable) into SSA.
27209 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27211 * doc/install.texi (Specific): opensource.apple.com uses https
27212 now. Remove trailing slash.
27214 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27216 * README.Portability: Remove note on an Irix compatibility issue.
27218 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
27220 * gcov.c (INCLUDE_ALGORITHM): Define.
27221 (INCLUDE_VECTOR): Define.
27222 No longer include <vector> and <algorithm> directly.
27224 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
27226 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
27228 * doc/invoke.texi (Code Gen Options): Ditto.
27230 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
27233 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
27235 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
27237 rtl-optimization/79125
27238 * cprop.c (local_cprop_pass): Handle cases where we make an
27239 unconditional trap.
27241 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
27245 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
27246 read from, for big endian.
27248 2017-01-20 Jiong Wang <jiong.wang@arm.com>
27250 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
27251 register pauth builtins for LP64 only.
27253 2017-01-20 Marek Polacek <polacek@redhat.com>
27256 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
27259 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
27261 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
27263 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
27264 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
27265 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
27267 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27270 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
27271 in big-endian mode when they are not a single duplicated value.
27273 2017-01-20 Richard Biener <rguenther@suse.de>
27275 * BASE-VER: Bump to 7.0.1.
27277 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
27279 * omp-low.c (omplow_simd_context): New struct. Use it...
27280 (lower_rec_simd_input_clauses): ...here and...
27281 (lower_rec_input_clauses): ...here to hold common data. Adjust all
27282 references to idx, lane, max_vf, is_simt.
27284 2017-01-20 Graham Markall <graham.markall@embecosm.com>
27286 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
27289 2017-01-20 Martin Jambor <mjambor@suse.cz>
27291 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
27292 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
27294 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
27295 (GTFILES): Rename hsa.c to hsa-common.c.
27296 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
27297 * hsa-dump.c: Likewise.
27298 * hsa-gen.c: Likewise.
27299 * hsa-regalloc.c: Likewise.
27300 * ipa-hsa.c: Likewise.
27301 * omp-expand.c: Likewise.
27302 * omp-low.c: Likewise.
27303 * toplev.c: Likewise.
27305 2017-01-20 Marek Polacek <polacek@redhat.com>
27308 * doc/invoke.texi: Document -Wduplicated-branches.
27309 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
27310 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
27311 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
27312 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
27313 return 0 only when not OEP_LEXICOGRAPHIC.
27314 (fold_build_cleanup_point_expr): Use the expression
27315 location when building CLEANUP_POINT_EXPR.
27316 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
27317 * tree.c (add_expr): Handle error_mark_node.
27319 2017-01-20 Martin Liska <mliska@suse.cz>
27322 * tree-profile.c (init_ic_make_global_vars): Do not call
27324 (gimple_init_gcov_profiler): Likewise.
27326 2017-01-20 Martin Liska <mliska@suse.cz>
27329 * cgraph.h (maybe_create_reference): Remove argument and
27331 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
27333 * ipa-cp.c (create_specialized_node): Likewise.
27334 * symtab.c (symtab_node::maybe_create_reference): Handle
27335 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
27337 2017-01-20 Martin Liska <mliska@suse.cz>
27339 * read-rtl-function.c (function_reader::create_function): Use
27340 build_decl instread of build_decl_stat.
27342 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
27344 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
27345 * config/i386/avx512dqintrin.h: Ditto.
27346 * config/i386/avx512fintrin.h: Ditto.
27347 * config/i386/i386-builtin-types.def: Add new types.
27348 * config/i386/i386.c: Handle new types.
27349 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
27350 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
27351 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
27352 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
27353 (__builtin_ia32_kshiftridi): New.
27354 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
27356 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
27360 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
27361 define to rs6000_init_stack_protect_guard.
27362 (rs6000_init_stack_protect_guard): New function.
27364 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27365 Yunqiang Su <yunqiang.su@imgtec.com>
27367 * config.gcc (supported_defaults): Add madd4.
27368 (with_madd4): Add validation.
27369 (all_defaults): Add madd4.
27370 * config/mips/mips.opt (mmadd4): New option.
27371 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27373 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27375 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
27376 (ISA_HAS_FUSED_MADD4): Likewise.
27377 * doc/invoke.texi (-mmadd4): Document the new option.
27378 * doc/install.texi (--with-madd4): Document the new option.
27380 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27382 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
27383 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
27384 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
27385 (aarch64_init_pauth_hint_builtins): New.
27386 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
27387 (aarch64_expand_builtin): Expand new builtins.
27389 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27391 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
27392 * combine-stack-adj.c (no_unhandled_cfa): Handle
27393 REG_CFA_TOGGLE_RA_MANGLE.
27394 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
27395 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
27396 info for return address signing.
27397 (aarch64_expand_epilogue): Likewise.
27399 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27401 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
27402 * config/aarch64/aarch64-protos.h
27403 (aarch64_return_address_signing_enabled): New declaration.
27404 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
27406 (aarch64_expand_prologue): Sign return address before it's pushed onto
27408 (aarch64_expand_epilogue): Authenticate return address fetched from
27410 (aarch64_override_options): Sanity check for ILP32 and ISA level.
27411 (aarch64_attributes): New function attributes for "sign-return-address".
27412 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
27413 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
27414 ("*do_return"): Generate combined instructions according to key index.
27415 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
27416 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
27418 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
27419 * config/aarch64/aarch64.opt (msign-return-address=): New.
27420 * doc/extend.texi (AArch64 Function Attributes): Documents
27421 "sign-return-address=".
27422 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
27424 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27426 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
27427 overall option summary.
27429 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27431 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
27432 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
27433 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
27434 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
27436 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
27438 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
27439 -mpower9-minmax by default for -mcpu=power9.
27440 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
27441 128-bit floating point.
27443 2017-01-20 Alan Modra <amodra@gmail.com>
27445 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
27446 optimizing for size.
27448 2017-01-20 Alan Modra <amodra@gmail.com>
27451 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
27452 for strcmp and strncmp from corresponding builtin decl.
27454 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27456 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
27457 instead of i386/rtems-64.h.
27458 * config/i386/rtems-64.h: Remove.
27460 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27464 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
27466 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
27468 2017-01-19 Tamar Christina <tamar.christina@arm.com>
27470 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
27471 Change int to HOST_WIDE_INT.
27472 * config/aarch64/aarch64-protos.h
27473 (aarch64_simd_gen_const_vector_dup): Likewise.
27474 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
27476 2017-01-19 David Malcolm <dmalcolm@redhat.com>
27478 * langhooks-def.h (lhd_type_for_size): New decl.
27479 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
27480 * langhooks.c (lhd_type_for_size): New function, taken from
27483 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
27485 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
27486 define_bypass for CR latency.
27487 (power9-cracked-alu): Update bypass latency and remove power9-branch.
27488 (power9-alu2): Add define_bypass for CR latency.
27490 (power9-mul): Update insn latency.
27491 (power9-mul-compare): Update insn latency, bypass latency and remove
27494 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27496 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
27498 * config/aarch64/aarch64.md
27499 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
27500 aarch64_nopcrelative_literal_loads.
27501 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
27503 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
27505 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
27506 TARGET_LOONGSON_3A.
27507 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
27509 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
27512 * config.gcc (supported_defaults): Add lxc1-sxc1.
27513 (with_lxc1_sxc1): Add validation.
27514 (all_defaults): Add lxc1-sxc1.
27515 * config/mips/mips.opt (mlxc1-sxc1): New option.
27516 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27518 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27519 __mips_no_lxc1_sxc1.
27520 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
27521 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
27522 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
27524 2017-01-19 Richard Biener <rguenther@suse.de>
27526 PR tree-optimization/72488
27527 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
27528 sure to restore SSA info.
27529 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
27531 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
27533 PR rtl-optimization/79121
27534 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
27535 of the inner type when shifting an extended value.
27537 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27540 * symtab.c (symtab_node::equal_address_to): Fix comparing of
27541 interposable aliases.
27543 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
27546 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
27547 Use the evmergelohi instruction.
27548 (mov_si<mode>_e500_subreg4_2_le): Likewise.
27549 (mov_sitf_e500_subreg8_2_be): Likewise.
27550 (mov_sitf_e500_subreg12_2_le): Likewise.
27551 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
27552 (mov_si<mode>_e500_subreg4_2_be): Likewise.
27553 (mov_sitf_e500_subreg8_2_le): Likewise.
27554 (mov_sitf_e500_subreg12_2_be): Likewise.
27556 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27558 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
27559 attribute from vecsimple to vecperm.
27560 (altivec_vbpermq2): Likewise.
27562 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27565 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
27567 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27568 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
27569 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
27570 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
27571 case where N arg is SIZE_MAX.
27572 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
27573 (cmpstrsi): Add pattern.
27575 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27577 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27578 __builtin_vec_revb builtins.
27579 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
27580 built-in functions to support generation of the ISA 3.0 XXBR<x>
27581 vector byte reverse instructions.
27582 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
27583 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
27584 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
27585 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
27586 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
27587 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
27588 (P9V_BUILTIN_VEC_REVB): Likewise.
27589 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
27590 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
27591 (p9_xxbrq_v16qi): Likewise.
27592 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
27593 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
27594 (p9_xxbrh_v8hi): Likewise.
27595 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
27596 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
27597 vec_revb built-in functions.
27599 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
27601 PR rtl-optimization/78952
27602 * config/i386/i386.md (any_extract): New code iterator.
27603 (*insvqi_2): Use any_extract for source operand.
27604 (*insvqi_3): Use any_shiftrt for source operand.
27606 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
27608 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
27610 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
27612 2017-01-18 Matthias Klose <doko@ubuntu.com>
27614 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
27616 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27618 * config/rs6000/altivec.h (vec_bperm): Change #define.
27619 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
27620 (altivec_vbpermq2): New define_insn.
27621 (altivec_vbpermd): Likewise.
27622 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
27623 function interface.
27624 (VBPERMD): Likewise.
27625 (VBPERM): New polymorphic function interface.
27626 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
27627 Add entries for P9V_BUILTIN_VEC_VBPERM.
27628 * doc/extend.texi: Add interfaces for vec_bperm.
27630 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27632 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
27633 first letter of error messages.
27634 (s390_resolve_overloaded_builtin): Likewise.
27635 * config/s390/s390.c (s390_expand_builtin): Likewise.
27636 (s390_invalid_arg_for_unprototyped_fn): Likewise.
27637 (s390_valid_target_attribute_inner_p): Likewise.
27638 * config/s390/s390.md ("tabort"): Likewise.
27640 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
27642 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
27643 (ISA_AVOID_DIV_HILO): New macro.
27644 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
27645 (ISA_HAS_DDIV): Likewise.
27647 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27649 * doc/invoke.texi (fabi-version): Correct number of occurrences.
27651 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27653 * doc/invoke.texi (fabi-version): Spelling fix.
27655 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27658 * doc/invoke.texi (fabi-version): Mention mangling fix for
27661 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27664 * doc/invoke.texi (fabi-version): Document discriminator mangling.
27666 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
27669 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
27670 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
27672 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
27674 (stack_protect_test): Ditto.
27675 * config/rs6000/rs6000.opt (mstack-protector-guard=,
27676 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
27678 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
27679 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
27680 -mstack-protector-guard-offset=.
27681 (RS/6000 and PowerPC Options): Ditto.
27683 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27685 * config/i386/i386.h (MASK_CLASS_P): New define.
27686 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
27687 there are no registers from different register sets also when
27688 mask registers are used. Update function comment.
27689 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
27690 to (*k/*r) and (*k/*km) alternatives.
27692 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
27694 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
27695 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
27696 (EH_RETURN_HANDLER_RTX): New define.
27697 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
27698 Force frame pointer in EH return functions.
27699 (aarch64_expand_epilogue): Add barrier for eh_return.
27700 (aarch64_final_eh_return_addr): Remove.
27701 (aarch64_eh_return_handler_rtx): New function.
27702 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
27704 (aarch64_eh_return_handler_rtx): New prototype.
27706 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27708 * config/rs6000/altivec.h (vec_rlmi): New #define.
27709 (vec_vrlnm): Likewise.
27710 (vec_rlnm): Likewise.
27711 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
27712 (UNSPEC_VRLNM): Likewise.
27713 (VIlong): New mode iterator.
27714 (altivec_vrl<VI_char>mi): New define_insn.
27715 (altivec_vrl<VI_char>nm): Likewise.
27716 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
27718 (VRLDNM): Likewise.
27719 (RLNM): New polymorphic function entry.
27720 (VRLWMI): New monomorphic function entry.
27721 (VRLDMI): Likewise.
27722 (RLMI): New polymorphic function entry.
27723 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
27724 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
27725 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
27728 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27731 * dwarf2out.c (field_byte_offset): Restore the
27732 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
27733 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
27734 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
27737 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27740 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
27742 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27745 * dwarf2out.c (add_data_member_location_attribute): For constant
27746 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
27747 instead of DW_AT_data_member_location, DW_AT_bit_offset and
27748 DW_AT_byte_size attributes.
27750 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
27752 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
27753 after forcing to constant memory when the code model is medium.
27755 2017-01-17 Julia Koval <julia.koval@intel.com>
27758 * config/i386/avx512fintrin.h
27759 (_mm512_i32gather_ps): Change __addr type to void const*.
27760 (_mm512_mask_i32gather_ps): Ditto.
27761 (_mm512_i32gather_pd): Ditto.
27762 (_mm512_mask_i32gather_pd): Ditto.
27763 (_mm512_i64gather_ps): Ditto.
27764 (_mm512_mask_i64gather_ps): Ditto.
27765 (_mm512_i64gather_pd): Ditto.
27766 (_mm512_mask_i64gather_pd): Ditto.
27767 (_mm512_i32gather_epi32): Ditto.
27768 (_mm512_mask_i32gather_epi32): Ditto.
27769 (_mm512_i32gather_epi64): Ditto.
27770 (_mm512_mask_i32gather_epi64): Ditto.
27771 (_mm512_i64gather_epi32): Ditto.
27772 (_mm512_mask_i64gather_epi32): Ditto.
27773 (_mm512_i64gather_epi64): Ditto.
27774 (_mm512_mask_i64gather_epi64): Ditto.
27775 (_mm512_i32scatter_ps): Change __addr type to void*.
27776 (_mm512_mask_i32scatter_ps): Ditto.
27777 (_mm512_i32scatter_pd): Ditto.
27778 (_mm512_mask_i32scatter_pd): Ditto.
27779 (_mm512_i64scatter_ps): Ditto.
27780 (_mm512_mask_i64scatter_ps): Ditto.
27781 (_mm512_i64scatter_pd): Ditto.
27782 (_mm512_mask_i64scatter_pd): Ditto.
27783 (_mm512_i32scatter_epi32): Ditto.
27784 (_mm512_mask_i32scatter_epi32): Ditto.
27785 (_mm512_i32scatter_epi64): Ditto.
27786 (_mm512_mask_i32scatter_epi64): Ditto.
27787 (_mm512_i64scatter_epi32): Ditto.
27788 (_mm512_mask_i64scatter_epi32): Ditto.
27789 (_mm512_i64scatter_epi64): Ditto.
27790 (_mm512_mask_i64scatter_epi64): Ditto.
27791 * config/i386/avx512pfintrin.h
27792 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
27793 (_mm512_mask_prefetch_i32gather_ps): Ditto.
27794 (_mm512_mask_prefetch_i64gather_pd): Ditto.
27795 (_mm512_mask_prefetch_i64gather_ps): Ditto.
27796 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
27797 (_mm512_prefetch_i32scatter_ps): Ditto.
27798 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
27799 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
27800 (_mm512_prefetch_i64scatter_pd): Ditto.
27801 (_mm512_prefetch_i64scatter_ps): Ditto.
27802 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
27803 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
27804 * config/i386/avx512vlintrin.h
27805 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
27806 (_mm_mmask_i32gather_ps): Ditto.
27807 (_mm256_mmask_i32gather_pd): Ditto.
27808 (_mm_mmask_i32gather_pd): Ditto.
27809 (_mm256_mmask_i64gather_ps): Ditto.
27810 (_mm_mmask_i64gather_ps): Ditto.
27811 (_mm256_mmask_i64gather_pd): Ditto.
27812 (_mm_mmask_i64gather_pd): Ditto.
27813 (_mm256_mmask_i32gather_epi32): Ditto.
27814 (_mm_mmask_i32gather_epi32): Ditto.
27815 (_mm256_mmask_i32gather_epi64): Ditto.
27816 (_mm_mmask_i32gather_epi64): Ditto.
27817 (_mm256_mmask_i64gather_epi32): Ditto.
27818 (_mm_mmask_i64gather_epi32): Ditto.
27819 (_mm256_mmask_i64gather_epi64): Ditto.
27820 (_mm_mmask_i64gather_epi64): Ditto.
27821 (_mm256_i32scatter_ps): Change __addr type to void*.
27822 (_mm256_mask_i32scatter_ps): Ditto.
27823 (_mm_i32scatter_ps): Ditto.
27824 (_mm_mask_i32scatter_ps): Ditto.
27825 (_mm256_i32scatter_pd): Ditto.
27826 (_mm256_mask_i32scatter_pd): Ditto.
27827 (_mm_i32scatter_pd): Ditto.
27828 (_mm_mask_i32scatter_pd): Ditto.
27829 (_mm256_i64scatter_ps): Ditto.
27830 (_mm256_mask_i64scatter_ps): Ditto.
27831 (_mm_i64scatter_ps): Ditto.
27832 (_mm_mask_i64scatter_ps): Ditto.
27833 (_mm256_i64scatter_pd): Ditto.
27834 (_mm256_mask_i64scatter_pd): Ditto.
27835 (_mm_i64scatter_pd): Ditto.
27836 (_mm_mask_i64scatter_pd): Ditto.
27837 (_mm256_i32scatter_epi32): Ditto.
27838 (_mm256_mask_i32scatter_epi32): Ditto.
27839 (_mm_i32scatter_epi32): Ditto.
27840 (_mm_mask_i32scatter_epi32): Ditto.
27841 (_mm256_i32scatter_epi64): Ditto.
27842 (_mm256_mask_i32scatter_epi64): Ditto.
27843 (_mm_i32scatter_epi64): Ditto.
27844 (_mm_mask_i32scatter_epi64): Ditto.
27845 (_mm256_i64scatter_epi32): Ditto.
27846 (_mm256_mask_i64scatter_epi32): Ditto.
27847 (_mm_i64scatter_epi32): Ditto.
27848 (_mm_mask_i64scatter_epi32): Ditto.
27849 (_mm256_i64scatter_epi64): Ditto.
27850 (_mm256_mask_i64scatter_epi64): Ditto.
27851 (_mm_i64scatter_epi64): Ditto.
27852 (_mm_mask_i64scatter_epi64): Ditto.
27853 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
27854 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
27855 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
27856 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
27857 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
27858 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
27859 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
27860 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
27861 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
27862 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
27863 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
27864 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
27865 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
27866 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
27867 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
27868 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
27869 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
27870 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
27871 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
27872 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
27873 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
27874 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
27875 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
27876 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
27877 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
27878 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
27879 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
27880 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
27881 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
27882 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
27883 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
27884 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
27885 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
27886 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
27887 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
27888 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
27889 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
27890 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
27891 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
27892 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
27893 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
27894 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
27895 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
27896 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
27897 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
27898 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
27899 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
27900 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
27901 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
27902 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
27903 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
27904 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
27905 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
27906 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
27907 definitions accordingly.
27909 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
27910 Kuan-Lin Chen <kuanlinchentw@gmail.com>
27913 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
27916 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
27919 * ira-conflicts.c (ira_build_conflicts): Update total conflict
27920 hard regs for inner regno.
27922 2017-01-17 Martin Liska <mliska@suse.cz>
27925 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
27926 assumption and add comment.
27928 2017-01-17 Nathan Sidwell <nathan@acm.org>
27930 * ipa-visibility.c (localize_node): New function, broken out of ...
27931 (function_and_variable_visibility): ... here. Call it.
27933 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27935 PR middle-end/77445
27936 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
27937 correctly set frequency of oudgoing edge.
27938 (duplicate_thread_path): Fix profile updating.
27940 2017-01-17 Jakub Jelinek <jakub@redhat.com>
27943 * configure.ac: Add GCC_BASE_VER.
27944 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
27945 version from BASE-VER file.
27946 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
27947 (gcc.o): Depend on $(BASEVER).
27948 * common.opt (dumpfullversion): New option.
27949 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
27950 * doc/invoke.texi: Document -dumpfullversion.
27951 * doc/install.texi: Document --with-gcc-major-version-only.
27952 * configure: Regenerated.
27954 2017-01-17 Richard Biener <rguenther@suse.de>
27956 PR tree-optimization/71433
27957 * tree-vrp.c (register_new_assert_for): Merge same asserts
27958 on all incoming edges.
27959 (process_assert_insertions_for): Handle insertions at the
27962 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
27964 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
27965 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
27967 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
27970 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
27973 2017-01-17 Alan Modra <amodra@gmail.com>
27976 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
27977 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
27978 symbolic stack limit when pic.
27980 2017-01-16 Martin Sebor <msebor@redhat.com>
27982 PR tree-optimization/78608
27983 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
27985 2017-01-16 Jeff Law <law@redhat.com>
27988 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
27989 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
27990 for several include directories that may be relative to sysroot.
27991 * config/i386/x-mingw32 (gplus_includedir): Define.
27992 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
27993 (native_system_includedir): Likewise.
27994 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
27995 override if TARGET_SYSTEM_ROOT is defined.
27996 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
27998 PR tree-optimization/79090
27999 PR tree-optimization/33562
28000 PR tree-optimization/61912
28001 PR tree-optimization/77485
28002 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
28003 and computed trims into the dump file.
28005 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
28007 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
28009 2017-01-16 Jakub Jelinek <jakub@redhat.com>
28012 * gimplify.c (gimplify_init_constructor): If want_value and
28013 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
28017 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
28018 sequence. Formatting fixes.
28019 (doloop_optimize): Formatting fixes.
28022 * gcc.c (debug_level_greater_than_spec_func): New function.
28023 (static_spec_functions): Add debug-level-gt spec function.
28024 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
28026 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
28027 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
28028 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
28029 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
28030 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
28031 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
28033 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
28035 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
28036 QImode fixups to general and mask registers only.
28038 2017-01-16 Carl Love <cel@us.ibm.com>
28040 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
28041 for built-in functions
28042 vector signed char vec_nabs (vector signed char)
28043 vector signed short vec_nabs (vector signed short)
28044 vector signed int vec_nabs (vector signed int)
28045 vector signed long long vec_nabs (vector signed long long)
28046 vector float vec_nabs (vector float)
28047 vector double vec_nabs (vector double)
28048 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
28050 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
28051 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
28052 * doc/extend.texi: Update the documentation file for the new built-in
28055 2017-01-16 Martin Sebor <msebor@redhat.com>
28057 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
28060 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28062 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
28063 UNSPEC_VSX__XXSPLTD to require special splat handling.
28065 2017-01-16 David Malcolm <dmalcolm@redhat.com>
28068 * system.h: Poison strndup.
28070 2017-01-16 Alan Modra <amodra@gmail.com>
28073 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
28076 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
28078 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
28080 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
28082 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
28083 call recog here. Assert that INSN_CODE (insn) is non-negative.
28085 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28088 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
28090 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
28091 in the currently scheduled RTL fragment.
28093 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28095 PR rtl-optimization/78751
28096 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
28099 2017-01-14 Jeff Law <law@redhat.com>
28101 PR tree-optimization/79090
28102 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
28103 variable length stores.
28104 (compute_trims): Delete dead assignment to *trim_tail.
28105 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
28108 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
28110 PR rtl-optimization/78626
28111 PR rtl-optimization/78727
28112 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
28113 of a block, and split such blocks after everything else is finished.
28115 2017-01-14 Alan Modra <amodra@gmail.com>
28118 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
28119 target legitimate_combined_insn.
28120 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
28121 (rs6000_legitimate_combined_insn): New function.
28122 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
28124 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
28125 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
28126 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
28128 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
28130 * doc/frontends.texi (G++ and GCC): Remove references to Java.
28132 2017-01-13 Jeff Law <law@redhat.com>
28134 PR tree-optimization/33562
28135 PR tree-optimization/61912
28136 PR tree-optimization/77485
28137 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
28139 (delete_dead_assignment): Likewise.
28140 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
28141 statement to delete_dead_call and delete_dead_assignment.
28143 2017-01-13 David Malcolm <dmalcolm@redhat.com>
28146 * substring-locations.c (format_warning_va): Strengthen case 1 so
28147 that both endpoints of the substring must be within the format
28148 range for just the substring to be printed.
28150 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
28152 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
28153 * config/i386/i386.c (ix86_target_string): Add missing options
28154 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
28155 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
28156 flags_other and ix86_target_other to flags2_other. Display unknown
28158 (ix86_valid_target_attribute_inner_p): Add missing options and
28159 reorder options by implied ISAs, as in ix86_target_string.
28161 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28163 * hash-table.h (hash_table::too_empty_p): New function.
28164 (hash_table::expand): Use it.
28165 (hash_table::traverse): Likewise.
28166 (hash_table::empty_slot): Use sizeof (value_type) instead of
28167 sizeof (PTR) to convert bytes to elements. Shrink the table
28168 if the current size is excessive for the current number of
28171 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28173 * ira-costs.c (record_reg_classes): Break from the inner loop
28174 early once alt_fail is known to be true. Update outer loop
28175 handling accordingly.
28177 2017-01-13 Jeff Law <law@redhat.com>
28179 * tree-ssa-dse.c (decrement_count): New function.
28180 (increment_start_addr, maybe_trim_memstar_call): Likewise.
28181 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
28182 when we know the partially dead statement is a mem* function.
28184 PR tree-optimization/61912
28185 PR tree-optimization/77485
28186 * tree-ssa-dse.c: Include expr.h.
28187 (maybe_trim_constructor_store): New function.
28188 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
28190 PR tree-optimization/33562
28191 PR tree-optimization/61912
28192 PR tree-optimization/77485
28193 * doc/invoke.texi: Document new dse-max-object-size param.
28194 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
28195 * tree-ssa-dse.c: Include params.h.
28196 (dse_store_status): New enum.
28197 (initialize_ao_ref_for_dse): New, partially extracted from
28199 (valid_ao_ref_for_dse, normalize_ref): New.
28200 (setup_live_bytes_from_ref, compute_trims): Likewise.
28201 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
28202 (maybe_trim_partially_dead_store): Likewise.
28203 (maybe_trim_complex_store): Likewise.
28204 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
28205 Track what bytes live from the original store. Return tri-state
28206 for dead, partially dead or live.
28207 (dse_dom_walker): Add constructor, destructor and new private members.
28208 (delete_dead_call, delete_dead_assignment): New extracted from
28210 (dse_optimize_stmt): Make a member of dse_dom_walker.
28211 Use initialize_ao_ref_for_dse.
28213 PR tree-optimization/33562
28214 PR tree-optimization/61912
28215 PR tree-optimization/77485
28216 * sbitmap.h (bitmap_count_bits): Prototype.
28217 (bitmap_clear_range, bitmap_set_range): Likewise.
28218 * sbitmap.c (bitmap_clear_range): New function.
28219 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
28221 2017-01-13 Martin Liska <mliska@suse.cz>
28224 * function.c (set_cfun): Add new argument force.
28225 * function.h (set_cfun): Likewise.
28226 * ipa-inline-transform.c (inline_call): Use the function when
28227 strict alising from is dropped for function we inline to.
28229 2017-01-13 Richard Biener <rguenther@suse.de>
28231 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
28232 for dumping GIMPLE INTEGER_CSTs.
28234 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28236 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
28237 to 201112L since C++17.
28239 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
28242 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
28243 if -fsanitize=kernel-address is present.
28245 2017-01-13 Richard Biener <rguenther@suse.de>
28247 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
28248 as _Literal ( type ) number in case usual suffixes do not
28249 preserve all information.
28251 2017-01-13 Richard Biener <rguenther@suse.de>
28253 PR tree-optimization/77283
28254 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
28255 and ssa-iterators.h.
28256 (is_feasible_trace): Implement a cost model based on joiner
28259 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
28262 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
28263 char or short to __float128/_Float128 directly.
28265 2017-01-12 Martin Sebor <msebor@redhat.com>
28267 to -Wformat-overflow.
28268 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
28269 (min_bytes_remaining): Same.
28270 (get_string_length): Same.
28271 (format_string): Same.
28272 (format_directive): Same.
28274 (pass_sprintf_length::handle_gimple_call): Same.
28276 2017-01-12 Jakub Jelinek <jakub@redhat.com>
28278 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
28279 info.nowrite calls with no lhs that can't throw. Return bool
28280 whether gsi_remove has been called or not.
28281 (pass_sprintf_length::handle_gimple_call): Return bool whether
28282 try_substitute_return_value called gsi_remove. Formatting fix.
28283 (pass_sprintf_length::execute): Don't use gsi_remove if
28284 handle_gimple_call returned true.
28287 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
28288 be removed due to side-effects, don't remove following barrier nor
28289 turn the successor edge into fallthru edge.
28291 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28294 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
28295 element-reversing loads and stores as not swappable.
28297 2017-01-12 Nathan Sidwell <nathan@acm.org>
28298 Nicolai Stange <nicstange@gmail.com>
28300 * combine.c (try_combine): Don't ignore result of overlap checking
28301 loop. Combine overlap & asm check into single loop.
28303 2017-01-12 Richard Biener <rguenther@suse.de>
28305 * tree-pretty-print.c (dump_generic_node): Provide -gimple
28306 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
28308 2017-01-12 Richard Biener <rguenther@suse.de>
28310 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
28311 and TS_TARGET_OPTION directly derive from TS_BASE.
28312 * tree-core.h (tree_optimization_option): Derive from tree_base.
28313 (tree_target_option): Likewise.
28315 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
28317 * config/i386/i386.c (memory_address_length): Increase len
28318 only when rip_relative_addr_p returns false.
28320 2017-01-11 Julia Koval <julia.koval@intel.com>
28322 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
28323 (OPTION_MASK_ISA_SGX_SET): New.
28324 (ix86_handle_option): Handle OPT_msgx.
28325 * config.gcc: Added sgxintrin.h.
28326 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
28327 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
28328 * config/i386/i386.c (ix86_target_string): Add -msgx.
28330 (ix86_option_override_internal): Handle new options.
28331 (ix86_valid_target_attribute_inner_p): Add sgx.
28332 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
28333 * config/i386/i386.opt: Add msgx.
28334 * config/i386/sgxintrin.h: New file.
28335 * config/i386/x86intrin.h: Add sgxintrin.h.
28337 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28340 * fold-const.c (maybe_nonzero_address): Return 1 for function
28342 (tree_single_nonzero_warnv_p): Don't handle function local objects
28346 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
28349 2017-01-11 David Malcolm <dmalcolm@redhat.com>
28352 * opts.c: Include "spellcheck.h"
28353 (struct string_fragment): New struct.
28354 (struct edit_distance_traits<const string_fragment &>): New
28356 (get_closest_sanitizer_option): New function.
28357 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
28359 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28361 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
28363 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
28364 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
28365 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
28366 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
28367 for initial die_offset if dwarf_split_debug_info.
28368 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
28369 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
28371 (output_skeleton_debug_sections): Formatting fix. Use
28372 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
28373 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
28375 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
28377 * config/arm/cortex-a53.md: Add bypasses for
28378 cortex_a53_r2f_cvt.
28379 (cortex_a53_r2f): Only use for transfers.
28380 (cortex_a53_f2r): Likewise.
28381 (cortex_a53_r2f_cvt): Add reservation for conversions.
28382 (cortex_a53_f2r_cvt): Likewise.
28384 2017-01-11 Tamar Christina <tamar.christina@arm.com>
28386 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
28387 to all inlined functions, change static to extern.
28389 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
28392 * config/arm/arm.c (legitimize_pic_address): Handle reference to
28394 (arm_assemble_integer): Likewise.
28396 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
28398 * config.gcc: Use new awk script to check CPU, FPU and architecture
28399 parameters for --with-... options.
28400 * config/arm/parsecpu.awk: New file
28401 * config/arm/arm-cpus.in: New file.
28402 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
28404 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
28406 * config/arm/t-arm: Update dependency rules.
28407 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
28408 of processing .def files.
28409 * config/arm/genopt.sh: Deleted.
28410 * config/arm/gentune.sh: Deleted.
28411 * config/arm/arm-cores.def: Deleted.
28412 * config/arm/arm-arches.def: Deleted.
28413 * config/arm/arm-fpus.def: Deleted.
28414 * config/arm/arm-tune.md: Regenerated.
28415 * config/arm/arm-tables.opt: Regenerated.
28416 * config/arm/arm-cpu.h: New generated file.
28417 * config/arm/arm-cpu-data.h: New generated file.
28418 * config/arm/arm-cpu-cdata.h: New generated file.
28420 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
28423 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
28425 (input_varpool_node): Unpack dynamically_initialized bit.
28427 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
28429 PR rtl-optimization/79032
28430 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
28431 the alignment of the adjusted memory reference against that of MODE,
28432 instead of the alignment of the original memory reference.
28434 2017-01-11 Martin Jambor <mjambor@suse.cz>
28436 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
28438 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
28439 decorated functions.
28441 2017-01-11 Richard Biener <rguenther@suse.de>
28443 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
28444 set range/nonnull info for PHI results. Do not set it on
28445 stmts marked for removal.
28447 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
28449 * expr.c (store_field): In the bitfield case, fetch the return value
28450 from the registers before applying a single big-endian adjustment.
28451 Always do a final load for a BLKmode value not larger than a word.
28453 2017-01-10 David Malcolm <dmalcolm@redhat.com>
28456 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28457 that we correctly handle column numbers greater than
28458 LINE_MAP_MAX_COLUMN_NUMBER.
28460 2017-01-10 Martin Sebor <msebor@redhat.com>
28462 PR middle-end/78245
28463 * gimple-ssa-sprintf.c (get_destination_size): Call
28464 {init,fini}object_sizes.
28465 * tree-object-size.c (addr_object_size): Adjust.
28466 (pass_through_call): Adjust.
28467 (pass_object_sizes::execute): Adjust.
28468 * tree-object-size.h (fini_object_sizes): Declare.
28470 2017-01-10 Martin Sebor <msebor@redhat.com>
28472 PR tree-optimization/78775
28473 * builtins.c (get_size_range): Move...
28474 * calls.c: ...to here.
28475 (alloc_max_size): Accept zero argument.
28476 (operand_signed_p): Remove.
28477 (maybe_warn_alloc_args_overflow): Call get_size_range.
28478 * calls.h (get_size_range): Declare.
28480 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
28482 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
28483 from TI's devices.csv file as of September 2016.
28484 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
28486 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
28488 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
28489 * doc/invoke.texi: Likewise.
28490 * doc/md.texi: Likewise.
28491 * doc/objc.texi: Likewise.
28493 2017-01-10 Joshua Conner <joshconner@google.com>
28495 * config/arm/fuchsia-elf.h: New file.
28496 * config/fuchsia.h: New file.
28497 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
28498 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
28500 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
28502 2016-01-10 Richard Biener <rguenther@suse.de>
28504 PR tree-optimization/79034
28505 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28506 Propagate out degenerate PHIs in the joiner.
28508 2017-01-10 Martin Liska <mliska@suse.cz>
28510 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
28511 (sort_congruence_classes_by_decl_uid): Likewise.
28512 (sort_congruence_class_groups_by_decl_uid): Likewise.
28513 (sem_item_optimizer::merge_classes): Sort class, groups in these
28514 classes and members in the groups by DECL_UID of declarations.
28515 This would make merge operations stable.
28517 2017-01-10 Martin Liska <mliska@suse.cz>
28519 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
28520 usage of m_classes_vec.
28521 (sem_item_optimizer::~sem_item_optimizer): Likewise.
28522 (sem_item_optimizer::get_group_by_hash): Likewise.
28523 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
28524 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
28525 (sem_item_optimizer::verify_classes): Likewise.
28526 (sem_item_optimizer::process_cong_reduction): Likewise.
28527 (sem_item_optimizer::dump_cong_classes): Likewise.
28528 (sem_item_optimizer::merge_classes): Likewise.
28529 * ipa-icf.h (congruence_class_hash): Rename from
28530 congruence_class_group_hash. Remove declaration of m_classes_vec.
28532 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
28534 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
28535 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
28536 * config.gcc: Add avx512vpopcntdqintrin.h.
28537 * config/i386/avx512vpopcntdqintrin.h: New.
28538 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
28539 * config/i386/i386-builtin-types.def: Add new types.
28540 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
28541 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
28542 __builtin_ia32_vpopcountq_v8di_mask): New.
28543 * config/i386/i386-c.c (ix86_target_macros_internal): Define
28544 __AVX512VPOPCNTDQ__.
28545 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
28546 (PTA_AVX512VPOPCNTDQ): Define.
28547 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
28548 TARGET_AVX512VPOPCNTDQ_P): Define.
28549 * config/i386/i386.opt: Add mavx512vpopcntdq.
28550 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
28551 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
28553 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28555 PR middle-end/77484
28556 * predict.def (PRED_CALL): Set to 67.
28558 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
28560 * expr.c (store_field): In the bitfield case, if the value comes from
28561 a function call and is of an aggregate type returned in registers, do
28562 not modify the field mode; extract the value in all cases if the mode
28563 is BLKmode and the size is not larger than a word.
28565 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
28568 * config/i386/cpuid.h: Fix undefined behavior.
28570 2017-01-04 Jeff Law <law@redhat.com>
28572 PR tree-optimization/79007
28573 PR tree-optimization/67955
28574 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
28575 conservative for pt.null when flag_non_call_exceptions is on.
28577 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28579 PR translation/79019
28580 PR translation/79020
28581 * params.def (PARAM_INLINE_MIN_SPEEDUP,
28582 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
28583 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
28585 * config/avr/avr.opt (maccumulate-args): Likewise.
28586 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
28587 * common.opt (freport-bug): Likewise.
28588 * cif-code.def (CIF_FINAL_ERROR): Likewise.
28589 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
28590 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
28591 translatable string.
28592 * config/i386/i386.c (function_value_32): Likewise.
28593 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
28594 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
28596 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
28597 * common/config/msp430/msp430-common.c (msp430_handle_option):
28599 * symtab.c (symtab_node::verify_base): Likewise.
28600 * opts.c (set_debug_level): Likewise.
28601 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
28602 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
28603 missing whitespace to translatable strings.
28604 * config/avr/avr.md (bswapsi2): Fix typo in comment.
28605 * config/sh/superh.h: Likewise.
28606 * config/i386/xopintrin.h: Likewise.
28607 * config/i386/znver1.md: Likewise.
28608 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
28609 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
28610 * double-int.h (struct double_int): Likewise.
28611 * double-int.c (div_and_round_double): Likewise.
28612 * wide-int.cc: Likewise.
28613 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
28614 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
28615 * cfgcleanup.c (crossjumps_occured): Renamed to ...
28616 (crossjumps_occurred): ... this.
28617 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
28620 PR tree-optimization/78899
28621 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
28622 returning bool return struct loop *, NULL for failure and the new
28624 (versionable_outer_loop_p): Don't version outer loop if it has
28625 dont_vectorized bit set.
28626 (tree_if_conversion): When versioning outer loop, ensure
28627 tree_if_conversion is performed also on the inner loop of the
28628 non-vectorizable outer loop copy.
28629 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
28630 LOOP_VECTORIZED in inner loop of the scalar outer loop and
28631 prevent vectorization of it.
28632 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
28633 the outer loop vectorization of the non-scalar version is attempted
28634 before vectorization of the inner loop in scalar version. If
28635 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
28636 vectorization of its inner loop.
28637 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
28638 has 2 inner loops, rename also on edges from bb whose single pred
28639 is outer_loop->header. Fix typo in function comment.
28641 2017-01-09 Martin Sebor <msebor@redhat.com>
28644 * asan.c (asan_emit_stack_protection): Increase local buffer size
28645 to avoid snprintf truncation warning.
28647 2017-01-09 Andrew Pinski <apinski@cavium.com>
28649 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
28650 to reference thunderx2t99 for the tuning structure
28651 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
28653 (thunderx2t99_extra_costs): This.
28654 * config/aarch64/aarch64-tune.md: Regenerate.
28655 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
28656 (vulcan_addrcost_table): This.
28657 (vulcan_regmove_cost): Rename to ...
28658 (thunderx2t99_regmove_cost): This.
28659 (vulcan_vector_cost): Rename to ...
28660 (thunderx2t99_vector_cost): this.
28661 (vulcan_branch_cost): Rename to ...
28662 (thunderx2t99_branch_cost): This.
28663 (vulcan_tunings): Rename to ...
28664 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
28665 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
28667 2017-01-09 Martin Jambor <mjambor@suse.cz>
28671 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
28672 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
28673 (propagate_vr_accross_jump_function): Use the above function for all
28674 value range computations for pass-through jump functions and type
28675 converasion from explicit value range values.
28676 (ipcp_propagate_stage): Do not attempt to deduce types of formal
28677 parameters from TYPE_ARG_TYPES.
28678 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
28679 (ipa_write_node_info): Stream type of the actual argument.
28680 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
28682 2017-01-09 Martin Liska <mliska@suse.cz>
28685 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
28686 (lookup_compiler): Do not show error message with have_E.
28688 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28690 PR tree-optimization/78938
28691 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
28692 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
28693 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
28694 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
28697 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28699 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
28702 2017-01-09 Richard Biener <rguenther@suse.de>
28704 PR tree-optimization/78997
28705 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
28706 name condition properly.
28708 2017-01-09 Richard Biener <rguenther@suse.de>
28711 * dwarf2out.c (is_cxx): New overload with context.
28712 (is_naming_typedef_decl): Use it.
28714 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28716 * invoke.texi (Option Summary): Correct spacing in option lists
28717 and add line breaks to fix over-long lines.
28719 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28721 PR middle-end/17660
28723 * extend.texi (Common Variable Attributes): Add xref to GCC
28724 Internals manual to explain mode attribute keywords.
28726 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
28729 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
28730 and Preprocessor Options.
28731 (Options for Linking): Document -pthread here....
28732 (RS/6000 and PowerPC Options): ...not here.
28733 (Solaris 2 Options): ...or here.
28734 * doc/cppopts.texi: Document -pthread.
28736 2017-01-08 Martin Sebor <msebor@redhat.com>
28738 PR middle-end/77708
28739 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
28740 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
28741 New member functions.
28742 (format_directive): Used them.
28744 (pass_sprintf_length::handle_gimple_call): Same.
28745 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
28746 to avoid truncation for any argument.
28747 (extract_affine_mul): Same.
28748 * tree.c (get_file_function_name): Same.
28750 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28752 PR middle-end/77484
28753 * predict.def (PRED_INDIR_CALL): Set to 86.
28755 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28757 PR preprocessor/54124
28758 * doc/cppopts.texi: Reformat -d subtable to list the full name
28759 of the options. Add cross-reference to the docs for the general
28760 compiler -d options.
28761 * doc/invoke.texi (Developer Options): Add cross-reference to the
28762 preprocessor-specific -d option documentation.
28764 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28766 PR preprocessor/13498
28767 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
28768 redudant material, and reflect new command-line options.
28769 (System Headers): Likewise.
28771 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
28773 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
28774 -isystem, and -idirafter. Copy-edit.
28775 * doc/cppopts.texi: Copy-edit. Remove contradiction about
28776 default for -ftrack-macro-expansion. Delete obsolete and
28777 badly-formatted implementation details about -fdebug-cpp output.
28778 * doc/cppwarnopts.texi: Copy-edit.
28780 2017-01-07 David Malcolm <dmalcolm@redhat.com>
28783 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28784 that the transition from a max line width >= 1<<10 to narrower
28785 lines works correctly.
28787 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
28789 * doc/options.texi (PerFunction): New.
28790 * opt-functions.awk (switch_flags): Map both Optimization and
28791 PerFunction to CL_OPTIMIZATION.
28792 * opth-gen.awk: Test for PerFunction flag along with
28794 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
28795 it only when the latter is present. Skip those that don't in
28796 the hash function generator.
28797 * common.opt (fvar-tracking): Mark as PerFunction instead of
28799 (fvar-tracking-assignments): Likewise.
28800 (fvar-tracking-assignments-toggle): Likewise.
28801 (fvar-tracking-uninit): Likewise.
28803 2017-01-07 Jakub Jelinek <jakub@redhat.com>
28805 PR translation/79018
28806 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
28809 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
28812 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
28813 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
28814 TARGET_LONG_JUMP_TABLE_OFFSETS.
28815 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
28816 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
28817 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
28818 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
28819 * config/m68k/m68k.md (tablejump expander): Likewise.
28820 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
28821 TARGET_LONG_JUMP_TABLE_OFFSETS.
28822 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
28823 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
28825 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
28826 David Holsgrove <david.holsgrove@xilinx.com>
28828 * common/config/microblaze/microblaze-common.c
28829 (TARGET_EXCEPT_UNWIND_INFO): Remove.
28830 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
28832 * config/microblaze/microblaze.c (microblaze_must_save_register)
28833 (microblaze_expand_epilogue, microblaze_return_addr): Handle
28835 (microblaze_eh_return): New function.
28836 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
28837 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
28838 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
28839 * config/microblaze/microblaze.md (eh_return): New pattern.
28841 2017-01-06 Jakub Jelinek <jakub@redhat.com>
28843 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
28844 GCC_DIAGNOSTIC_STRINGIFY): Define.
28846 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
28848 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28850 * config/arm/arm.md (<mcrr>): New.
28852 * config/arm/arm.c (arm_arch5te): New.
28853 (arm_option_override): Set arm_arch5te.
28854 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
28856 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
28857 (arm_mcrr_qualifiers): ... this. New.
28858 (MRRC_QUALIFIERS): Define to...
28859 (arm_mrrc_qualifiers): ... this. New.
28860 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
28862 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
28863 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
28864 (MRRCI, mrrc, MRRC): New.
28865 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
28866 VUNSPEC_MRRC2): New.
28868 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28870 * config/arm/arm.md (<mcr>): New.
28872 * config/arm/arm.c (arm_coproc_builtin_available): Add
28873 support for mcr, mrc, mcr2 and mrc2.
28874 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
28875 (arm_mcr_qualifiers): ... this. New.
28876 (MRC_QUALIFIERS): Define to ...
28877 (arm_mrc_qualifiers): ... this. New.
28878 (MCR_QUALIFIERS): Define to ...
28879 (arm_mcr_qualifiers): ... this. New.
28880 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
28882 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
28883 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
28884 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
28885 VUNSPEC_MRC2): New.
28887 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28889 * config/arm/arm.md (*ldc): New.
28893 * config/arm/arm.c (arm_coproc_builtin_available): Add
28894 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
28895 (arm_coproc_ldc_stc_legitimate_address): New.
28896 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
28897 'qualifier_const_pointer'.
28898 (LDC_QUALIFIERS): Define to...
28899 (arm_ldc_qualifiers): ... this. New.
28900 (STC_QUALIFIERS): Define to...
28901 (arm_stc_qualifiers): ... this. New.
28902 * config/arm/arm-protos.h
28903 (arm_coproc_ldc_stc_legitimate_address): New.
28904 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
28905 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
28906 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
28907 stc2, stcl, stc2l): New.
28908 * config/arm/constraints.md (Uz): New.
28909 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
28910 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
28911 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
28912 VUNSPEC_STC2L): New.
28914 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28916 * config/arm/arm.md (<cdp>): New.
28917 * config/arm/arm.c (neon_const_bounds): Rename this ...
28918 (arm_const_bounds): ... this.
28919 (arm_coproc_builtin_available): New.
28920 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
28921 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
28922 (CDP_QUALIFIERS): Define to...
28923 (arm_cdp_qualifiers): ... this. New.
28925 (arm_expand_builtin_args): Add case for 6 arguments.
28926 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
28927 (arm_const_bounds): ... this.
28928 (arm_coproc_builtin_available): New.
28929 * config/arm/arm_acle.h (__arm_cdp): New.
28931 * config/arm/arm_acle_builtins.def (cdp): New.
28933 * config/arm/iterators.md (CDPI,CDP,cdp): New.
28934 * config/arm/neon.md: Rename all 'neon_const_bounds' to
28935 'arm_const_bounds'.
28936 * config/arm/types.md (coproc): New.
28937 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
28938 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
28939 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
28940 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
28942 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28944 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
28945 (UBINOP_QUALIFIERS): New.
28947 (acle_builtin_data): New. Change comment.
28948 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
28949 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
28950 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
28951 arm_acle_builtins.def.
28952 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
28953 (arm_init_acle_builtins): New.
28954 (CRC32_BUILTIN): Remove.
28955 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
28956 crc32cb, crc32ch and crc32cw.
28957 (arm_init_crc32_builtins): Remove.
28958 (arm_init_builtins): Use arm_init_acle_builtins rather
28959 than arm_init_crc32_builtins.
28960 (arm_expand_acle_builtin): New.
28961 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
28962 * config/arm/arm_acle_builtins.def: New.
28964 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
28966 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
28967 (arm_builtin_datum): ... this.
28968 (arm_init_neon_builtin): Rename to ...
28969 (arm_init_builtin): ... this. Add a new parameters PREFIX
28970 and USE_SIG_IN_NAME.
28971 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
28972 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
28973 'arm_builtin_datum'.
28974 (arm_init_vfp_builtins): Likewise.
28975 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
28976 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
28977 (arm_expand_neon_args): Rename to ...
28978 (arm_expand_builtin_args): ... this. Rename builtin_arg
28979 enum values and differentiate between ARG_BUILTIN_MEMORY
28980 and ARG_BUILTIN_NEON_MEMORY.
28981 (arm_expand_neon_builtin_1): Rename to ...
28982 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
28983 values, arm_expand_builtin_args and add bool parameter NEON.
28984 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
28985 (arm_expand_vfp_builtin): Likewise.
28986 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
28988 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28990 PR middle-end/77484
28991 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
28992 * predict.c (tree_estimate_probability_bb): Reverse direction of
28993 polymorphic call predictor.
28995 2017-01-06 David Malcolm <dmalcolm@redhat.com>
28997 * passes.c (execute_one_pass): Split out pass-skipping logic into...
28998 (determine_pass_name_match): ...this new function and...
28999 (should_skip_pass_p): ...this new function.
29001 2017-01-06 Nathan Sidwell <nathan@acm.org>
29003 * ipa-visibility.c (function_and_variable_visibility): Reformat
29004 comments and long lines. Remove extrneous if.
29005 * symtab.c (symtab_node::make_decl_local): Fix code format.
29006 (symtab_node::set_section_for_node): Fix comment typo.
29008 2017-01-06 Martin Liska <mliska@suse.cz>
29011 * lra-constraints.c: Rename invariant to lra_invariant.
29012 * predict.c (set_even_probabilities): Initialize e to NULL.
29014 2017-01-05 Martin Sebor <msebor@redhat.com>
29016 PR tree-optimization/78910
29017 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
29018 (format_integer): Correct off-by-one error in the handling
29019 of precision with negative numbers in signed conversions..
29021 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
29023 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
29025 2017-01-05 Jakub Jelinek <jakub@redhat.com>
29027 PR tree-optimization/71016
29028 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
29029 factor_out_conditional_conversion. Formatting fix.
29030 (factor_out_conditional_conversion): Add cond_stmt argument.
29031 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
29032 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
29035 2017-01-05 David Malcolm <dmalcolm@redhat.com>
29037 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
29038 read-rtl-function.o, and selftest-rtl.o.
29039 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
29040 (selftest::aarch64_test_loading_full_dump): New function.
29041 (selftest::aarch64_run_selftests): New function.
29042 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
29043 selftest::aarch64_run_selftests.
29044 * config/i386/i386.c
29045 (selftest::ix86_test_loading_dump_fragment_1): New function.
29046 (selftest::ix86_test_loading_call_insn): New function.
29047 (selftest::ix86_test_loading_full_dump): New function.
29048 (selftest::ix86_test_loading_unspec): New function.
29049 (selftest::ix86_run_selftests): Call the new functions.
29050 * emit-rtl.c (maybe_set_max_label_num): New function.
29051 * emit-rtl.h (maybe_set_max_label_num): New decl.
29052 * function.c (instantiate_decls): Guard call to
29053 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
29054 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
29056 * gensupport.c (gen_reader::gen_reader): Pass "false"
29057 for new "compact" param of rtx_reader.
29058 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
29059 rather than an empty string for NULL strings.
29060 * read-md.c: Potentially include config.h rather than bconfig.h.
29061 Wrap include of errors.h with #ifdef GENERATOR_FILE.
29062 (have_error): New global, copied from errors.c.
29063 (md_reader::read_name): Rename to...
29064 (md_reader::read_name_1): ...this, adding "out_loc" param,
29065 and converting "missing name or number" to returning false, rather
29067 (md_reader::read_name): Reimplement in terms of read_name_1.
29068 (md_reader::read_name_or_nil): New function.
29069 (md_reader::read_string): Handle "(nil)" by returning NULL.
29070 (md_reader::md_reader): Add new param "compact".
29071 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
29072 (md_reader::read_file): New method.
29073 * read-md.h (md_reader::md_reader): Add new param "compact".
29074 (md_reader::read_file): New method.
29075 (md_reader::is_compact): New accessor.
29076 (md_reader::read_name): Convert return type from void to file_location.
29077 (md_reader::read_name_or_nil): New decl.
29078 (md_reader::read_name_1): New decl.
29079 (md_reader::m_compact): New field.
29080 (noop_reader::noop_reader): Pass "false" for new "compact" param
29082 (rtx_reader::rtx_reader): Add new "compact" param.
29083 (rtx_reader::read_rtx_operand): Make virtual and convert return
29084 type from void to rtx.
29085 (rtx_reader::read_until): New decl.
29086 (rtx_reader::handle_any_trailing_information): New virtual function.
29087 (rtx_reader::postprocess): New virtual function.
29088 (rtx_reader::finalize_string): New virtual function.
29089 (rtx_reader::m_in_call_function_usage): New field.
29090 (rtx_reader::m_reuse_rtx_by_id): New field.
29091 * read-rtl-function.c: New file.
29092 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
29093 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
29094 (selftest::verify_three_block_rtl_cfg): New decl.
29095 * read-rtl-function.h: New file.
29096 * read-rtl.c: Potentially include config.h rather than bconfig.h.
29097 For host, include function.h, memmodel.h, and emit-rtl.h.
29098 (one_time_initialization): New function.
29099 (struct compact_insn_name): New struct.
29100 (compact_insn_names): New array.
29101 (find_code): Handle insn codes in compact dumps.
29102 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
29103 (bind_subst_iter_and_attr): Likewise.
29104 (add_condition_to_string): Likewise.
29105 (add_condition_to_rtx): Likewise.
29106 (apply_attribute_uses): Likewise.
29107 (add_current_iterators): Likewise.
29108 (apply_iterators): Likewise.
29109 (initialize_iterators): Guard usage of apply_subst_iterator with
29110 #ifdef GENERATOR_FILE.
29111 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
29112 (md_reader::read_mapping): Likewise.
29113 (add_define_attr_for_define_subst): Likewise.
29114 (add_define_subst_attr): Likewise.
29115 (read_subst_mapping): Likewise.
29116 (check_code_iterator): Likewise.
29117 (rtx_reader::read_rtx): Likewise. Move one-time initialization
29119 (one_time_initialization): New function.
29120 (rtx_reader::read_until): New method.
29121 (read_flags): New function.
29122 (parse_reg_note_name): New function.
29123 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
29124 Handle reuse_rtx ids.
29125 Wrap iterator lookup within #ifdef GENERATOR_FILE.
29126 Add parsing support for RTL dumps, mirroring the special-cases in
29127 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
29128 values, and calling handle_any_trailing_information.
29129 (rtx_reader::read_rtx_operand): Convert return type from void
29130 to rtx, returning return_rtx. Handle case 'e'. Call
29131 finalize_string on XSTR and XTMPL fields.
29132 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
29133 "(nil)" values were omitted. Call the postprocess vfunc on the
29135 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
29136 class ctor. Initialize m_in_call_function_usage. Call
29137 one_time_initialization.
29138 * rtl-tests.c (selftest::test_uncond_jump): Call
29139 set_new_first_and_last_insn.
29140 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
29141 * selftest-rtl.c: New file.
29142 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
29143 (selftest::get_insn_by_uid): New decl.
29144 * selftest-run-tests.c (selftest::run_tests): Call
29145 read_rtl_function_c_tests.
29146 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
29147 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
29150 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
29152 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
29153 operands in a special way. Assert that pos+len <= mode precision.
29155 2017-01-05 Jakub Jelinek <jakub@redhat.com>
29157 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
29158 3 argument Alias with unlimited for the negative form.
29159 (fno-vect-cost-model): Removed.
29161 2017-01-05 Martin Liska <mliska@suse.cz>
29163 * hsa-gen.c (gen_hsa_divmod): New function.
29164 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
29166 2017-01-05 Martin Liska <mliska@suse.cz>
29169 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
29172 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29174 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
29175 small constant length operands.
29177 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29179 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
29180 between loop iterations.
29182 2017-01-05 Martin Liska <mliska@suse.cz>
29185 * gimplify.c (gimplify_decl_expr): Compare to
29186 asan_poisoned_variables instread of checking flags.
29187 (gimplify_target_expr): Likewise.
29188 (gimplify_expr): Likewise.
29189 (gimplify_function_tree): Conditionally initialize
29190 asan_poisoned_variables.
29192 2017-01-04 Jeff Law <law@redhat.com>
29194 PR tree-optimizatin/78812
29195 * rtl.h (contains_mem_rtx_p): Prototype.
29196 * ifcvt.c (containts_mem_rtx_p): Move from here to...
29197 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
29198 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
29199 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
29200 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
29202 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29204 * input.c (assert_char_at_range): Default-initialize actual_range.
29206 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29208 * df-scan.c (df_ref_create_structure): Make regno unsigned,
29209 to match the caller.
29211 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29213 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
29214 insns after final jump in test to emit dummy move.
29216 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29218 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
29219 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
29221 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29223 * multiple_target.c (create_dispatcher_calls): Init e_next.
29224 * tree-ssa-loop-split.c (split_loop): Init border.
29225 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
29228 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
29233 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
29234 (altivec_register_operand): Do not return true if the operand
29235 contains a SUBREG mixing SImode and SFmode.
29236 (vsx_register_operand): Likewise.
29237 (vsx_reg_sfsubreg_ok): New predicate.
29238 (vfloat_operand): Do not return true if the operand contains a
29239 SUBREG mixing SImode and SFmode.
29240 (vint_operand): Likewise.
29241 (vlogical_operand): Likewise.
29242 (gpc_reg_operand): Likewise.
29243 (int_reg_operand): Likewise.
29244 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
29245 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
29246 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
29248 (rs6000_emit_move_si_sf_subreg): New helper function.
29249 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
29250 fixup SUBREGs involving SImode and SFmode.
29251 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
29252 numbers for the new peephole2 optimization.
29253 (peephole2 for SFmode unions): New peephole2 to optimize cases in
29254 the GLIBC math library that do AND/IOR/XOR operations on single
29255 precision floating point.
29256 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
29257 target macros to say whether we need to avoid SUBREGs mixing
29259 (TARGET_ALLOW_SF_SUBREG): Likewise.
29260 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
29261 (UNSPEC_SI_FROM_SF): Likewise.
29262 (iorxor): Change spacing.
29263 (and_ior_xor): New iterator for AND, IOR, and XOR.
29264 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
29265 (movdi_from_sf_zero_ext): Likewise.
29266 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
29267 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
29268 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
29269 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
29270 (fms<mode>4): Likewise.
29271 (fnma<mode>4): Likewise.
29272 (fnms<mode>4): Likewise.
29273 (nfma<mode>4): Likewise.
29274 (nfms<mode>4): Likewise.
29276 2017-01-04 Marek Polacek <polacek@redhat.com>
29279 * doc/invoke.texi: Document -Wpointer-compare.
29281 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29283 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
29286 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
29287 descriptions for -gdwarf-5 and emit them as uleb128 instead of
29290 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
29293 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
29294 documentation of the powerpc_popcntb_ok attribute.
29295 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29296 code to issue warning messages if a requested CPU configuration is
29297 not supported by the binary (assembler and loader) toolchain.
29298 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
29299 made to define a built-in function that has been disabled.
29300 (paired_init_builtins): Add assertion to prevent ICE if attempt is
29301 made to define a built-in function that has been disabled.
29302 (altivec_init_builtins): Add comment explaining why definition
29303 of the DST built-in functions is not preceded by an assertion
29304 check. Add assertions to prevent ICE if attempts are made to
29305 define an altivec predicate or an abs* built-in function that has
29307 (htm_init_builtins): Add comment explaining why definition of the
29308 htm built-in functions is not preceded by an assertion check.
29310 2017-01-04 Jeff Law <law@redhat.com>
29312 PR tree-optimizatin/67955
29313 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
29314 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
29315 the points-to solution does not include pt_null. Use DECL_PT_UID
29318 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
29320 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
29321 Use gen_int_mode instead of gen_lopwart for const_int operands.
29323 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29325 PR tree-optimization/71563
29326 * match.pd: Simplify X << Y into X if Y is known to be 0 or
29327 out of range value - has low bits known to be zero.
29329 2017-01-04 Alan Modra <amodra@gmail.com>
29331 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
29332 * configure: Regenerate.
29333 * config.in: Regenerate.
29335 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29338 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
29339 a substring of the message, but strcmp with the whole message. Ifdef
29340 ENABLE_NLS, translate the message first using dgettext.
29342 2017-01-03 Jeff Law <law@redhat.com>
29344 PR tree-optimizatin/78856
29345 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
29346 (mark_threaded_blocks): Remove code to truncate thread paths that
29347 cross multiple loop headers. Instead invalidate the cached loop
29348 iteration information and handle case of a thread path walking
29349 into an irreducible region.
29351 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
29354 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
29355 assertions. Add support for doing the signbit if the IEEE 128-bit
29356 floating point value is in a GPR.
29357 * config/rs6000/rs6000.md (Fsignbit): Delete.
29358 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
29359 Update the length attribute if the value is in a GPR.
29360 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
29361 the sign or zero extension instruction, since the value is always 0/1.
29362 (signbit<mode>2_dm2): Delete using <Fsignbit>.
29365 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
29366 extracting SImode to a GPR register so that we can generate a
29367 store, limit the vector to be in a traditional Altivec register
29368 for the vextuwrx instruction.
29370 2017-01-03 Ian Lance Taylor <iant@google.com>
29372 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
29374 2017-01-03 Martin Sebor <msebor@redhat.com>
29376 PR tree-optimization/78696
29377 * gimple-ssa-sprintf.c (format_floating): Correct handling of
29378 precision. Use MPFR for %f for greater fidelity. Correct handling
29380 (pass_sprintf_length::compute_format_length): Set width and precision
29381 specified by asrerisk to void_node for vararg functions.
29382 (try_substitute_return_value): Adjust dump output.
29384 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
29386 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
29388 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
29390 * doc/invoke.texi (SPARC options): Document -mlra as the default.
29391 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
29392 -mlra/-mno-lra was passed to the compiler.
29394 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
29396 PR rtl-optimization/65618
29397 * emit-rtl.c (try_split): Move initialization of "before" and
29398 "after" to just before the call to emit_insn_after_setloc.
29400 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
29402 * doc/md.texi (Standard Names): Remove reference to Java frontend.
29404 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
29406 * dwarf2out.c (gen_enumeration_type_die): When
29407 -gno-strict-dwarf, add a DW_AT_encoding attribute.
29409 2017-01-03 Jakub Jelinek <jakub@redhat.com>
29411 PR tree-optimization/78965
29412 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
29413 Change first argument from const call_info & to call_info &. For %n
29414 set info.nowrite to false.
29416 PR middle-end/78901
29417 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
29418 possibly throwing calls.
29420 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
29421 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
29422 and fns handling, rather than in a separate case SSA_NAME.
29424 2017-01-02 Jeff Law <law@redhat.com>
29426 * config/darwin-driver.c (darwin_driver_init): Const-correctness
29427 fixes for first_period and second_period variables.
29429 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
29432 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
29433 (*insvqi_1): New insn pattern.
29434 (*insvqi_1_mem_rex64): Ditto.
29435 (*insvqi_2): Ditto.
29436 (*insvqi_3): Rename from *insvqi.
29438 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
29440 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
29442 * doc/cfg.texi (Edges): Remove reference to Java.
29443 (Maintaining the CFG): Ditto.
29445 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29447 PR middle-end/77674
29448 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
29449 transparent aliases.
29451 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29453 PR middle-end/77484
29454 * predict.def (PRED_CALL): Update hitrate.
29455 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
29456 * predict.c (tree_estimate_probability_bb): Split CALL predictor
29457 into direct/indirect/polymorphic variants.
29459 2017-01-01 Jakub Jelinek <jakub@redhat.com>
29461 Update copyright years.
29463 * gcc.c (process_command): Update copyright notice dates.
29464 * gcov-dump.c (print_version): Ditto.
29465 * gcov.c (print_version): Ditto.
29466 * gcov-tool.c (print_version): Ditto.
29467 * gengtype.c (create_file): Ditto.
29468 * doc/cpp.texi: Bump @copying's copyright year.
29469 * doc/cppinternals.texi: Ditto.
29470 * doc/gcc.texi: Ditto.
29471 * doc/gccint.texi: Ditto.
29472 * doc/gcov.texi: Ditto.
29473 * doc/install.texi: Ditto.
29474 * doc/invoke.texi: Ditto.
29476 Copyright (C) 2017 Free Software Foundation, Inc.
29478 Copying and distribution of this file, with or without modification,
29479 are permitted in any medium without royalty provided the copyright
29480 notice and this notice are preserved.