1 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4 * expr.c (store_field): Don't optimize away bitsize == 0 store
5 from CALL_EXPR with addressable return type.
7 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
10 * config/rs6000/rs6000.md (sel): Delete mode attribute.
11 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
13 TARGET_ISEL instead of TARGET_ISEL<sel>.
15 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
17 * config/rs6000/rs6000.c
18 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
20 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
22 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
24 (reg_or_cint_operand): Fix comment.
25 (reg_or_zero_operand): New predicate.
26 * config/rs6000/rs6000-protos.h (output_isel): Delete.
27 * config/rs6000/rs6000.c (output_isel): Delete.
28 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
29 instead of reg_or_cint_operand. Output instruction directly (not via
31 (isel_unsigned_<mode>): Ditto.
32 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
33 gpc_reg_operand. Add an instruction alternative for this. Output
35 (*isel_reversed_unsigned_<mode>): Ditto.
37 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
39 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
40 (TARGET_CANONICALIZE_COMPARISON): Define.
42 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
45 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
46 Check whether the dest is REG before adding REG_EQUIV note.
48 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
51 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
53 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
54 (make_hard_regno_born, make_hard_regno_dead): Update
55 bb_killed_pseudos and bb_gen_pseudos.
57 2017-10-11 Nathan Sidwell <nathan@acm.org>
59 * incpath.h (enum incpath_kind): Name enum, prefix values.
60 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
61 * incpath.c (heads, tails): Use INC_MAX.
62 (add_env_var_paths, add_standard_paths): Use incpath_kind.
63 (merge_include_chains, split_quote_chain,
64 register_include_chains): Update incpath_kind names.
65 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
66 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
68 (add_framework_path, darwin_register_objc_includes): Likewise.
69 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
71 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
73 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
74 Do not use float_operator operator predicate.
75 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
76 * config/i386/predicates.md (float_operator): Remove predicate.
78 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
80 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
82 (*jcc<mode>_i387): Ditto.
83 (*jccu<mode>_i387): Ditto.
84 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
85 (*jcc_*_i387 splitters): Remove.
86 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
87 * config/i386/i386.c (ix86_split_fp_branch): Remove.
88 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
91 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
93 * profile-count.h (slow_safe_scale_64bit): New function.
94 (safe_scale_64bit): New inline.
95 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
96 * profile-count.c: Include wide-int.h
97 (slow_safe_scale_64bit): New.
99 2017-10-11 Nathan Sidwell <nathan@acm.org>
101 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
102 HAS_DECL_ASSEMBLER_NAME_P.
103 * gimple-expr.c (gimple_decl_printable_name: Check
104 HAS_DECL_ASSEMBLER_NAME_P too.
105 * ipa-utils.h (type_in_anonymous_namespace_p): Check
106 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
107 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
108 * passes.c (rest_of_decl_compilation): Check
109 HAS_DECL_ASSEMBLER_NAME_P too.
110 * recog.c (verify_changes): Likewise.
111 * tree-pretty-print.c (dump_decl_name): Likewise.
112 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
114 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
115 (SET_DECL_ASSEMBLER_NAME): Use it.
116 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
117 (COPY_DECL_ASSEMBLER_NAME): Likewise.
118 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
120 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
122 * config.gcc (i386, x86_64): Add extra objects.
123 * i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
124 (ix86_min_insn_size): Declare.
125 (ix86_issue_rate): Declare.
126 (ix86_adjust_cost): Declare.
127 (ia32_multipass_dfa_lookahead): Declare.
128 (ix86_macro_fusion_p): Declare.
129 (ix86_macro_fusion_pair_p): Declare.
130 (ix86_bd_has_dispatch): Declare.
131 (ix86_bd_do_dispatch): Declare.
132 (ix86_core2i7_init_hooks): Declare.
133 (ix86_atom_sched_reorder): Declare.
134 * i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
135 (COSTS_N_BYTES): Move to x86-tune-costs.h.
136 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
137 (rip_relative_addr_p): Rename to ...
138 (ix86_rip_relative_addr_p): ... this one; export.
139 (memory_address_length): Update.
140 (ix86_issue_rate): Move to x86-tune-sched.c.
141 (ix86_flags_dependent): Move to x86-tune-sched.c.
142 (ix86_agi_dependent): Move to x86-tune-sched.c.
143 (exact_dependency_1): Move to x86-tune-sched.c.
144 (exact_store_load_dependency): Move to x86-tune-sched.c.
145 (ix86_adjust_cost): Move to x86-tune-sched.c.
146 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
147 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
148 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
149 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
150 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
151 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
152 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
153 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
154 (min_insn_size): Rename to ...
155 (ix86_min_insn_size): ... this one; export.
156 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
157 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
158 (core2i7_first_cycle_multipass_backtrack): Move to
159 x86-tune-sched-core.c.
160 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
161 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
162 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
163 (ix86_avoid_jump_mispredicts): Update.
164 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
165 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
166 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
167 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
168 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
169 (MAX_INSN): Move to ix86-tune-sched-bd.c.
170 (MAX_IMM): Move to ix86-tune-sched-bd.c.
171 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
172 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
173 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
174 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
175 (MAX_STORE): Move to ix86-tune-sched-bd.c.
176 (BIG): Move to ix86-tune-sched-bd.c.
177 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
178 (enum insn_path): Move to ix86-tune-sched-bd.c.
179 (get_mem_group): Move to ix86-tune-sched-bd.c.
180 (is_cmp): Move to ix86-tune-sched-bd.c.
181 (dispatch_violation): Move to ix86-tune-sched-bd.c.
182 (is_branch): Move to ix86-tune-sched-bd.c.
183 (is_prefetch): Move to ix86-tune-sched-bd.c.
184 (init_window): Move to ix86-tune-sched-bd.c.
185 (allocate_window): Move to ix86-tune-sched-bd.c.
186 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
187 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
188 (process_end_window): Move to ix86-tune-sched-bd.c.
189 (allocate_next_window): Move to ix86-tune-sched-bd.c.
190 (find_constant): Move to ix86-tune-sched-bd.c.
191 (get_num_immediates): Move to ix86-tune-sched-bd.c.
192 (has_immediate): Move to ix86-tune-sched-bd.c.
193 (get_insn_path): Move to ix86-tune-sched-bd.c.
194 (get_insn_group): Move to ix86-tune-sched-bd.c.
195 (count_num_restricted): Move to ix86-tune-sched-bd.c.
196 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
197 (add_insn_window): Move to ix86-tune-sched-bd.c.
198 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
199 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
200 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
201 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
202 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
203 (do_dispatch): Move to ix86-tune-sched-bd.c.
204 (has_dispatch): Move to ix86-tune-sched-bd.c.
205 * i386/t-i386: Add new object files.
206 * i386/x86-tune-costs.h: New file.
207 * i386/x86-tune-sched-atom.c: New file.
208 * i386/x86-tune-sched-bd.c: New file.
209 * i386/x86-tune-sched-core.c: New file.
210 * i386/x86-tune-sched.c: New file.
212 2017-10-11 Liu Hao <lh_mouse@126.com>
214 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
215 the generic version below instead.
216 (should_colorize): Recognize Windows consoles as terminals
218 * pretty-print.c [__MINGW32__] (write_all): New function.
219 [__MINGW32__] (find_esc_head): Likewise.
220 [__MINGW32__] (find_esc_terminator): Likewise.
221 [__MINGW32__] (eat_esc_sequence): Likewise.
222 [__MINGW32__] (mingw_ansi_fputs): New function that handles
224 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
227 2017-10-11 Richard Biener <rguenther@suse.de>
229 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
230 Properly call analyze_scalar_evolution with the loop of the stmt.
232 2017-10-11 Richard Biener <rguenther@suse.de>
234 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
235 * tree-core.h (tree_base): Add chrec_var union member.
236 * tree.h (CHREC_VAR): Remove.
237 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
238 * tree-chrec.h (build_polynomial_chrec): Adjust.
239 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
240 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
242 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
244 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
245 * match.pd: ... here.
246 ((T) X == (T) Y): Relax condition.
248 2017-10-11 Bin Cheng <bin.cheng@arm.com>
250 PR tree-optimization/82472
251 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
253 (break_alias_scc_partitions): Update postorder number.
255 2017-10-11 Martin Liska <mliska@suse.cz>
258 * opts.c (parse_no_sanitize_attribute): Do not use error_value
260 * opts.h (parse_no_sanitize_attribute): Remove last argument.
262 2017-10-11 Martin Liska <mliska@suse.cz>
264 * print-rtl.c (print_insn): Move declaration of idbuf
265 to same scope as name.
267 2017-10-11 Martin Liska <mliska@suse.cz>
272 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
275 2017-10-11 Martin Liska <mliska@suse.cz>
278 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
281 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
283 * config.gcc (default_gnu_indirect_function): Default to yes for
284 arm*-*-linux* with glibc.
286 2017-10-11 Richard Biener <rguenther@suse.de>
288 * tree-scalar-evolution.c (get_scalar_evolution): Handle
289 default-defs and types we do not want to analyze.
290 (interpret_loop_phi): Replace unreachable code with an assert.
291 (compute_scalar_evolution_in_loop): Remove and inline ...
292 (analyze_scalar_evolution_1): ... here, replacing condition with
293 what makes the intent clearer. Remove handling of cases
294 get_scalar_evolution now handles.
296 2017-10-10 Jim Wilson <wilson@tuliptree.org>
298 PR rtl-optimization/81434
299 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
300 comment for main loop. In sched_group_found if, also add checks for
301 pass and min_cost_group.
303 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
305 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
306 (rs6000_insn_cost): New function.
307 * config/rs6000/rs6000.md (cost): New attribute.
309 2017-10-10 Jakub Jelinek <jakub@redhat.com>
310 H.J. Lu <hongjiu.lu@intel.com>
314 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
315 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
316 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
317 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
318 to OPTION_MASK_ISA_AVX512VL - builtins that have both
319 OPTION_MASK_ISA_MMX and some other bit set require both
320 mmx and the ISAs without the mmx bit.
321 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
322 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
323 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
324 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
325 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
326 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
327 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
328 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
329 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
330 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
331 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
332 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
333 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
334 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
335 Add OPTION_MASK_ISA_MMX.
337 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
339 * config.gcc (armv7*-*-freebsd*): New target.
340 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
342 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
344 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
345 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
348 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
350 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
351 a template parameter.
352 (WIDE_INT_REF_FOR): Update accordingly.
353 * tree.h (wi::int_traits <const_tree>): Delete.
354 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
355 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
356 (wi::tree_to_wide_ref): New typedef.
357 (wi::to_wide): New function.
358 * calls.c (get_size_range): Use wi::to_wide when operating on
360 * cgraph.c (cgraph_node::create_thunk): Likewise.
361 * config/i386/i386.c (ix86_data_alignment): Likewise.
362 (ix86_local_alignment): Likewise.
363 * dbxout.c (stabstr_O): Likewise.
364 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
365 * expr.c (const_vector_from_tree): Likewise.
366 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
367 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
368 (fold_negate_expr_1, int_const_binop_1, const_binop)
369 (fold_convert_const_int_from_real, optimize_bit_field_compare)
370 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
371 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
372 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
373 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
374 (fold_not_const, round_up_loc): Likewise.
375 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
376 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
377 (alloca_call_type): Likewise.
378 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
379 * godump.c (go_output_typedef): Likewise.
380 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
381 * internal-fn.c (get_min_precision): Likewise.
382 * ipa-cp.c (ipcp_store_vr_results): Likewise.
383 * ipa-polymorphic-call.c
384 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
385 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
386 (ipa_modify_call_arguments): Likewise.
387 * match.pd: Likewise.
388 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
389 * print-tree.c (print_node_brief, print_node): Likewise.
390 * stmt.c (expand_case): Likewise.
391 * stor-layout.c (layout_type): Likewise.
392 * tree-affine.c (tree_to_aff_combination): Likewise.
393 * tree-cfg.c (group_case_labels_stmt): Likewise.
394 * tree-data-ref.c (dr_analyze_indices): Likewise.
395 (prune_runtime_alias_test_list): Likewise.
396 * tree-dump.c (dequeue_and_dump): Likewise.
397 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
398 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
399 * tree-pretty-print.c (dump_generic_node): Likewise.
400 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
401 (simple_iv_with_niters): Likewise.
402 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
403 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
404 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
405 * tree-ssa-loop-niter.c (split_to_var_and_offset)
406 (refine_value_range_using_guard, number_of_iterations_ne_max)
407 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
408 (get_cst_init_from_scev, record_nonwrapping_iv)
409 (scev_var_range_cant_overflow): Likewise.
410 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
411 * tree-ssa-pre.c (compute_avail): Likewise.
412 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
413 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
414 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
415 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
416 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
417 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
418 (dump_case_nodes, try_switch_expansion): Likewise.
419 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
420 (vect_do_peeling): Likewise.
421 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
422 * tree-vect-stmts.c (vectorizable_load): Likewise.
423 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
424 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
425 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
426 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
427 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
428 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
429 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
430 (evrp_dom_walker::before_dom_children): Likewise.
431 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
432 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
433 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
434 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
435 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
436 (get_range_pos_neg): Likewise.
437 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
438 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
439 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
440 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
441 * config/avr/avr.c (avr_fold_builtin): Likewise.
442 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
443 * config/msp430/msp430.c (msp430_attr): Likewise.
444 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
445 * config/powerpcspe/powerpcspe-c.c
446 (altivec_resolve_overloaded_builtin): Likewise.
447 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
448 (rs6000_expand_ternop_builtin): Likewise.
449 * config/rs6000/rs6000-c.c
450 (altivec_resolve_overloaded_builtin): Likewise.
451 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
452 (rs6000_expand_ternop_builtin): Likewise.
453 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
455 2017-10-10 Bin Cheng <bin.cheng@arm.com>
457 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
458 when copying loop nest with only one inner loop.
460 2017-10-10 Richard Biener <rguenther@suse.de>
462 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
463 blocks if SCEV is active.
464 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
466 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
467 (scev_initialize): Assert we are not yet initialized.
469 2017-10-10 Bin Cheng <bin.cheng@arm.com>
471 * tree-loop-distribution.c (generate_loops_for_partition): Remove
472 inner loop's exit stmt by making it always exit the loop, otherwise
473 we would generate an infinite empty loop.
475 2017-10-10 Bin Cheng <bin.cheng@arm.com>
477 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
478 renaming variables in new preheader if it's deleted.
480 2017-10-10 Bin Cheng <bin.cheng@arm.com>
482 * tree-loop-distribution.c (struct partition): Remove unused field
483 loops of the structure.
484 (partition_alloc, partition_free): Ditto.
485 (build_rdg_partition_for_vertex): Ditto.
487 2017-10-09 Jeff Law <law@redhat.com>
489 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
490 return type to match prototype and documentation.
492 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
494 * config/rs6000/rs6000.c (processor_costs): Move to ...
495 * config/rs6000/rs6000.h: ... here.
496 (rs6000_cost): Declare.
498 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
500 * except.c (setjmp_fn): New global variable.
501 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
502 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
503 if DONT_USE_BUILTIN_SETJMP is defined.
505 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
507 * target.def (insn_cost): New hook.
508 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
509 * doc/tm.texi: Regenerate.
510 * rtlanal.c (insn_cost): Use the new hook.
512 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
514 * combine.c (combine_validate_cost): Compute the new insn_cost,
515 not just pattern_cost.
516 (try_combine): Adjust comment.
518 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
520 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
522 * combine.c (uid_insn_cost): Adjust comment.
523 (combine_validate_cost): Adjust comment. Use pattern_cost instead
525 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
526 * dse.c (find_shift_sequence): Ditto.
527 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
528 (bb_valid_for_noce_process_p): Use pattern_cost.
529 * rtl.h (insn_rtx_cost): Delete.
530 (pattern_cost): New prototype.
531 (insn_cost): New prototype.
532 * rtlanal.c (insn_rtx_cost): Rename to...
533 (pattern_cost): ... this.
536 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
538 * config/i386/i386.md (*jcc_2): Remove insn pattern.
539 (*jcc<mode>_0_r_i387): Ditto.
540 (*jccxf_r_i387): Ditto.
541 (*jcc<mode>_r_i387): Ditto.
542 (*jccu<mode>_r_i387): Ditto.
543 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
544 (*jcc): Rename from *jcc_1.
546 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
548 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
549 deferred rescans after the lvx/stvx recombination pre-pass.
551 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
553 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
554 memory operation instruction support.
555 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
556 (rs6000-ibm-aix[789]*): Likewise.
557 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
558 Document new functions.
560 2017-10-09 Richard Biener <rguenther@suse.de>
562 PR tree-optimization/82397
563 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
564 equality only for semantically equal trees.
566 2017-10-09 Richard Biener <rguenther@suse.de>
568 PR tree-optimization/82449
569 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
570 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
571 allow constant addresses.
572 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
575 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
577 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
580 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
583 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
586 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
589 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
591 2017-10-09 Jakub Jelinek <jakub@redhat.com>
594 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
595 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
597 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
599 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
600 (WI_BINARY_PREDICATE_RESULT): Likewise.
601 (wi::binary_traits::operator_result): New type.
602 (wi::binary_traits::predicate_result): Likewise.
603 (generic_wide_int::operator~, unary generic_wide_int::operator-)
604 (generic_wide_int::operator==, generic_wide_int::operator!=)
605 (generic_wide_int::operator&, generic_wide_int::and_not)
606 (generic_wide_int::operator|, generic_wide_int::or_not)
607 (generic_wide_int::operator^, generic_wide_int::operator+
608 (binary generic_wide_int::operator-, generic_wide_int::operator*):
610 (operator~, unary operator-, operator==, operator!=, operator&)
611 (operator|, operator^, operator+, binary operator-, operator*): New
613 * expr.c (get_inner_reference): Use wi::bit_and_not.
614 * fold-const.c (fold_binary_loc): Likewise.
615 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
616 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
617 (bit_value_binop): Likewise.
618 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
619 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
620 (extract_range_from_binary_expr_1): Likewise.
621 (masked_increment): Likewise.
622 (simplify_bit_ops_using_ranges): Likewise.
624 2017-10-09 Martin Jambor <mjambor@suse.cz>
627 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
628 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
629 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
631 (hsa_fixup_mov_insn_type): New function.
632 (hsa_op_with_type::get_in_type): Use it.
633 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
634 immediates in an assert.
635 (hsa_op_with_type::extend_int_to_32bit): New method.
636 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
637 types. Convert to dest type if necessary.
638 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
639 (reg_for_gimple_ssa): Pass false as min32int to
640 hsa_type_for_scalar_tree_type.
641 (gen_hsa_addr): Fixup type when creating addresable temporary.
642 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
643 (gen_hsa_unary_operation): Extend operands and convert to dest type if
644 necessary. Call hsa_fixup_mov_insn_type.
645 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
646 extend operands and convert to dest type if necessary.
647 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
648 to dest type if necessary.
649 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
650 if conversion nt necessary and size matches.
651 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
652 to dest type if necessary.
653 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
654 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
656 (gen_hsa_clrsb): Likewise.
657 (gen_hsa_ffs): Likewise.
658 (gen_hsa_divmod): Extend operands and convert to dest type if
660 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
662 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
664 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
665 Remove empty default arguments. Use a brace block as output
667 (conditional return): Ditto.
669 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
670 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
672 (group_ending_nop): Ditto.
674 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
675 (splitters for those): Ditto.
677 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
679 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
680 a conditional jump (and the compare for it) so that pc_rtx is the
682 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
683 for the deleted and renamed ctr<mode>_internal[234] patterns.
684 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
685 Delete second conditional return pattern.
686 (ctr<mode>_internal2): Delete this second bdnz pattern.
687 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
688 (ctr<mode>_internal4): Delete this second bdz pattern.
690 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
692 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
693 (always_initialized_rtx_for_ssa_name_p): New predicate.
694 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
695 (finish_out_of_ssa): Free new field of SA.
696 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
697 * tree-ssa-coalesce.c: Include tree-ssa.h.
698 (get_parm_default_def_partitions): Remove extern keyword.
699 (get_undefined_value_partitions): New function.
700 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
701 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
703 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
705 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
707 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
709 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
711 * i386.c (ix86_expand_set_or_movmem): Disable 512bit loops for targets
714 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
716 * i386.c (has_dispatch): Disable for Ryzen.
718 2017-10-08 Olivier Hainque <hainque@adacore.com>
720 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
721 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
723 (thumb_set_return_address): Likewise.
725 2017-10-08 Olivier Hainque <hainque@adacore.com>
727 * common/config/arm/arm-common.c (arm_except_unwind_info):
728 Handle DWARF2_UNWIND_INFO.
730 2017-10-07 Michael Collison <michael.collison@arm.com>
732 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
735 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
737 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
738 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
739 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
740 defined, force the creation of a new block for a dispatch label.
742 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
744 * invoke.texi (Wsuggest-attribute=cold): Document.
745 * common.opt (Wsuggest-attribute=cold): New
746 * ipa-pure-const.c (warn_function_cold): New function.
747 * predict.c (compute_function_frequency): Use it.
748 * predict.h (warn_function_cold): Declare.
750 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
752 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
755 2017-10-06 Martin Liska <mliska@suse.cz>
757 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
758 keyword for member functions.
759 (struct sanopt_tree_couple): New struct.
760 (struct sanopt_tree_couple_hash): New function.
761 (struct sanopt_ctx): Add new hash_map.
762 (has_dominating_ubsan_ptr_check): New function.
763 (record_ubsan_ptr_check_stmt): Likewise.
764 (maybe_optimize_ubsan_ptr_ifn): Likewise.
765 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
766 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
768 2017-10-06 Sudakshina Das <sudi.das@arm.com>
771 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
772 aarch64_simd_valid_immediate on CONST_VECTORs.
773 (aarch64_reg_or_bic_imm): Likewise.
775 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
777 PR rtl-optimization/82396
778 * haifa-sched.c (ready_sort_real): Disable qsort checking.
780 2017-10-06 Sebastian Pop <sebpop@gmail.com>
782 * graphite-dependences.c (scop_get_reads): Move code to...
783 (scop_get_must_writes): Move code to...
784 (scop_get_may_writes): Move code to...
785 (scop_get_reads_and_writes): ... here.
786 (scop_get_dependences): Call scop_get_reads_and_writes.
788 2017-10-06 Jakub Jelinek <jakub@redhat.com>
790 PR tree-optimization/82434
791 * fold-const.h (can_native_encode_type_p,
792 can_native_encode_string_p): Remove.
793 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
794 don't encode anything, just return what would be otherwise returned.
795 (native_encode_fixed, native_encode_complex, native_encode_vector):
797 (native_encode_string): Likewise. Inline by hand
798 can_native_encode_string_p.
799 (can_native_encode_type_p): Remove.
800 (can_native_encode_string_p): Remove.
801 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
802 STRING_CSTs using can_native_encode_string_p, test all
803 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
804 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
805 argument from native_encode_expr.
806 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
807 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
810 2017-10-06 Richard Biener <rguenther@suse.de>
812 PR tree-optimization/82397
813 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
814 operand_equal_p but rely on data_ref_compare_tree for detecting
816 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
817 to match up with dr_group_sort_cmp.
819 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
822 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
824 * config/s390/s390-builtin-types.def: Regenerate.
826 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
829 * config/s390/s390-builtin-types.def: Regenerate.
830 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
831 Change flag from B_VXE to B_VX.
832 (s390_vec_min_dbl): Remove B_VXE flag.
834 2017-10-06 Richard Biener <rguenther@suse.de>
836 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
837 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
838 translate_isl_ast_to_gimple::is_valid_rename,
839 translate_isl_ast_to_gimple::get_rename,
840 translate_isl_ast_to_gimple::get_def_bb_for_const,
841 translate_isl_ast_to_gimple::get_new_name,
842 translate_isl_ast_to_gimple::collect_all_ssa_names,
843 translate_isl_ast_to_gimple::copy_loop_phi_args,
844 translate_isl_ast_to_gimple::collect_all_ssa_names,
845 translate_isl_ast_to_gimple::copy_loop_phi_args,
846 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
847 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
848 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
849 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
850 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
851 translate_isl_ast_to_gimple::copy_cond_phi_args,
852 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
853 translate_isl_ast_to_gimple::edge_for_new_close_phis,
854 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
855 translate_isl_ast_to_gimple::rename_uses,
856 translate_isl_ast_to_gimple::rename_all_uses): Remove.
857 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
858 (set_rename_for_each_def): Likewise.
859 (graphite_copy_stmts_from_block): Handle debug stmt resetting
860 here. Handle rewriting SCEV analyzable uses here.
861 (copy_bb_and_scalar_dependences): Generate code for PHI
863 (graphite_regenerate_ast_isl): Adjust.
864 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
865 (add_write, add_read): New functions.
866 (build_cross_bb_scalars_def): Use it and simplify.
867 (build_cross_bb_scalars_use): Likewise.
868 (graphite_find_cross_bb_scalar_vars): Inline into...
869 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
870 simulating out-of-SSA. Compute liveout and add dependencies.
871 (build_scops): Force an empty entry block.
872 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
874 (sese_build_liveouts): Declare.
875 (sese_trivially_empty_bb_p): Likewise.
876 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
877 compute liveout and debug_liveout.
878 (sese_bad_liveouts_use): Remove.
879 (sese_reset_debug_liveouts_bb): Likewise.
880 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
881 (sese_build_liveouts): Build liveout and debug_liveout and store
883 (new_sese_info): Adjust.
884 (free_sese_info): Likewise.
885 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
886 do not build liveout here.
887 (move_sese_in_condition): Adjust region entry.
888 (scev_analyzable_p): Match up with chrec_apply requirements.
889 (sese_trivially_empty_bb_p): New.
890 * tree-into-ssa.c (get_reaching_def): Properly support generating
891 default-defs for incremental rewrite of anonymous names.
893 2017-10-06 Richard Biener <rguenther@suse.de>
895 * graphite-sese-to-poly.c (extract_affine): For casts increasing
896 precision do not perform modulo reduction.
898 2017-10-06 Richard Biener <rguenther@suse.de>
900 PR tree-optimization/82436
901 * tree-vect-slp.c (vect_supported_load_permutation_p): More
902 conservatively choose the vectorization factor when checking
903 whether we can perform the required load permutation.
904 (vect_transform_slp_perm_load): Assert when we may not fail.
906 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
908 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
909 message for incompatible -msdata=* and -mcall-* options.
911 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
913 * i386.c (ia32_multipass_dfa_lookahead): Default to issue rate
914 for post-reload scheduling.
916 2017-10-05 Tamar Christina <tamar.christina@arm.com>
918 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
920 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
922 * i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
923 to improve monte carlo in scimark.
925 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
927 * i386.c (ix86_size_cost, i386_cost, i486_cost, pentium_cost,
928 lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
929 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
930 pentium4_cost, nocona_cost): Set reassociation width to 1.
931 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
932 width to 2 for fp operations and 1 otherwise.
933 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
935 (atom_cost): Set reassociation width to 2.
936 (slm_cost, generic_cost): Set fp reassociation width
937 to 2 and 1 otherwise.
938 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
939 (core_cost): Set fp reassociation width to 4 and vector to 2.
940 (ix86_reassociation_width): Rewrite using cost table; special case
941 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
942 and TARGET_AVX128_OPTIMAL.
943 * i386.h (processor_costs): Add
944 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
945 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
946 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
947 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
948 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
949 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
951 2017-10-05 Nathan Sidwell <nathan@acm.org>
953 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
955 2017-10-05 Tamar Christina <tamar.christina@arm.com>
957 * config/arm/arm.c (arm_test_fpu_data): New.
958 (arm_run_selftests): Call arm_test_fpu_data.
960 2017-10-04 Nathan Sidwell <nathan@acm.org>
962 * toplev.c (toplev::main): Remove excess parens on pretty_printer
964 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
966 2017-10-04 Sudakshina Das <sudi.das@arm.com>
968 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
969 check type for aarch64_simd_valid_immediate.
970 (aarch64_output_simd_mov_immediate): Update prototype.
971 (aarch64_simd_valid_immediate): Update prototype.
972 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
973 support for ORR-immediate.
974 (and<mode>3): modified pattern to add support for BIC-immediate.
975 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
976 now checks for valid immediate for BIC and ORR based on new enum
978 (aarch64_output_simd_mov_immediate): Function now used to output
979 BIC/ORR imm as well based on new enum argument.
980 * config/aarch64/constraints.md (Do): New vector immediate constraint.
982 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
983 (aarch64_reg_or_bic_imm): Likewise.
985 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
987 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
988 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
990 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
994 PR rtl-optimization/82396
995 * haifa-sched.c (autopref_multipass_init): Simplify
997 (autopref_rank_data): Simplify sort order.
998 * sched-int.h (autopref_multipass_data_): Remove
999 multi_mem_insn_p, min_offset and max_offset.
1001 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1003 * doc/sourcebuild.texi: Document vect_peeling_profitable.
1005 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1007 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
1010 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1012 * doc/sourcebuild.texi: Document vect_long_mult.
1014 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
1016 PR tree-optimization/82413
1017 * fold-const.c (build_range_check): Use widest_int when comparing
1018 the maximum ETYPE value with HIGH.
1020 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
1022 PR rtl-optimization/82396
1023 * haifa-sched.c (autopref_multipass_init): Simplify
1025 (autopref_rank_data): Simplify sort order.
1026 * sched-int.h (autopref_multipass_data_): Remove
1027 multi_mem_insn_p, min_offset and max_offset.
1029 2017-10-04 Jakub Jelinek <jakub@redhat.com>
1031 PR tree-optimization/82381
1032 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
1033 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
1036 PR tree-optimization/82374
1037 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
1038 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
1039 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
1040 current_function_decl to the new decl.
1042 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
1044 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
1045 helper macro for IEEE float128 hardware built-in functions.
1046 (SQRTF128_ODD): Add built-in functions with the round-to-odd
1048 (TRUNCF128_ODD): Likewise.
1049 (ADDF128_ODD): Likewise.
1050 (SUBF128_ODD): Likewise.
1051 (MULF128_ODD): Likewise.
1052 (DIVF128_ODD): Likewise.
1053 (FMAF128_ODD): Likewise.
1054 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
1055 UNSPEC_TRUNC_ROUND_TO_ODD.
1056 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
1057 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
1058 floating point round to odd instructions.
1059 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
1060 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
1061 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
1062 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
1063 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
1064 (trunc<mode>sf2_hw): Change the truncate with round to odd
1065 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
1066 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
1067 to odd hardware instructions.
1068 (sub<mode>3_odd): Likewise.
1069 (mul<mode>3_odd): Likewise.
1070 (div<mode>3_odd): Likewise.
1071 (sqrt<mode>2_odd): Likewise.
1072 (fma<mode>4_odd): Likewise.
1073 (fms<mode>4_odd): Likewise.
1074 (nfma<mode>4_odd): Likewise.
1075 (nfms<mode>4_odd): Likewise.
1076 (trunc<mode>df2_odd): Change the truncate with round to odd
1077 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
1079 * doc/extend.texi (PowerPC built-in functions): Update documentation
1080 for existing IEEE float128-bit built-in functions. Add built-in
1081 functions that generate the IEEE 128-bit floating point round to
1084 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
1086 PR rtl-optimization/77729
1087 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
1088 to (X&(C1&~C2))|C2 transformations.
1090 2017-10-03 Martin Jambor <mjambor@suse.cz>
1092 PR tree-optimization/82363
1093 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
1094 mismatch, mark lacc written regardless of racc.
1096 2017-10-03 Jakub Jelinek <jakub@redhat.com>
1098 PR tree-optimization/82381
1099 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
1100 stmt_to_insert nor wheather SSA_NAMEs are default defs.
1101 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
1102 fallthrough into reassoc_stmt_dominates_stmt_p.
1105 * combine.c (combine_instructions): Don't combine in unreachable
1108 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
1111 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
1112 function to not use the have_cpu variable. Do not set cpu_index,
1113 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
1115 (rs6000_valid_attribute_p): Remove duplicate initializations of
1116 old_optimize and func_optimize.
1117 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
1118 (rs6000_activate_target_options): Make global.
1119 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
1122 2017-10-02 Jakub Jelinek <jakub@redhat.com>
1124 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
1125 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
1126 Set *poffset to 0 and *psize and *pmax_size to -1 if
1127 *poffset + *psize overflows in HOST_WIDE_INT.
1129 PR tree-optimization/82387
1130 PR tree-optimization/82388
1131 PR tree-optimization/82389
1132 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
1133 instead of live_bytes non-NULL.
1135 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
1138 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
1141 2017-10-02 Richard Biener <rguenther@suse.de>
1143 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
1144 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
1145 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
1147 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1149 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
1150 requested precision matches the type's.
1151 * calls.c (alloc_max_size): Calculate the new candidate size as
1152 a widest_int and use wi::to_widest when comparing it with the
1153 current candidate size.
1154 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
1155 zero rather than integer_zero_node.
1156 * match.pd: Check for a no-op conversion before using wi::add
1157 rather than after. Use tree_to_uhwi when summing small shift
1158 counts into an unsigned int.
1160 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1161 Alan Hayward <alan.hayward@arm.com>
1162 David Sherwood <david.sherwood@arm.com>
1165 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
1166 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
1167 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
1168 POINTER_AND_FP_REGS.
1170 2017-10-02 Richard Biener <rguenther@suse.de>
1172 PR tree-optimization/82355
1173 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1174 a mapping for the enclosing loop but avoid generating one for
1176 (copy_bb_and_scalar_dependences): Remove premature codegen
1177 error on PHIs in blocks duplicated into multiple places.
1178 * graphite-scop-detection.c
1179 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1180 in the region use it as loop and nest to analyze the DR in.
1181 (try_generate_gimple_bb): Likewise.
1182 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1183 (add_loop_constraints): For blocks in a loop not in the region
1184 create a dimension with a single iteration.
1185 * sese.h (gbb_loop_at_index): Remove assert.
1187 2017-10-01 Kevin Buettner <kevinb@redhat.com>
1189 * omp-expand.c (adjust_context_scope): New function.
1190 (expand_parallel_call): Call adjust_context_scope.
1192 2017-10-01 Jeff Law <law@redhat.com>
1194 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
1195 dom_opt_dom_walker class with direct access to private members.
1196 Add comments. Call test_for_singularity.
1197 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
1198 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
1199 m_dummy_cond anymore.
1200 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
1202 (pass_dominator:execute): Build the dummy_cond here and pass it
1203 to the dom_opt_dom_walker ctor.
1204 (test_for_singularity): New function.
1206 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
1207 Maya Rashish <coypu@sdf.org>
1209 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
1210 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
1211 (alpha*-*-netbsd*) Use nbsd_tm_file.
1212 (arm*-*-netbsdelf*) Likewise.
1213 (i[34567]86-*-netbsdelf*) Likewise.
1214 (x86_64-*-netbsd*) Likewise.
1215 (mips*-*-netbsd*) Likewise.
1216 (powerpc-*-netbsd*) Likewise.
1217 (sh*-*-netbsd*) Likewise.
1218 (sparc-*-netbsdelf*) Likewise.
1219 (sparc64-*-netbsd*) Likewise.
1220 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
1222 (vax-*-netbsdelf*) Likewise.
1223 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
1224 (UINT_FAST8_TYPE) Likewise.
1225 (INT_FAST16_TYPE) Check CHAR_FAST16.
1226 (UINT_FAST16_TYPE) Likewise.
1228 2017-09-30 Jakub Jelinek <jakub@redhat.com>
1231 * config/i386/i386.md
1232 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
1233 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
1234 *divmodsi4_zext_2): New define_insn_and_split.
1235 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
1236 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
1237 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
1238 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
1239 New define_insn_and_split.
1240 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
1241 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
1242 operands[1] having DImode when mode is SImode.
1244 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
1245 always SImode for DIV and MOD in REG_EQUAL notes.
1247 2017-09-29 Yury Gribov <tetra2005@gmail.com>
1250 * match.pd: Fix handling of NaNs in pattern.
1252 2017-09-29 Jeff Law <law@redhat.com>
1254 * sbitmap.c (bitmap_bit_in_range_p): New function.
1255 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
1256 * tree-ssa-dse.c (live_bytes_read): New function.
1257 (dse_classify_store): Ignore reads of dead bytes.
1259 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
1260 typos and whitespace errors.
1261 * config/i386/predicates.md (address_no_seg_operand): Likewise.
1262 * config/s390/s390.c (s390_emit_prologue): Likewise.
1264 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
1267 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
1268 with a symbol for LRA.
1270 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
1272 PR rtl-optimization/82338
1273 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
1275 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
1277 * genmodes.c (calc_wider_mode): Suppress qsort macro.
1278 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
1279 (qsort_chk): Declare.
1280 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
1281 (qsort_chk): New function.
1283 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1285 PR tree-optimization/82337
1286 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
1287 phi definition if the PHI result appears in an abnormal PHI.
1288 (find_basis_for_base_expr): Don't record a basis if the LHS of the
1289 basis appears in an abnormal PHI.
1291 2017-09-29 Richard Biener <rguenther@suse.de>
1293 * graphite-isl-ast-to-gimple.c
1294 (translate_isl_ast_to_gimple::set_codegen_error): New function.
1295 (binary_op_to_tree): Use it.
1296 (get_rename_from_scev): Likewise.
1297 (copy_loop_phi_nodes): Likewise.
1298 (copy_bb_and_scalar_dependences): Likewise.
1299 (translate_pending_phi_nodes): Likewise.
1301 2017-09-29 Jakub Jelinek <jakub@redhat.com>
1304 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
1305 for movabsq $(i32 << shift), r64.
1307 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
1309 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
1310 index when encoding %esp as %rsp to avoid 0x67 prefix.
1312 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
1314 * config/i386/i386.md (*movsf_internal, *movdf_internal):
1315 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
1317 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
1319 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
1320 Extensions with more than 16 double VFP registers.
1321 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
1322 to_clear_mask and all code related to it. Replace the remaining
1323 entry by a sbitmap and adapt code accordingly.
1325 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
1327 * brig-builtins.def: Change pure attributes to const.
1329 2017-09-28 Joseph Myers <joseph@codesourcery.com>
1331 * config.gcc (default_gnu_indirect_function): Default to yes for
1332 sparc*-*-linux* with glibc.
1334 2017-09-28 Joseph Myers <joseph@codesourcery.com>
1336 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
1337 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
1338 when creating .init_array and .fini_array sections with priority
1341 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
1344 * config/aarch64/aarch64.c
1345 (aarch64_builtin_support_vector_misalignment): Always return false
1346 when misalignment is unknown.
1348 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
1350 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
1351 this function to return false if the definition used by the swap
1352 instruction is artificial, or if the memory address from which the
1353 constant value is loaded is not represented by a base address held
1354 in a register or if the base address register is a frame or stack
1355 pointer. Additionally, return false if the base address of the
1356 loaded constant is a SYMBOL_REF but is not considered to be a
1358 (replace_swapped_load_constant): New function.
1359 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
1360 loaded constant vector with a load of a swapped constant vector.
1362 2017-09-27 Carl Love <cel@us.ibm.com>
1364 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
1365 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
1366 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
1369 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
1371 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
1372 first, always call autopref_rank_data otherwise.
1374 2017-09-27 Richard Biener <rguenther@suse.de>
1376 * graphite-scop-detection.c (find_scop_parameters): Move
1377 loop bound handling ...
1378 (gather_bbs::before_dom_children): ... here, avoiding the need
1379 to build scop_info->loop_nest.
1380 (record_loop_in_sese): Remove.
1381 * sese.h (sese_info_t::loop_nest): Remove.
1382 * sese.c (new_sese_info): Do not allocate loop_nest.
1383 (free_sese_info): Do not free loop_nest.
1385 2017-09-27 Jakub Jelinek <jakub@redhat.com>
1388 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
1389 lhs from calls if the lhs has addressable type.
1391 2017-09-27 Richard Biener <rguenther@suse.de>
1393 * graphite.h (scop::max_alias_set): New member.
1394 * graphite-scop-detection.c: Remove references to non-existing
1395 --param in comments.
1396 (build_alias_sets): Record the maximum alias set used for drs.
1397 (build_scops): Support zero as unlimited for
1398 --param graphite-max-arrays-per-scop.
1399 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
1401 (build_poly_sr_1): ... here. Compute alias set based on the
1402 maximum alias set used for drs rather than
1403 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
1405 2017-09-27 Richard Biener <rguenther@suse.de>
1407 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
1408 --param loop-block-tile-size=0 to disable tiling.
1410 2017-09-27 Richard Biener <rguenther@suse.de>
1412 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
1413 (graphite-max-nb-scop-params): Document special value zero.
1414 * domwalk.h (dom_walker::STOP): New symbolical constant.
1415 (dom_walker::dom_walker): Add optional parameter for bb to
1417 (dom_walker::~dom_walker): Declare.
1418 (dom_walker::before_dom_children): Document STOP return value.
1419 (dom_walker::m_user_bb_to_rpo): New member.
1420 (dom_walker::m_bb_to_rpo): Likewise.
1421 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
1422 mapping here if not provided by the user.
1423 (dom_walker::~dom_walker): Free bb to RPO mapping if not
1424 provided by the user.
1425 (dom_walker::STOP): Define.
1426 (dom_walker::walk): Do not compute bb to RPO mapping here.
1427 Support STOP return value from before_dom_children to stop
1429 * graphite-optimize-isl.c (optimize_isl): If the schedule
1430 is the same still generate code if -fgraphite-identity
1431 or -floop-parallelize-all are given.
1432 * graphite-scop-detection.c: Include cfganal.h.
1433 (gather_bbs::gather_bbs): Get and pass through bb to RPO
1435 (gather_bbs::before_dom_children): Return STOP for BBs
1437 (build_scops): Compute bb to RPO mapping and pass it to
1438 the domwalk. Treat --param graphite-max-nb-scop-params=0
1439 as not limiting the number of params.
1440 * graphite.c (graphite_initialize): Remove limit on the
1441 number of basic-blocks in a function.
1442 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
1443 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
1444 default value of 10.
1446 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1448 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
1449 Adjust code to eliminate needing to do the shift right 32-bits
1450 operation after XSCVDPSPN.
1452 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1454 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
1455 ((X / Y) != 0 -> X >= Y): Likewise.
1457 2017-09-26 Carl Love <cel@us.ibm.com>
1459 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
1460 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
1461 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
1462 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
1463 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
1464 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
1465 definitions and overloading.
1466 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
1467 statement for P9V_BUILTIN_XST_LEN_R.
1468 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
1469 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
1470 define_expand and define_insn for the instructions and builtins.
1471 * doc/extend.texi: Update the built-in documentation file for the new
1473 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
1474 define_insn for the instructions
1476 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
1479 * gcc/config/netbsd-protos.h: New file.
1480 * gcc/config/netbsd.c: New file.
1481 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
1482 * gcc/config/t-netbsd: New file.
1483 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
1484 (tmake_file) Add t-netbsd.
1485 (extra_objs) Add netbsd.o.
1487 2017-09-26 Janus Weil <janus@gcc.gnu.org>
1491 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
1493 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1495 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
1496 sign extension from a vector register to a GPR by doing a 32-bit
1497 direct move and then an EXTSW.
1498 (extendsi<mode>2 splitter): Likewise.
1499 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
1500 right or vector extract after doing XSCVDPSPN. Use
1501 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
1503 (movdi_from_sf_zero_ext): Likewise.
1504 (reload_gpr_from_vsxsf): Likewise.
1505 (p8_mfvsrd_4_disf): Delete, no longer used.
1506 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
1507 then needing to move the SFmode to a GPR to use the XSCVDPSP
1508 instruction instead of FRSP and XSCVDPSPN.
1509 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
1510 it is adjacent to the other XSCVSPDP insns.
1511 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
1512 SFmode to be in traditional Altivec registers.
1513 (vsx_xscvdpspn): Eliminate useless alternative constraint.
1514 (vsx_xscvspdpn): Likewise.
1515 (vsx_xscvspdpn_scalar): Likewise.
1517 2017-09-26 Martin Jambor <mjambor@suse.cz>
1519 * tree-sra.c (compare_access_positions): Put integral types first,
1520 stabilize sorting of integral types, remove conditions putting
1521 non-full-precision integers last.
1522 (sort_and_splice_var_accesses): Disable scalarization if a
1523 non-integert would be represented by a non-full-precision integer.
1525 2017-09-26 Joseph Myers <joseph@codesourcery.com>
1527 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
1528 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
1529 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
1530 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
1531 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
1532 conditionals inside the function instead of around it. Call
1533 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
1534 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
1536 2017-09-26 Richard Biener <rguenther@suse.de>
1538 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
1540 (scop_detection::build_scop_breadth): ... this. Removed.
1541 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
1542 (scop_detection::harmful_stmt_in_bb): Likewise.
1543 (scop_detection::graphite_can_represent_stmt): Likewise.
1544 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
1545 (scop_detection::can_represent_loop): Remove recursion, fold in ...
1546 (scop_detection::can_represent_loop_1): ... this. Removed.
1547 (scop_detection::harmful_loop_in_region): Simplify after inlining
1548 the above and remove more quadraticness.
1549 (build_scops): Adjust.
1550 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
1553 2017-09-26 Jakub Jelinek <jakub@redhat.com>
1556 * config/i386/i386.c (ix86_print_operand_address_as): Only test
1557 REGNO (base) == SP_REG if base is a REG.
1560 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
1561 if it is different SSA_NAME.
1562 (optimize_range_tests_cmp_bitwise): New function.
1563 (optimize_range_tests): Call it.
1565 2017-09-26 Richard Biener <rguenther@suse.de>
1567 PR tree-optimization/82321
1568 * graphite.c (canonicalize_loop_closed_ssa): Properly check
1569 for the def being inside the loop.
1571 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1573 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
1575 * config/s390/s390-builtins.def: Fix constraint on op4.
1577 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1579 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
1580 independent expanders.
1581 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
1582 ("vec_ordered", "vec_unordered"): New expanders.
1584 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1586 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
1589 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1591 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
1592 vec_unpacks_lo_v16qi.
1593 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
1595 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1597 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
1598 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
1599 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
1601 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1603 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
1605 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
1607 ("*vec_slb<mode>"): New insn pattern.
1608 ("vec_shr_<mode>"): New expander.
1609 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
1610 and force the shift count operand to V16QImode.
1611 ("vec_srb<mode>"): Set shift count mode to V16QI.
1613 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1615 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
1616 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
1617 ("vec_widen_smult_hi_<mode>"): New expander definitions.
1619 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
1622 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
1624 2017-09-26 Richard Biener <rguenther@suse.de>
1626 PR tree-optimization/82320
1627 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
1630 2017-09-25 Jeff Law <law@redhat.com>
1632 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
1633 prototype for new argument.
1634 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
1635 mostly extracted from rs6000_emit_allocate_stack.
1636 (rs6000_emit_probe_stack_range_stack_clash): New function.
1637 (rs6000_emit_allocate_stack): Call
1638 rs6000_emit_probe_stack_range_stack_clash as needed.
1639 (rs6000_emit_probe_stack_range): Add additional argument
1640 to call to gen_probe_stack_range{si,di}.
1641 (output_probe_stack_range): New.
1642 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
1643 (output_probe_stack_range_stack_clash): New.
1644 (rs6000_emit_prologue): Emit notes into dump file as requested.
1645 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
1646 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
1647 Add additional operand and pass it to output_probe_stack_range.
1649 2017-09-25 Bin Cheng <bin.cheng@arm.com>
1651 PR tree-optimization/82163
1652 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
1653 (checking_verify_loop_closed_ssa): New parameter.
1654 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
1655 (check_loop_closed_ssa_stmt): Delete.
1656 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
1657 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
1658 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
1661 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
1663 * brig-builtins.def: Treat HSAIL barrier builtins as
1664 setjmp/longjump style functions.
1666 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1668 * target.def (constant_alignment): New hook.
1669 * defaults.h (CONSTANT_ALIGNMENT): Delete.
1670 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
1671 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
1672 * doc/tm.texi: Regenerate.
1673 * targhooks.h (default_constant_alignment): Declare.
1674 (constant_alignment_word_strings): Likewise.
1675 * targhooks.c (default_constant_alignment): New function.
1676 (constant_alignment_word_strings): Likewise.
1677 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
1678 instead of CONSTANT_ALIGNMENT.
1679 * varasm.c (align_variable, get_variable_align, build_constant_desc)
1680 (force_const_mem): Likewise.
1681 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
1682 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
1683 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
1684 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1685 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
1687 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
1688 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1689 constant_alignment_word_strings.
1690 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
1691 (CONSTANT_ALIGNMENT): Likewise.
1692 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1693 (arm_constant_alignment): New function.
1694 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
1695 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1696 constant_alignment_word_strings.
1697 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
1698 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1699 constant_alignment_word_strings.
1700 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
1701 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1702 (cris_constant_alignment): New function.
1703 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
1704 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1705 (epiphany_constant_alignment): New function.
1706 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
1707 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1708 constant_alignment_word_strings.
1709 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
1710 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1711 constant_alignment_word_strings.
1712 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
1713 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1714 constant_alignment_word_strings.
1715 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
1716 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
1717 * config/i386/i386.c (ix86_constant_alignment): Make static.
1718 Use the same interface as the target hook.
1719 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1720 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
1721 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1722 constant_alignment_word_strings.
1723 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
1724 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
1725 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1726 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
1727 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1728 constant_alignment_word_strings.
1729 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
1730 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1731 constant_alignment_word_strings.
1732 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
1733 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1734 constant_alignment_word_strings.
1735 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
1736 * config/microblaze/microblaze.c (microblaze_constant_alignment):
1738 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1739 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
1740 * config/mips/mips.c (mips_constant_alignment): New function.
1741 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1742 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
1743 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
1744 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1745 (mmix_constant_alignment): Make static. Use the same interface
1747 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
1748 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1749 constant_alignment_word_strings.
1750 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
1751 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1752 constant_alignment_word_strings.
1753 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
1754 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1755 constant_alignment_word_strings.
1756 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
1757 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1758 (rs6000_constant_alignment): New function.
1759 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
1760 * config/riscv/riscv.c (riscv_constant_alignment): New function.
1761 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1762 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
1763 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1764 (rs6000_constant_alignment): New function.
1765 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
1766 * config/s390/s390.c (s390_constant_alignment): New function.
1767 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1768 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
1769 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1770 constant_alignment_word_strings.
1771 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
1772 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1773 (sparc_constant_alignment): New function.
1774 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
1775 * config/spu/spu.c (spu_constant_alignment): New function.
1776 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1777 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
1778 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1779 constant_alignment_word_strings.
1780 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
1781 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1782 constant_alignment_word_strings.
1783 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
1784 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
1785 constant_alignment_word_strings.
1786 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
1787 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1788 (visium_constant_alignment): New function.
1789 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
1790 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
1791 (xtensa_constant_alignment): New function.
1792 * system.h (CONSTANT_ALIGNMENT): Poison.
1794 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
1796 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
1797 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
1798 (rs6000_builtin_valid_without_lhs): New helper function.
1799 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1800 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
1802 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
1804 * target.h (vec_perm_indices): Use unsigned short rather than
1806 (auto_vec_perm_indices): Likewise.
1807 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
1808 Use unsigned int rather than unsigned char.
1809 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
1811 2017-09-25 Richard Biener <rguenther@suse.de>
1813 * cfgloop.h (sort_sibling_loops): Declare.
1814 * cfgloop.c (sort_sibling_loops_cmp): New helper.
1815 (sort_sibling_loops): New function sorting the sibling loop list
1817 * graphite.c (graphite_transform_loops): Sort sibling loops.
1819 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
1821 * target.def (vec_perm_const_ok): Change sel parameter to
1823 * optabs-query.c (can_vec_perm_p): Update accordingly.
1824 * doc/tm.texi: Regenerate.
1825 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
1826 auto_vec_perm_indices and remove separate nelt field.
1827 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
1828 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
1829 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
1830 (aarch64_expand_vec_perm_const): Update accordingly.
1831 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
1832 to vec_perm_indices.
1833 * config/arm/arm.c (expand_vec_perm_d): Change perm to
1834 auto_vec_perm_indices and remove separate nelt field.
1835 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
1836 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
1837 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
1839 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
1840 to vec_perm_indices.
1841 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
1842 sel to vec_perm_indices.
1843 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
1844 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
1845 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
1847 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
1850 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
1853 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
1854 on the FUNCTION_DECL function context if it has a DIE that is a
1857 2017-09-25 Richard Biener <rguenther@suse.de>
1859 PR tree-optimization/82285
1860 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
1863 2017-09-25 Tom de Vries <tom@codesourcery.com>
1867 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
1870 2017-09-25 Richard Biener <rguenther@suse.de>
1872 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
1873 ISL errors other than isl_error_quota happen. Dump if the
1874 schedule is the same.
1875 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
1876 errors instead of aborting inside ISL.
1878 2017-09-25 Iain Sandoe <iain@codesourcery.com>
1881 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
1882 of libgcc_eh for m64.
1883 * config/i386/darwin64.h: Likewise.
1885 2017-09-25 Richard Biener <rguenther@suse.de>
1888 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
1889 attribute for incomplete types nor twice for complete ones.
1891 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
1894 * config/i386/i386.c (ix86_print_operand_address_as): Encode
1895 %esp as %rsp to avoid 0x67 prefix if there is no index or base
1898 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
1901 * config/i386/i386.opt (mprefer-avx256): Use
1902 ix86_target_flags variable.
1903 * config/i386/i386.c (ix86_target_string): Move
1904 -mprefer-avx256 to flag2_opts.
1906 2017-09-22 Jakub Jelinek <jakub@redhat.com>
1909 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
1910 and x != -1 | y != -1 into (x & y) != -1.
1912 2017-09-22 Steve Ellcey <sellcey@cavium.com>
1914 * config.gcc: Add new case statement to set
1915 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
1916 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
1917 s390x-*-linux* case statements. Added aarch64 to the list of
1918 supported architectures.
1920 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1922 PR tree-optimization/82289
1923 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
1924 STMT_VINFO_RELEVANT_P.
1926 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1927 Alan Hayward <alan.hayward@arm.com>
1928 David Sherwood <david.sherwood@arm.com>
1930 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
1931 for VR_RANGE only; don't allow VR_ANTI_RANGE.
1932 (extract_range_from_binary_expr_1): Don't call
1933 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
1935 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1936 Alan Hayward <alan.hayward@arm.com>
1937 David Sherwood <david.sherwood@arm.com>
1939 * target.def (preferred_vector_alignment): New hook.
1940 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
1942 * doc/tm.texi: Regenerate.
1943 * targhooks.h (default_preferred_vector_alignment): Declare.
1944 * targhooks.c (default_preferred_vector_alignment): New function.
1945 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
1947 (DR_TARGET_ALIGNMENT): New macro.
1948 (aligned_access_p): Update commentary.
1949 (vect_known_alignment_in_bytes): New function.
1950 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
1952 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
1953 Calculate the misalignment based on the target alignment rather than
1955 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
1956 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
1957 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
1958 the target alignment, rather than masking the element misalignment
1959 with the number of elements in a vector. Also use the target
1960 alignment when calculating the maximum number of peels.
1961 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
1962 instead of TYPE_ALIGN_UNIT.
1963 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
1964 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
1965 (vect_create_addr_base_for_vector_ref): Update call accordingly.
1966 (vect_create_data_ref_ptr): Likewise.
1967 (vect_setup_realignment): Realign by ANDing with
1968 -DR_TARGET_MISALIGNMENT.
1969 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
1970 the number of peels based on DR_TARGET_ALIGNMENT.
1971 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
1972 with the guaranteed alignment boundary when deciding whether
1974 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
1975 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
1976 (ensure_base_align): Remove stmt_info parameter. Get the
1977 target base alignment from DR_TARGET_ALIGNMENT.
1978 (vectorizable_store): Update call accordingly. Interpret
1979 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
1981 (vectorizable_load): Likewise.
1983 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
1984 Alan Hayward <alan.hayward@arm.com>
1985 David Sherwood <david.sherwood@arm.com>
1987 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
1988 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
1989 (vect_enhance_data_refs_alignment): Likewise.
1991 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
1993 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
1994 error. Only quit immediately if parsing is complete.
1995 (BEGIN): Initialize fatal_err and parse_done.
1996 (begin fpu, end fpu): Check number of arguments.
1997 (begin arch, end arch): Likewise.
1998 (begin cpu, end cpu): Likewise.
1999 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
2000 (optalias): Likewise.
2002 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
2004 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
2005 * config/arm/arm-isa.h: Delete. Move definitions to ...
2006 * arm-cpus.in: ... here. Use new feature and fgroup values.
2007 * config/arm/arm.c (arm_option_override): Use lower case for feature
2009 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
2010 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
2011 * config/arm/parsecpu.awk (END): Add new command 'isa'.
2013 (print_isa_bits_for): New function.
2014 (gen_isa): New function.
2015 (gen_comm_data): Use print_isa_bits_for.
2016 (define feature): New keyword.
2017 (define fgroup): New keyword.
2018 * config/arm/t-arm (TM_H): Remove.
2019 (GTM_H): Add arm-isa.h.
2020 (arm-isa.h): Add rule to generate file.
2021 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
2022 case for feature bit names.
2024 2017-09-22 Richard Biener <rguenther@suse.de>
2026 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
2028 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
2029 print of no dependency loops ...
2030 * graphite.c (graphite_transform_loops): ... here.
2031 (canonicalize_loop_closed_ssa_form): Work from inner to outer
2033 (same_close_phi_node, remove_duplicate_close_phi,
2034 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
2035 (canonicalize_loop_closed_ssa): ... here and simplify.
2036 * graphite-optimize-isl.c: Include tree-vectorizer.h.
2037 (optimize_isl): Use dump_printf_loc to tell when we stopped
2038 optimizing because of an ISL timeout.
2040 2017-09-22 Richard Biener <rguenther@suse.de>
2042 PR tree-optimization/82291
2043 * tree-if-conv.c (predicate_mem_writes): Make sure to
2044 remove writes in blocks predicated with false.
2046 2017-09-22 Richard Biener <rguenther@suse.de>
2048 * sese.c: Include cfganal.h.
2049 (if_region_set_false_region): Remove.
2050 (create_if_region_on_edge): Likewise.
2051 (move_sese_in_condition): Re-implement without destroying
2054 2017-09-22 Richard Biener <rguenther@suse.de>
2056 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
2057 Verify both BBs contain loop PHI nodes before dispatching to
2059 (graphite_regenerate_ast_isl): Do not recompute dominators,
2060 do not verify three times. Restructure for clarity.
2061 * graphite-scop-detection.c (same_close_phi_node,
2062 remove_duplicate_close_phi, make_close_phi_nodes_unique,
2063 defined_in_loop_p, canonicalize_loop_closed_ssa,
2064 canonicalize_loop_closed_ssa_form): Simplify, remove excess
2065 checking and SSA rewrite, move to ...
2066 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
2067 (graphite_initialize): Do not pass in ctx, do not reset the
2068 SCEV cache, compute only dominators.
2069 (graphite_transform_loops): Allocate ISL ctx after
2070 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
2071 Maintain post-dominators only around build_scops.
2072 * sese.c (if_region_set_false_region): Make static. Free
2073 and recompute dominators.
2074 (move_sese_in_condition): Assert we don't get called with
2075 post-dominators computed.
2076 * sese.h (if_region_set_false_region): Remove.
2078 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
2080 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
2081 mode attribute for TARGET_AVX512VL.
2083 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
2085 * config/i386/i386.opt (mprefer-avx256): New option.
2086 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
2088 (ix86_preferred_simd_mode): Return 256-bit AVX modes
2089 for TARGET_PREFER_AVX256.
2090 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
2092 2017-09-21 Jeff Law <law@redhat.com>
2094 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
2095 Fix dump output if the only stack space is for pushed registers.
2097 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2099 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
2102 2017-09-21 Martin Sebor <msebor@redhat.com>
2105 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
2106 code (in C++) or code that triggers warnings.
2108 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
2110 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
2112 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
2114 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
2115 * sched-rgn.c: Ditto.
2116 * sel-sched-ir.c: Ditto.
2118 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
2120 * toplev.h (set_random_seed): Adjust return type.
2121 * toplev.c (init_local_tick): Move eager initialization of random_seed
2122 to get_random_seed. Adjust comment.
2123 (init_random_seed): Inline to get_random_seed, delete.
2124 (get_random_seed): Initialize random_seed lazily.
2125 (set_random_seed): Do not return previous value.
2126 (print_switch_value): Do not call get_random_seed.
2128 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
2130 * cgraph.c (delete_function_version): New, broken out from...
2131 (cgraph_node::delete_function_version): ...here. Rename to
2132 cgraph_node::delete_function_version_by_decl. Update all uses.
2133 (cgraph_node::remove): Call delete_function_version.
2135 2017-09-21 Jakub Jelinek <jakub@redhat.com>
2138 * tree-inline.c (expand_call_inline): Emit clobber stmts for
2139 VAR_DECLs to which addressable non-volatile parameters are mapped
2140 and for id->retvar after the return value assignment. Clear
2141 id->retval and id->retbnd after inlining.
2143 2017-09-21 Richard Biener <rguenther@suse.de>
2145 PR tree-optimization/82276
2146 PR tree-optimization/82244
2147 * tree-vrp.c (build_assert_expr_for): Set
2148 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
2150 (remove_range_assertions): Revert earlier change.
2152 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
2155 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
2157 2017-09-21 Richard Biener <rguenther@suse.de>
2159 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2160 Restore valid IL after code generation errors.
2161 * graphite.c (graphite_transform_loops): Diagnose code
2162 generation issues as MSG_MISSED_OPTIMIZATION and continue
2163 with processing SCOPs.
2165 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2166 Alan Hayward <alan.hayward@arm.com>
2167 David Sherwood <david.sherwood@arm.com>
2169 * calls.c (compute_argument_addresses): Use simplify_gen_binary
2170 rather than choosing between plus_constant and gen_rtx_<CODE>.
2171 * expr.c (emit_push_insn): Likewise.
2172 (expand_expr_real_2): Likewise.
2174 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2175 Alan Hayward <alan.hayward@arm.com>
2176 David Sherwood <david.sherwood@arm.com>
2178 * loop-unroll.c (split_iv): Call copy_rtx on the step.
2180 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2181 Alan Hayward <alan.hayward@arm.com>
2182 David Sherwood <david.sherwood@arm.com>
2184 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
2185 calling tree_to_uhwi.
2187 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2188 Alan Hayward <alan.hayward@arm.com>
2189 David Sherwood <david.sherwood@arm.com>
2191 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
2192 INTEGER_CST rather than a negative test for ADDR_EXPR.
2194 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2195 Alan Hayward <alan.hayward@arm.com>
2196 David Sherwood <david.sherwood@arm.com>
2198 * tree-vrp.c (extract_range_from_binary_expr_1): Check
2199 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
2201 2017-09-21 Richard Biener <rguenther@suse.de>
2203 PR tree-optimization/71351
2204 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
2205 graphite_create_new_loop_guard): Remove, fold remaining parts
2207 (translate_isl_ast_node_for): ... here and simplify.
2209 2017-09-21 Jakub Jelinek <jakub@redhat.com>
2212 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
2213 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
2214 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
2215 alternative always use QI mode, for -Os imov (=R,R) alternative
2216 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
2219 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2220 Jeff Law <law@redhat.com>
2222 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
2223 (allocate_stack_space): New function, partially extracted from
2225 (s390_emit_prologue): Track offset to most recent stack probe.
2226 Code to allocate space moved into allocate_stack_space.
2227 Dump actions when no stack is allocated.
2228 (s390_prologue_plus_offset): New function.
2229 (s390_emit_stack_probe): Likewise.
2231 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
2233 * common.opt (Wa, Wl, Wp, g, gz=): Add
2235 (gno-column-info): Remove.
2236 (gcolumn-info): Drop RejectNegative.
2238 (gno-record-gcc-switches): Remove.
2239 (grecord-gcc-switches): Drop RejectNegative.
2240 (gno-split-dwarf): Remove.
2241 (gsplit-dwarf): Drop RejectNegative.
2242 (gno-strict-dwarf): Remove.
2243 (gstrict-dwarf): Drop RejectNegative.
2244 * config/darwin.opt (gfull, gused): Add RejectNegative.
2245 * dwarf2out.c (gen_producer_string): Drop
2246 gno-record-gcc-switches handler.
2247 * optc-gen.awk: Add g to prefixes with negative forms.
2248 * opts-common.c (remapping_prefix_p): New.
2249 (find_opt): Check it.
2250 (generate_canonical_option): Test g prefix.
2251 (option_map): Add -gno- mapping.
2252 (add_misspelling_candidates): Check remapping_prefix_p.
2254 2017-09-20 Jeff Law <law@redhat.com>
2256 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
2257 thinko in stack clash protection support.
2259 * explow.c (compute_stack_clash_protection_loop_data): Use
2260 CONST_INT_P instead of explicit test. Verify object is a
2261 CONST_INT_P before looking at INTVAL.
2262 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
2263 instead of explicit test.
2265 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
2268 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
2269 address instead of to r1 and r11.
2271 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
2273 * config.gcc: Support "knm".
2274 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
2275 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2277 * config/i386/i386.c (m_KNM): Define.
2278 (processor_target_table): Add "knm".
2280 (ix86_option_override_internal): Add "knm".
2281 (ix86_issue_rate): Add PROCESSOR_KNM.
2282 (ix86_adjust_cost): Ditto.
2283 (ia32_multipass_dfa_lookahead): Ditto.
2284 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
2285 (fold_builtin_cpu): Add M_INTEL_KNM.
2286 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
2287 (processor_type): Add PROCESSOR_KNM.
2288 * config/i386/x86-tune.def: Add m_KNM.
2289 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
2291 2017-09-20 Richard Biener <rguenther@suse.de>
2293 PR tree-optimization/80213
2294 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
2295 are allowed in empty BBs as well.
2296 (canonicalize_loop_closed_ssa): Also look for other complex
2298 (scop_detection::get_sese): Include the loop-closed PHI block
2300 (scop_detection::merge_sese): Remove code adding extra blocks.
2301 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
2302 (build_scops): Assert the final returned scop is invalid.
2304 2017-09-20 Richard Biener <rguenther@suse.de>
2306 PR tree-optimization/82264
2307 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
2309 (vn_phi_lookup): Likewise.
2310 (vn_phi_insert): Likewise.
2312 2017-09-20 Jakub Jelinek <jakub@redhat.com>
2314 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
2315 that fits into uhwi or shwi, add DW_AT_const_value regardless
2316 of early_dwarf without going through RTL, using add_AT_unsigned
2319 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
2320 (DEBUG_LTO_ABBREV_SECTION): Likewise.
2321 (DEBUG_LTO_MACINFO_SECTION): Likewise.
2322 (DEBUG_MACRO_SECTION): Likewise.
2323 (DEBUG_LTO_MACRO_SECTION): Likewise.
2324 (DEBUG_STR_DWO_SECTION): Likewise.
2325 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
2326 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
2327 (DEBUG_LTO_DWO_LINE_SECTION): Define.
2328 (DEBUG_LTO_LINE_STR_SECTION): Define.
2329 (init_sections_and_labels): Initialize debug_line_str_section
2330 variable. Initialize debug_loc_section for -gdwarf-5 to
2331 DEBUG_LOCLISTS_SECTION. Formatting fixes.
2333 2017-09-20 Richard Biener <rguenther@suse.de>
2335 * graphite-sese-to-poly.c (extract_affine): Properly handle
2336 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
2338 2017-09-20 Richard Biener <rguenther@suse.de>
2340 PR tree-optimization/81373
2341 * graphite-scop-detection.c (build_cross_bb_scalars_def):
2342 Force SESE live-out defs to be handled even if they are
2345 2017-09-19 Jeff Law <law@redhat.com>
2347 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
2348 nothing for stack adjustments with REG_STACK_CHECK.
2349 * sched-deps.c (parse_add_or_inc): Reject insns with
2350 REG_STACK_CHECK from dependency breaking.
2351 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
2352 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
2353 * reg-notes.def (STACK_CHECK): New note.
2355 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
2356 (ix86_expand_prologue): Dump stack clash info as needed.
2357 Call ix86_adjust_stack_and_probe_stack_clash as needed.
2359 * function.c (dump_stack_clash_frame_info): New function.
2360 * function.h (dump_stack_clash_frame_info): Prototype.
2361 (enum stack_clash_probes): New enum.
2363 * config/alpha/alpha.c (alpha_expand_prologue): Also check
2364 flag_stack_clash_protection.
2365 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
2366 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
2367 (arm_frame_pointer_required): Likewise.
2368 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
2369 (ia64_expand_prologue): Likewise.
2370 * config/mips/mips.c (mips_expand_prologue): Likewise.
2371 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
2372 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
2373 (sparc_flat_expand_prologue): Likewise.
2374 * config/spu/spu.c (spu_expand_prologue): Likewise.
2376 * explow.c: Include "params.h".
2377 (anti_adjust_stack_and_probe_stack_clash): New function.
2378 (get_stack_check_protect): Likewise.
2379 (compute_stack_clash_protection_loop_data): Likewise.
2380 (emit_stack_clash_protection_loop_start): Likewise.
2381 (emit_stack_clash_protection_loop_end): Likewise.
2382 (allocate_dynamic_stack_space): Use get_stack_check_protect.
2383 Use anti_adjust_stack_and_probe_stack_clash.
2384 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
2385 (emit_stack_clash_protection_loop_start): Likewise.
2386 (emit_stack_clash_protection_loop_end): Likewise.
2387 * rtl.h (get_stack_check_protect): Prototype.
2388 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
2389 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
2390 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
2392 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
2394 * doc/tm.texi: Rebuilt.
2395 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
2396 get_stack_check_protect.
2397 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
2398 * config/arm/arm.c (arm_expand_prologue): Likewise.
2399 (arm_frame_pointer_required): Likewise.
2400 * config/i386/i386.c (ix86_expand_prologue): Likewise.
2401 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
2402 * config/mips/mips.c (mips_expand_prologue): Likewise.
2403 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
2404 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
2405 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
2406 (sparc_flat_expand_prologue): Likewise.
2408 * common.opt (-fstack-clash-protection): New option.
2409 * flag-types.h (enum stack_check_type): Note difference between
2410 -fstack-check= and -fstack-clash-protection.
2411 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
2412 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
2413 * toplev.c (process_options): Issue warnings/errors for cases
2414 not handled with -fstack-clash-protection.
2415 * doc/invoke.texi (-fstack-clash-protection): Document new option.
2416 (-fstack-check): Note additional problem with -fstack-check=generic.
2417 Note that -fstack-check is primarily for Ada and refer users
2418 to -fstack-clash-protection for stack-clash-protection.
2419 Document new params for stack clash protection.
2421 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
2423 * config/i386/i386.c (ix86_split_long_move): Do not handle
2424 address used for LEA in a special way.
2426 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
2428 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
2430 2017-09-19 Martin Sebor <msebor@redhat.com>
2433 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
2434 of incompatible types.
2436 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
2438 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
2439 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
2440 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2441 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
2443 2017-09-19 Richard Biener <rguenther@suse.de>
2445 PR tree-optimization/82244
2446 * tree-vrp.c (remove_range_assertions): Do not propagate
2447 a constant to abnormals but replace the assert with a copy.
2449 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
2451 PR rtl-optimization/57878
2452 PR rtl-optimization/68988
2453 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
2454 avoidance test involving non_reload_pseudos. Move frequency test
2455 below the general fragmentation avoidance test.
2457 2017-09-19 Richard Biener <rguenther@suse.de>
2459 PR tree-optimization/69728
2460 * graphite-sese-to-poly.c (schedule_error): New global.
2461 (add_loop_schedule): Handle empty domain by failing the
2463 (build_original_schedule): Handle schedule_error.
2465 2017-09-19 Richard Biener <rguenther@suse.de>
2467 * graphite-scop-detection.c (scop_detection::can_represent_loop):
2468 Do not iterate to sibling loops but only to siblings of inner
2471 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
2474 * config/m68k/m68k.md (moveq feeding equality comparison): Check
2475 that the registers are different.
2477 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
2479 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
2480 to processor_model and "amdfam17h" to arch_names_table.
2481 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
2483 2017-09-18 Jakub Jelinek <jakub@redhat.com>
2486 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
2488 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2489 Alan Hayward <alan.hayward@arm.com>
2490 David Sherwood <david.sherwood@arm.com>
2492 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
2494 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
2496 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
2497 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
2498 vect_schedule_slp_instance.
2499 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
2500 Update call to vect_slp_analyze_node_operations. Simplify return
2502 (vect_slp_analyze_bb_1): Update call accordingly.
2503 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
2504 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
2505 (vect_schedule_slp): Update call accordingly.
2507 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2508 Alan Hayward <alan.hayward@arm.com>
2509 David Sherwood <david.sherwood@arm.com>
2511 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
2512 with types that aren't in fact scalar.
2514 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2516 * tree-vect-slp.c (vect_record_max_nunits): New function,
2518 (vect_build_slp_tree_1): ...here.
2519 (vect_build_slp_tree_2): Call it for phis too.
2521 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2523 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
2524 to vect_get_vec_def_for_operand when getting the mask operand.
2526 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2527 Alan Hayward <alan.hayward@arm.com>
2528 David Sherwood <david.sherwood@arm.com>
2530 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
2533 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2534 Alan Hayward <alan.hayward@arm.com>
2535 David Sherwood <david.sherwood@arm.com>
2537 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
2538 calculation for vector booleans.
2540 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2541 Alan Hayward <alan.hayward@arm.com>
2542 David Sherwood <david.sherwood@arm.com>
2544 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
2546 (vect_transform_stmt): ...here.
2547 (vect_analyze_stmt): Use it instead of calling
2548 vectorizable_live_operation directly.
2550 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
2552 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
2553 non-SIMT targets in acc vector loops.
2555 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
2557 * configure.ac: Add arc and check if assembler supports gdwarf2.
2558 * configure: Regenerate.
2560 2017-09-18 Richard Biener <rguenther@suse.de>
2562 PR tree-optimization/82220
2563 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
2564 epilogue niters from the min_profitable_iters compute.
2566 2017-09-18 Jakub Jelinek <jakub@redhat.com>
2569 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
2570 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
2571 (ix86_init_pic_reg): Revert 2017-09-01 changes.
2573 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
2576 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
2577 switching to a new text section.
2579 2017-09-18 Richard Biener <rguenther@suse.de>
2581 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
2583 (build_alias_set): Reject aliases with no access function.
2585 2017-09-18 Richard Biener <rguenther@suse.de>
2587 PR tree-optimization/79622
2588 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
2590 (build_cross_bb_scalars_use): Likewise.
2592 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
2594 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
2596 2017-09-18 Alan Modra <amodra@gmail.com>
2599 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
2600 stack_pointer_rtx for count 0. Update comments. Break up
2601 large rtl expression.
2603 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
2605 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
2606 Increase to 20 bytes.
2607 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
2608 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
2609 or avx version of the stub.
2611 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
2614 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
2615 compute the minimum stack alignment. Also update preferred stack
2616 boundary for leaf functions.
2618 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
2620 PR tree-optimization/82228
2621 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
2624 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2626 * common/config/nds32/nds32-common.c
2627 (nds32_option_optimization_table): Refine formatting.
2628 (nds32_option_optimization_table): Use -fsched-pressure and
2629 -fomit-frame-pointer for specific optimization level.
2631 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
2633 * config/nds32/nds32.c: Refine formatting and comments.
2634 * config/nds32/nds32.h: Likewise.
2635 * config/nds32/nds32.md: Likewise.
2636 * config/nds32/nds32-cost.c: Likewise.
2637 * config/nds32/nds32-isr.c: Likewise.
2638 * config/nds32/nds32-md-auxiliary.c: Likewise.
2639 * config/nds32/nds32-multiple.md: Likewise.
2640 * config/nds32/nds32-predicates.c: Likewise.
2642 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
2643 Jakub Jelinek <jakub@redhat.com>
2645 Add support for -std=c++2a.
2646 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
2648 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
2650 2017-09-15 Steve Ellcey <sellcey@cavium.com>
2653 * doc/extend.texi (Common Function Attributes): Add
2654 references to ARM, AArch64, and S/390 specific attributes.
2655 (Function Specific Option Pragmas): Add AArch64 and S/390
2656 to list of back ends that support the target pragma.
2658 2017-09-15 Nathan Sidwell <nathan@acm.org>
2660 * doc/standards.texi: Fix C++17 description. Update URLs for
2663 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
2665 * common.opt (Wcast-align=strict): New warning option.
2666 * doc/invoke.texi: Document -Wcast-align=strict.
2668 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
2670 * cgraph.h (cgraph_thunk_info): Add comments.
2671 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
2672 assert for VIRTUAL_* arguments stricter.
2674 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
2676 PR tree-optimization/71026
2677 * match.pd: Move RDIV patterns from fold-const.c
2678 * fold-const.c (distribute_real_division): Removed.
2679 (fold_binary_loc): Remove calls to distribute_real_divison.
2681 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2683 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
2684 c++1z and gnu++1z as deprecated. Change other references to
2685 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
2686 Change -Wc++1z-compat to -Wc++17-compat.
2687 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
2688 * dwarf2out.c (highest_c_language): Handle C++17.
2689 (gen_compile_unit_die): Likewise.
2691 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2693 PR rtl-optimization/82192
2694 * combine.c (make_extraction): Don't look through non-paradoxical
2695 SUBREGs or TRUNCATE if pos + len is or might be bigger than
2698 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2699 Alan Hayward <alan.hayward@arm.com>
2700 David Sherwood <david.sherwood@arm.com>
2702 * target.def (function_arg_offset): New hook.
2703 * targhooks.h (default_function_arg_offset): Declare.
2704 * targhooks.c (default_function_arg_offset): New function.
2705 * function.c (locate_and_pad_parm): Use
2706 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
2707 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
2708 (TARGET_FUNCTION_ARG_OFFSET): ...this.
2709 * doc/tm.texi: Regenerate.
2710 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
2711 * config/spu/spu.c (spu_function_arg_offset): New function.
2712 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
2713 * system.h (FUNCTION_ARG_OFFSET): Poison.
2715 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2716 Alan Hayard <alan.hayward@arm.com>
2717 David Sherwood <david.sherwood@arm.com>
2719 * target.def (truly_noop_truncation): New hook.
2720 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
2721 than TRULY_NOOP_TRUNCATION.
2722 * hooks.h (hook_bool_uint_uint_true): Declare.
2723 * hooks.c (hook_bool_uint_uint_true): New function.
2724 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
2725 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
2726 * doc/tm.texi: Regenerate.
2727 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
2728 rather than TRULY_NOOP_TRUNCATION in comments.
2729 (simplify_comparison): Likewise.
2730 (record_truncated_value): Likewise.
2731 * expmed.c (extract_bit_field_1): Likewise.
2732 (extract_split_bit_field): Likewise.
2733 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
2734 instead of TRULY_NOOP_TRUNCATION.
2735 * function.c (assign_parm_setup_block): Likewise.
2736 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
2737 * rtlhooks.c: Include target.h.
2738 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
2739 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
2740 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
2741 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
2742 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
2743 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
2744 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
2745 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
2746 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
2747 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
2748 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
2749 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
2750 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
2751 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
2752 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
2753 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
2754 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
2755 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
2756 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
2757 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
2758 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
2759 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
2760 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
2761 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
2762 * config/mips/mips.c (mips_truly_noop_truncation): New function.
2763 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2764 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2765 rather than TRULY_NOOP_TRUNCATION in comments.
2766 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
2767 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
2768 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
2769 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
2770 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
2771 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
2772 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
2773 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
2774 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
2775 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
2776 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
2777 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2778 rather than TRULY_NOOP_TRUNCATION in comments.
2779 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
2780 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
2781 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
2782 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
2783 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
2784 TRULY_NOOP_TRUNCATION condition.
2785 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
2786 (TRULY_NOOP_TRUNCATION): Delete.
2787 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
2788 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
2789 * config/spu/spu.c (spu_truly_noop_truncation): New function.
2790 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2791 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
2792 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
2793 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
2794 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
2795 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
2796 rather than TRULY_NOOP_TRUNCATION in comments.
2797 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
2798 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
2799 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
2800 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
2801 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
2802 * system.h (TRULY_NOOP_TRUNCATION): Poison.
2804 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
2807 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
2808 (*cmp_ior): Likewise.
2809 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
2810 (*ior_scc_scc_cmp): Likewise.
2811 (*and_scc_scc): Likewise.
2812 (*and_scc_scc_cmp): Likewise.
2814 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
2815 Alan Hayard <alan.hayward@arm.com>
2816 David Sherwood <david.sherwood@arm.com>
2818 * target.def (can_change_mode_class): New hook.
2819 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
2820 (hard_regno_nregs): Likewise.
2821 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
2822 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
2823 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
2824 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
2825 (LOAD_EXTEND_OP): Update accordingly.
2826 * doc/tm.texi: Regenerate.
2827 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
2828 CANNOT_CHANGE_MODE_CLASS.
2829 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
2830 (REG_CAN_CHANGE_MODE_P): ...this new macro.
2831 * combine.c (simplify_set): Update accordingly.
2832 * emit-rtl.c (validate_subreg): Likewise.
2833 * recog.c (general_operand): Likewise.
2834 * regcprop.c (mode_change_ok): Likewise.
2835 * reload1.c (choose_reload_regs): Likewise.
2836 (inherit_piecemeal_p): Likewise.
2837 * rtlanal.c (simplify_subreg_regno): Likewise.
2838 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
2839 instead of CANNOT_CHANGE_MODE_CLASS.
2840 (reload_cse_simplify_operands): Likewise.
2841 * reload.c (push_reload): Use targetm.can_change_mode_class
2842 instead of CANNOT_CHANGE_MODE_CLASS.
2843 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
2844 REG_CANNOT_CHANGE_MODE_P.
2845 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2846 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
2847 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2848 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2849 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2850 (arm_can_change_mode_class): New function.
2851 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2852 than CANNOT_CHANGE_MODE_CLASS in comments.
2853 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2854 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
2855 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
2856 (ix86_can_change_mode_class): ...this new function, inverting the
2857 sense of the return value.
2858 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2859 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2860 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2861 (ia64_can_change_mode_class): New function.
2862 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2863 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
2864 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
2865 (m32c_can_change_mode_class): ...this new function, inverting the
2866 sense of the return value.
2867 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2868 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2869 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
2870 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
2871 (mips_can_change_mode_class): ...this new function, inverting the
2872 sense of the return value.
2873 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2874 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2875 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2876 (msp430_can_change_mode_class): New function.
2877 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2878 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
2879 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2880 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2881 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2882 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
2883 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2884 (pa_cannot_change_mode_class): Replace with...
2885 (pa_can_change_mode_class): ...this new function, inverting the
2886 sense of the return value.
2887 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
2888 than CANNOT_CHANGE_MODE_CLASS in comments.
2889 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2890 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
2891 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2892 (pdp11_cannot_change_mode_class): Replace with...
2893 (pdp11_can_change_mode_class): ...this new function, inverting the
2894 sense of the return value.
2895 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2896 * config/powerpcspe/powerpcspe-protos.h
2897 (rs6000_cannot_change_mode_class_ptr): Delete.
2898 * config/powerpcspe/powerpcspe.c
2899 (rs6000_cannot_change_mode_class_ptr): Delete.
2900 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2901 (rs6000_option_override_internal): Assign to
2902 targetm.can_change_mode_class instead of
2903 rs6000_cannot_change_mode_class_ptr.
2904 (rs6000_cannot_change_mode_class): Replace with...
2905 (rs6000_can_change_mode_class): ...this new function, inverting the
2906 sense of the return value.
2907 (rs6000_debug_cannot_change_mode_class): Replace with...
2908 (rs6000_debug_can_change_mode_class): ...this new function.
2909 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2910 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
2911 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2912 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2913 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
2915 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
2916 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2917 (rs6000_option_override_internal): Assign to
2918 targetm.can_change_mode_class instead of
2919 rs6000_cannot_change_mode_class_ptr.
2920 (rs6000_cannot_change_mode_class): Replace with...
2921 (rs6000_can_change_mode_class): ...this new function, inverting the
2922 sense of the return value.
2923 (rs6000_debug_cannot_change_mode_class): Replace with...
2924 (rs6000_debug_can_change_mode_class): ...this new function.
2925 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2926 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
2927 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
2928 (s390_can_change_mode_class): ...this new function, inverting the
2929 sense of the return value.
2930 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2931 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2932 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
2933 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2934 (sh_cannot_change_mode_class): Replace with...
2935 (sh_can_change_mode_class): ...this new function, inverting the
2936 sense of the return value.
2937 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2938 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2939 (sparc_can_change_mode_class): New function.
2940 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2941 * config/spu/spu.c (spu_can_change_mode_class): New function.
2942 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2943 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
2944 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
2945 (visium_can_change_mode_class): New function.
2946 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
2948 2017-09-15 Richard Biener <rguenther@suse.de>
2950 PR tree-optimization/82217
2951 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
2952 but not undefined case.
2954 2017-09-15 Jakub Jelinek <jakub@redhat.com>
2957 * postreload.c (reload_cse_simplify_operands): Skip
2958 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
2960 2017-09-15 Richard Biener <rguenther@suse.de>
2962 PR tree-optimization/68823
2963 * graphite-scop-detection.c (build_alias_set): If we have a
2964 possible dependence check whether we can handle them by just
2965 looking at the DRs DR_ACCESS_FNs.
2966 (build_scops): If build_alias_set fails, fail the SCOP.
2968 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2970 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
2971 to support float128 built-in functions that require the ISA 3.0
2973 (BU_FLOAT128_3_HW): Likewise.
2974 (SQRTF128): Add support for the IEEE 128-bit square root and fma
2976 (FMAF128): Likewise.
2978 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
2979 support for built-in functions that need the ISA 3.0 IEEE 128-bit
2980 floating point instructions.
2981 (rs6000_invalid_builtin): Likewise.
2982 (rs6000_builtin_mask_names): Likewise.
2983 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
2984 (RS6000_BTM_FLOAT128_HW): Likewise.
2985 (RS6000_BTM_COMMON): Likewise.
2986 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
2988 * doc/extend.texi (RS/6000 built-in functions): Document the
2989 IEEE 128-bit floating point square root and fused multiply-add
2992 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
2994 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
2995 reg (r2) isn't in the set of registers defined in the prologue.
2997 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
2998 Alan Hayward <alan.hayward@arm.com>
2999 David Sherwood <david.sherwood@arm.com>
3001 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
3002 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
3003 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
3004 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
3005 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
3007 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
3008 max_vectorization_factor.
3009 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
3011 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3012 Alan Hayward <alan.hayward@arm.com>
3013 David Sherwood <david.sherwood@arm.com>
3015 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
3016 (vect_worthwhile_without_simd_p): Declare.
3017 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
3018 (vectorizable_reduction): Use it.
3019 * tree-vect-stmts.c (vectorizable_shift): Likewise.
3020 (vectorizable_operation): Likewise.
3022 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3023 Alan Hayward <alan.hayward@arm.com>
3024 David Sherwood <david.sherwood@arm.com>
3026 * tree-vectorizer.h (vect_get_num_copies): New function.
3027 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
3028 * tree-vect-loop.c (vectorizable_reduction): Likewise.
3029 (vectorizable_induction): Likewise.
3030 (vectorizable_live_operation): Likewise.
3031 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3032 (vectorizable_bswap): Likewise.
3033 (vectorizable_call): Likewise.
3034 (vectorizable_conversion): Likewise.
3035 (vectorizable_assignment): Likewise.
3036 (vectorizable_shift): Likewise.
3037 (vectorizable_operation): Likewise.
3038 (vectorizable_store): Likewise.
3039 (vectorizable_load): Likewise.
3040 (vectorizable_condition): Likewise.
3041 (vectorizable_comparison): Likewise.
3042 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
3044 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3045 Alan Hayward <alan.hayward@arm.com>
3046 David Sherwood <david.sherwood@arm.com>
3048 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
3049 of vect_init_vector.
3051 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3052 Alan Hayward <alan.hayward@arm.com>
3053 David Sherwood <david.sherwood@arm.com>
3055 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
3056 an inline wrapper that provides a location.
3057 (gimple_build_vector): Likewise.
3058 * gimple-fold.c (gimple_build_vector_from_val): New function.
3059 (gimple_build_vector): Likewise.
3060 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
3061 functions to build the initial value. Always return a gimple value.
3062 (get_initial_defs_for_reduction): Likewise. Only compute
3064 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
3065 vect_init_vector on the results from get_initial_def(s)_for_reduction.
3066 (vectorizable_induction): Use gimple_build_vector rather than
3069 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3070 Alan Hayward <alan.hayward@arm.com>
3071 David Sherwood <david.sherwood@arm.com>
3073 * target.h (vec_perm_indices): New typedef.
3074 (auto_vec_perm_indices): Likewise.
3075 * optabs-query.h: Include target.h
3076 (can_vec_perm_p): Take a vec_perm_indices *.
3077 * optabs-query.c (can_vec_perm_p): Likewise.
3078 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
3079 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
3080 * tree-vect-generic.c (lower_vec_perm): Likewise.
3081 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
3082 (vect_grouped_load_supported): Likewise.
3083 (vect_shift_permute_load_chain): Likewise.
3084 (vect_permute_store_chain): Use auto_vec_perm_indices.
3085 (vect_permute_load_chain): Likewise.
3086 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
3087 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
3088 Update uses of can_vec_perm_p.
3089 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
3090 mode with a number of elements. Take a vec_perm_indices *.
3091 (vect_create_epilog_for_reduction): Update accordingly.
3092 Use auto_vec_perm_indices.
3093 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
3094 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
3095 (vect_transform_slp_perm_load): Likewise.
3096 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
3097 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
3098 (vect_gen_perm_mask_checked): Likewise.
3099 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
3100 (vect_gen_perm_mask_checked): Likewise.
3101 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
3102 (vectorizable_store): Likewise.
3103 (vectorizable_load): Likewise.
3104 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
3105 (vectorizable_bswap): Likewise.
3107 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3108 Alan Hayward <alan.hayward@arm.com>
3109 David Sherwood <david.sherwood@arm.com>
3111 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
3112 * tree.c (build_vector): Likewise.
3113 (build_vector_from_ctor): Update accordingly.
3114 (build_vector_from_val): Likewise.
3115 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
3116 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
3117 * tree-vect-generic.c (add_rshift): Likewise.
3118 (expand_vector_divmod): Likewise.
3119 (optimize_vector_constructor): Likewise.
3120 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
3121 (vect_transform_slp_perm_load): Likewise.
3122 (vect_schedule_slp_instance): Likewise.
3123 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
3124 (vectorizable_call): Likewise.
3125 (vect_gen_perm_mask_any): Likewise. Add elements in order.
3126 * expmed.c (make_tree): Likewise.
3127 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
3128 a vector passed to build_vector.
3129 (fold_convert_const): Likewise.
3130 (exact_inverse): Likewise.
3131 (fold_ternary_loc): Likewise.
3132 (fold_relational_const): Likewise.
3133 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
3134 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
3135 (const_unop): Likewise. Store the reduction accumulator in a
3136 variable rather than an array.
3137 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
3138 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
3139 the new vector, rather than constructing it after the input arrays.
3140 (native_interpret_vector): Use auto_vec<tree> when building
3141 a vector passed to build_vector. Add elements in order.
3142 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
3143 auto_vec<tree> when building a vector passed to build_vector.
3144 (vect_create_epilog_for_reduction): Likewise.
3145 (vectorizable_induction): Likewise.
3146 (get_initial_def_for_reduction): Likewise. Fix indentation of
3148 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
3150 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
3151 passed to build_vector.
3153 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3154 Alan Hayward <alan.hayward@arm.com>
3155 David Sherwood <david.sherwood@arm.com>
3157 * tree-core.h (tree_base::u): Add an "nelts" field.
3158 (tree_vector): Use VECTOR_CST_NELTS as the length.
3159 * tree.c (tree_size): Likewise.
3160 (make_vector): Initialize VECTOR_CST_NELTS.
3161 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
3162 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
3163 TYPE_VECTOR_SUBPARTS.
3164 * expr.c (const_vector_mask_from_tree): Consistently use "units"
3165 as the number of units, setting it from VECTOR_CST_NELTS.
3166 (const_vector_from_tree): Likewise.
3167 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
3168 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
3169 (fold_negate_expr_1): Likewise.
3170 (fold_convert_const): Likewise.
3171 (const_binop): Likewise. Differentiate the number of output and
3173 (const_unop): Likewise.
3174 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
3175 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
3176 in cases that did the opposite.
3178 2017-09-14 Richard Biener <rguenther@suse.de>
3180 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
3183 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
3185 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
3187 2017-09-14 Jakub Jelinek <jakub@redhat.com>
3190 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
3191 if and where to split a bb, except for splitting before debug insn
3192 sequences followed by non-label real insn. Delete debug insns
3193 in between basic blocks.
3195 * combine.c (make_compound_operation_int): Formatting fixes.
3197 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
3198 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3199 * config/netbsd.h (LINK_EH_SPEC): Likewise.
3200 * config/sol2.h (LINK_EH_SPEC): Likewise.
3201 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3202 * config/s390/linux.h (LINK_SPEC): Likewise.
3203 * config/freebsd.h (LINK_EH_SPEC): Likewise.
3204 * config/openbsd.h (LINK_EH_SPEC): Likewise.
3205 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3206 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
3207 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
3208 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3209 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
3210 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
3212 2017-09-13 Jakub Jelinek <jakub@redhat.com>
3214 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
3216 (ENDFILE_LINUX_SPEC): Likewise.
3217 (LINK_EH_SPEC): Likewise.
3218 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
3219 (LINK_OS_LINUX_SPEC32): Likewise.
3220 (LINK_OS_LINUX_SPEC64): Likewise.
3221 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
3222 (LINK_OS_LINUX_SPEC): Likewise.
3224 2017-09-13 Martin Liska <mliska@suse.cz>
3227 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
3228 CASE_HIGH is NULL_TREE.
3230 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
3231 Alan Hayward <alan.hayward@arm.com>
3232 David Sherwood <david.sherwood@arm.com>
3234 * target.def (secondary_memory_needed): New hook.
3235 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
3236 instead of SECONDARY_MEMORY_NEEDED.
3237 (secondary_memory_needed_mode): Likewise.
3238 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
3239 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
3240 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
3241 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
3242 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
3243 * doc/tm.texi: Regenerate.
3244 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
3245 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
3246 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3247 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
3248 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
3249 * config/i386/i386.c (inline_secondary_memory_needed): Put the
3250 mode argument first and change the reg_class arguments to reg_class_t.
3251 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
3252 Make static. Update the call to inline_secondary_memory_needed.
3253 (ix86_register_move_cost): Update the call to
3254 inline_secondary_memory_needed.
3255 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3256 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
3258 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
3259 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3261 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
3262 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
3263 * config/mips/mips.c (mips_secondary_memory_needed): Make static
3264 and match hook interface. Add comment from mips.h.
3265 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3266 * config/mmix/mmix.md (truncdfsf2): Refer to
3267 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3269 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
3270 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
3271 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3272 (pa_secondary_memory_needed): New function.
3273 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
3274 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
3275 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3276 (pdp11_secondary_memory_needed): Make static and match hook interface.
3277 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
3278 * config/powerpcspe/powerpcspe-protos.h
3279 (rs6000_secondary_memory_needed_ptr): Delete.
3280 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
3282 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3283 (rs6000_option_override_internal): Assign to
3284 targetm.secondary_memory_needed rather than
3285 rs6000_secondary_memory_needed_ptr.
3286 (rs6000_secondary_memory_needed): Match hook interface.
3287 (rs6000_debug_secondary_memory_needed): Likewise.
3288 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
3289 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
3290 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
3291 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3292 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
3293 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
3295 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
3296 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3297 (rs6000_option_override_internal): Assign to
3298 targetm.secondary_memory_needed rather than
3299 rs6000_secondary_memory_needed_ptr.
3300 (rs6000_secondary_memory_needed): Match hook interface.
3301 (rs6000_debug_secondary_memory_needed): Likewise.
3302 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
3303 * config/s390/s390.c (s390_secondary_memory_needed): New function.
3304 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3305 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
3306 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3307 (sparc_secondary_memory_needed): New function.
3308 * lra-constraints.c (check_and_process_move): Refer to
3309 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3311 (curr_insn_transform): Likewise.
3312 (process_alt_operands): Use targetm.secondary_memory_needed
3313 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3314 (check_secondary_memory_needed_p): Likewise.
3315 (choose_split_class): Likewise.
3316 * reload.c: Unconditionally include code that was previously
3317 conditional on SECONDARY_MEMORY_NEEDED.
3318 (push_secondary_reload): Use targetm.secondary_memory_needed
3319 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3320 (push_reload): Likewise.
3321 * reload1.c: Unconditionally include code that was previously
3322 conditional on SECONDARY_MEMORY_NEEDED.
3323 (choose_reload_regs): Use targetm.secondary_memory_needed
3324 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3325 (gen_reload): Likewise.
3326 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
3328 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
3329 Alan Hayward <alan.hayward@arm.com>
3330 David Sherwood <david.sherwood@arm.com>
3332 * target.def (secondary_memory_needed_mode): New hook:
3333 * targhooks.c (default_secondary_memory_needed_mode): Declare.
3334 * targhooks.h (default_secondary_memory_needed_mode): New function.
3335 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
3336 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
3337 * doc/tm.texi: Regenerate.
3338 * lra-constraints.c (check_and_process_move): Use
3339 targetm.secondary_memory_needed_mode instead of
3340 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
3341 (curr_insn_transform): Likewise.
3342 * reload.c (get_secondary_mem): Likewise.
3343 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3344 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
3346 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3347 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3348 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
3349 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3350 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
3352 * config/powerpcspe/powerpcspe-protos.h
3353 (rs6000_secondary_memory_needed_mode): Delete.
3354 * config/powerpcspe/powerpcspe.c
3355 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3356 (rs6000_secondary_memory_needed_mode): Make static.
3357 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3358 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
3360 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
3362 (rs6000_secondary_memory_needed_mode): Make static.
3363 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3364 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
3365 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3366 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3367 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
3369 (sparc_secondary_memory_needed_mode): New function.
3370 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
3372 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
3374 * config/aarch64/constraints.md (Umq): New constraint.
3375 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
3377 (mov<mode>): Update condition.
3379 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3381 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
3382 when bitposition is the same.
3384 2017-09-13 Richard Biener <rguenther@suse.de>
3386 * dwarf2out.c (output_die_symbol): Remove.
3387 (output_die): Do not output a DIEs symbol.
3389 2017-09-13 Richard Biener <rguenther@suse.de>
3392 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
3393 default-def to avoid breaking iterator update with the weird
3394 interaction with cgraph_update_edges_for_call_stmt_node.
3396 2017-09-13 Richard Biener <rguenther@suse.de>
3398 * tree-cfg.c (verify_gimple_assign_binary): Add verification
3399 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
3400 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
3401 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
3403 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
3405 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
3406 Disable pc relative literal load irrespective of
3407 TARGET_FIX_ERR_A53_84341 for default.
3409 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
3411 * config/sparc/sparc.c (output_return): Output the source location of
3412 the insn in the delay slot, if any.
3413 (output_sibcall): Likewise.
3415 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
3418 * common.opt (-static-pie): New alias.
3419 (shared): Negate static-pie.
3420 (-no-pie): Update help text.
3422 (static-pie): New option.
3423 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
3424 -static-pie support.
3425 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
3426 (LINK_EH_SPEC): Likewise.
3427 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3428 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
3429 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
3430 * gcc.c (LINK_COMMAND_SPEC): Likewise.
3431 (init_gcc_specs): Likewise.
3432 (init_spec): Likewise.
3433 (display_help): Update help message for -pie.
3434 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
3437 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
3439 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
3440 (movdi_aarch64): Likewise.
3441 (movti_aarch64): Likewise.
3443 2017-09-12 Simon Wright <simon@pushface.org>
3446 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
3447 calculation of the minor version, always output as 0.
3449 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3452 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
3453 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
3454 on it early, rather than manual conversion late. For
3455 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
3456 instead of performing manual conversion.
3458 2017-09-12 Carl Love <cel@us.ibm.com>
3460 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
3461 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
3463 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
3464 VMULOSW): Add definitions.
3465 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
3466 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
3467 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
3468 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
3469 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
3471 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3473 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
3475 (movti_aarch64): Likewise.
3476 (movdf_aarch64): Likewise.
3477 (movtf_aarch64): Likewise.
3478 (load_pairdi): Likewise.
3479 (store_pairdi): Likewise.
3480 (load_pairdf): Likewise.
3481 (store_pairdf): Likewise.
3482 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
3483 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
3484 (ldr_got_small_<mode>): Likewise.
3485 (ldr_got_small_28k_<mode>): Likewise.
3486 (ldr_got_tiny): Likewise.
3487 * config/aarch64/iterators.md (ldst_sz): New.
3488 (ldpstp_sz): Likewise.
3489 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
3491 (thunderx_load): Split load_8 to load_16.
3492 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
3494 (thunderx2t99_storepair_basic): Split store_8 to store_16.
3495 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
3496 (xgene1_store_pair): Split store_8 to store_16.
3497 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
3498 (falkor_st_0_st_sd): Split store_8 to store_16.
3500 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3502 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
3503 and store1/2/3/4 to store_4/8/12/16.
3504 * config/aarch64/aarch64.md: Update for rename.
3505 * config/arm/arm.md: Likewise.: Likewise.
3506 * config/arm/arm.c: Likewise.
3507 * config/arm/thumb1.md: Likewise.
3508 * config/arm/thumb2.md: Likewise.
3509 * config/arm/vfp.md: Likewise.
3510 * config/arm/arm-generic.md: Likewise.
3511 * config/arm/arm1020e.md: Likewise.
3512 * config/arm/arm1026ejs.md: Likewise.
3513 * config/arm/arm1136jfs.md: Likewise.
3514 * config/arm/arm926ejs.md: Likewise.
3515 * config/arm/cortex-a15.md: Likewise.
3516 * config/arm/cortex-a17.md: Likewise.
3517 * config/arm/cortex-a5.md: Likewise.
3518 * config/arm/cortex-a53.md: Likewise.
3519 * config/arm/cortex-a57.md: Likewise.
3520 * config/arm/cortex-a7.md: Likewise.
3521 * config/arm/cortex-a8.md: Likewise.
3522 * config/arm/cortex-a9.md: Likewise.
3523 * config/arm/cortex-m4.md: Likewise.
3524 * config/arm/cortex-m7.md: Likewise.
3525 * config/arm/cortex-r4.md: Likewise.
3526 * config/arm/exynos-m1.md: Likewise.
3527 * config/arm/fa526.md: Likewise.
3528 * config/arm/fa606te.md: Likewise.
3529 * config/arm/fa626te.md: Likewise.
3530 * config/arm/fa726te.md: Likewise.
3531 * config/arm/fmp626.md: Likewise.
3532 * config/arm/iwmmxt.md: Likewise.
3533 * config/arm/ldmstm.md: Likewise.
3534 * config/arm/marvell-pj4.md: Likewise.
3535 * config/arm/xgene1.md: Likewise.
3536 * config/aarch64/thunderx.md: Likewise.
3537 * config/aarch64/thunderx2t99.md: Likewise.
3538 * config/aarch64/falkor.md: Likewise.
3540 2017-09-12 Martin Liska <mliska@suse.cz>
3542 * attribs.c (private_lookup_attribute): New function.
3543 * attribs.h (private_lookup_attribute): Declared here.
3544 (lookup_attribute): Called from this place.
3546 2017-09-12 Richard Biener <rguenther@suse.de>
3548 PR tree-optimization/82157
3549 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
3550 stmts with side-effects.
3552 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3553 Alan Hayward <alan.hayward@arm.com>
3554 David Sherwood <david.sherwood@arm.com>
3556 * target.def (hard_regno_nregs): New hook.
3557 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
3558 * targhooks.h (default_hard_regno_nregs): Declare.
3559 * targhooks.c (default_hard_regno_nregs): New function.
3560 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
3561 (TARGET_HARD_REGNO_NREGS): ...this hook.
3562 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
3563 (CLASS_MAX_NREGS): Likewise.
3564 * doc/tm.texi: Regenerate.
3565 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
3566 instead of HARD_REGNO_NREGS.
3567 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
3568 HARD_REGNO_NREGS in the comment.
3569 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
3570 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
3571 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
3572 Return an unsigned int.
3573 (TARGET_HARD_REGNO_NREGS): Redefine.
3574 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
3575 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
3576 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
3577 (arc_hard_regno_nregs): New function.
3578 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
3579 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
3580 (arm_hard_regno_nregs): New function.
3581 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
3582 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
3583 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
3584 (TARGET_HARD_REGNO_NREGS): Redefine.
3585 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
3586 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
3587 (HARD_REGNO_NREGS): Delete.
3588 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
3589 (cr16_hard_regno_nregs): New function.
3590 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
3591 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
3592 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
3593 (cris_hard_regno_nregs): New function.
3594 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
3595 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
3596 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3597 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
3598 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
3599 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
3600 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
3601 (frv_hard_regno_nregs): Make static. Take and return an
3603 (frv_class_max_nregs): Remove outdated copy of documentation.
3604 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
3605 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
3606 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
3607 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
3608 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
3609 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
3610 (TARGET_HARD_REGNO_NREGS): Redefine.
3611 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
3612 (CLASS_MAX_NREGS): Update comment.
3613 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
3614 (ia64_hard_regno_nregs): New function.
3615 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
3616 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
3617 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
3618 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
3619 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
3621 (m32c_hard_regno_nregs): Likewise. Make static.
3622 (TARGET_HARD_REGNO_NREGS): Redefine.
3623 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
3624 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
3625 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
3626 (m68k_hard_regno_nregs): New function.
3627 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
3628 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
3629 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
3630 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
3631 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
3632 Take and return an unsigned int.
3633 (TARGET_HARD_REGNO_NREGS): Redefine.
3634 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
3635 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
3636 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
3637 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
3638 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
3639 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
3640 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
3641 (msp430_hard_regno_nregs): Make static. Take and return an
3643 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
3644 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
3645 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
3646 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
3647 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
3648 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
3649 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
3650 (TARGET_HARD_REGNO_NREGS): Redefine.
3651 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
3652 (PA_HARD_REGNO_NREGS): ...this.
3653 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
3654 (PA_HARD_REGNO_NREGS): ...this.
3655 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3656 (pa_hard_regno_nregs): New function.
3657 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
3658 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
3659 (pdp11_hard_regno_nregs): New function.
3660 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
3661 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
3662 (rs6000_hard_regno_nregs_hook): New function.
3663 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
3664 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
3665 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
3666 Take and return an unsigned int. Move earlier in file.
3667 (TARGET_HARD_REGNO_NREGS): Redefine.
3668 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
3669 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
3670 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
3671 (rl78_hard_regno_nregs): Make static. Take and return an
3673 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
3674 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
3675 (rs6000_hard_regno_nregs_hook): New function.
3676 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
3677 * config/rx/rx.c (rx_hard_regno_nregs): New function.
3678 (TARGET_HARD_REGNO_NREGS): Redefine.
3679 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
3680 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
3681 instead of HARD_REGNO_NREGS.
3682 (s390_hard_regno_nregs): New function.
3683 (s390_hard_regno_mode_ok): Add comment from s390.h.
3684 (TARGET_HARD_REGNO_NREGS): Redefine.
3685 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
3686 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
3687 (sh_hard_regno_nregs): New function.
3688 (sh_pass_in_reg_p): Use it.
3689 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
3690 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
3691 (sparc_hard_regno_nregs): New function.
3692 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
3693 * config/spu/spu.c (spu_hard_regno_nregs): New function.
3694 (spu_function_arg_advance): Use it, supplying a valid register number.
3695 (TARGET_HARD_REGNO_NREGS): Redefine.
3696 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
3697 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
3698 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
3699 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
3700 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
3701 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
3702 (CLASS_MAX_NREGS): Remove copy of old documentation.
3703 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
3704 (visium_hard_regno_nregs): New function.
3705 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
3706 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
3707 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
3708 xtensa_hard_regno_nregs): New function.
3709 * system.h (HARD_REGNO_NREGS): Poison.
3711 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3713 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
3714 hard_regno_nregs instead of HARD_REGNO_NREGS.
3715 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
3716 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
3717 (c6x_expand_epilogue): Likewise.
3718 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
3719 (frv_read_iacc_argument): Likewise.
3720 * config/sh/sh.c: Include regs.h.
3721 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3722 (regs_used): Likewise.
3723 (output_stack_adjust): Likewise.
3724 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
3725 * expmed.c: Include regs.h.
3726 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
3727 * ree.c: Include regs.h.
3728 (combine_reaching_defs): Use hard_regno_nregs instead of
3730 (add_removable_extension): Likewise.
3732 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3734 * regs.h (hard_regno_nregs): Turn into a function.
3735 (end_hard_regno): Update accordingly.
3736 * caller-save.c (setup_save_areas): Likewise.
3737 (save_call_clobbered_regs): Likewise.
3738 (replace_reg_with_saved_mem): Likewise.
3739 (insert_restore): Likewise.
3740 (insert_save): Likewise.
3741 * combine.c (can_change_dest_mode): Likewise.
3742 (move_deaths): Likewise.
3743 (distribute_notes): Likewise.
3744 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
3745 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
3746 (rs6000_split_multireg_move): Likewise.
3747 (rs6000_register_move_cost): Likewise.
3748 (rs6000_memory_move_cost): Likewise.
3749 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
3750 (rs6000_split_multireg_move): Likewise.
3751 (rs6000_register_move_cost): Likewise.
3752 (rs6000_memory_move_cost): Likewise.
3753 * cselib.c (cselib_reset_table): Likewise.
3754 (cselib_lookup_1): Likewise.
3755 * emit-rtl.c (set_mode_and_regno): Likewise.
3756 * function.c (aggregate_value_p): Likewise.
3757 * ira-color.c (setup_profitable_hard_regs): Likewise.
3758 (check_hard_reg_p): Likewise.
3759 (calculate_saved_nregs): Likewise.
3760 (assign_hard_reg): Likewise.
3761 (improve_allocation): Likewise.
3762 (calculate_spill_cost): Likewise.
3763 * ira-emit.c (modify_move_list): Likewise.
3764 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
3765 (ira_hard_reg_in_set_p): Likewise.
3766 * ira.c (setup_reg_mode_hard_regset): Likewise.
3767 (clarify_prohibited_class_mode_regs): Likewise.
3768 (check_allocation): Likewise.
3769 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3770 (lra_setup_reg_renumber): Likewise.
3771 (setup_try_hard_regno_pseudos): Likewise.
3772 (spill_for): Likewise.
3773 (assign_hard_regno): Likewise.
3774 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
3775 * lra-constraints.c (in_class_p): Likewise.
3776 (lra_constraint_offset): Likewise.
3777 (simplify_operand_subreg): Likewise.
3778 (lra_constraints): Likewise.
3779 (split_reg): Likewise.
3780 (split_if_necessary): Likewise.
3781 (invariant_p): Likewise.
3782 (inherit_in_ebb): Likewise.
3783 * lra-lives.c (process_bb_lives): Likewise.
3784 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
3785 (get_hard_regs): Likewise.
3786 (do_remat): Likewise.
3787 * lra-spills.c (assign_spill_hard_regs): Likewise.
3788 * mode-switching.c (create_pre_exit): Likewise.
3789 * postreload.c (reload_combine_recognize_pattern): Likewise.
3790 * recog.c (peep2_find_free_register): Likewise.
3791 * regcprop.c (kill_value_regno): Likewise.
3792 (set_value_regno): Likewise.
3793 (copy_value): Likewise.
3794 (maybe_mode_change): Likewise.
3795 (find_oldest_value_reg): Likewise.
3796 (copyprop_hardreg_forward_1): Likewise.
3797 * regrename.c (check_new_reg_p): Likewise.
3798 (regrename_do_replace): Likewise.
3799 * reload.c (push_reload): Likewise.
3800 (combine_reloads): Likewise.
3801 (find_dummy_reload): Likewise.
3802 (operands_match_p): Likewise.
3803 (find_reloads): Likewise.
3804 (find_equiv_reg): Likewise.
3805 (reload_adjust_reg_for_mode): Likewise.
3806 * reload1.c (count_pseudo): Likewise.
3807 (count_spilled_pseudo): Likewise.
3808 (find_reg): Likewise.
3809 (clear_reload_reg_in_use): Likewise.
3810 (free_for_value_p): Likewise.
3811 (allocate_reload_reg): Likewise.
3812 (choose_reload_regs): Likewise.
3813 (reload_adjust_reg_for_temp): Likewise.
3814 (emit_reload_insns): Likewise.
3815 (delete_output_reload): Likewise.
3816 * rtlanal.c (subreg_get_info): Likewise.
3817 * sched-deps.c (sched_analyze_reg): Likewise.
3818 * sel-sched.c (init_regs_for_mode): Likewise.
3819 (mark_unavailable_hard_regs): Likewise.
3820 (choose_best_reg_1): Likewise.
3821 (verify_target_availability): Likewise.
3822 * valtrack.c (dead_debug_insert_temp): Likewise.
3823 * var-tracking.c (track_loc_p): Likewise.
3824 (emit_note_insn_var_location): Likewise.
3825 * varasm.c (make_decl_rtl): Likewise.
3826 * reginfo.c (choose_hard_reg_mode): Likewise.
3827 (init_reg_modes_target): Refer directly to
3828 this_target_regs->x_hard_regno_nregs.
3830 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3832 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
3833 instead of hard_regno_nregs.
3835 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3837 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
3838 end_hard_regno instead of hard_regno_nregs.
3839 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
3840 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
3841 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
3842 * ira-color.c (improve_allocation): Likewise.
3843 * lra-assigns.c (find_hard_regno_for_1): Likewise.
3844 * lra-lives.c (mark_regno_live): Likewise.
3845 (mark_regno_dead): Likewise.
3846 * lra-remat.c (operand_to_remat): Likewise.
3847 * lra.c (collect_non_operand_hard_regs): Likewise.
3848 * postreload.c (reload_combine_note_store): Likewise.
3849 (move2add_valid_value_p): Likewise.
3850 * reload.c (regno_clobbered_p): Likewise.
3852 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3854 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
3856 * config/v850/v850.c (v850_reorg): Likewise.
3857 * reload.c (refers_to_regno_for_reload_p): Likewise.
3858 (find_equiv_reg): Likewise.
3859 * reload1.c (reload_reg_reaches_end_p): Likewise.
3861 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3863 * caller-save.c (add_used_regs): Use REG_NREGS instead of
3865 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
3866 * config/arm/arm.c (output_move_neon): Likewise.
3867 (arm_attr_length_move_neon): Likewise.
3868 (neon_split_vcombine): Likewise.
3869 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
3870 (c6x_mark_reg_written): Likewise.
3871 (c6x_dwarf_register_span): Likewise.
3872 * config/i386/i386.c (ix86_save_reg): Likewise.
3873 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
3874 (rws_access_reg): Likewise.
3875 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
3876 * mode-switching.c (create_pre_exit): Likewise.
3877 * ree.c (combine_reaching_defs): Likewise.
3878 (add_removable_extension): Likewise.
3879 * regcprop.c (find_oldest_value_reg): Likewise.
3880 (copyprop_hardreg_forward_1): Likewise.
3881 * reload.c (reload_inner_reg_of_subreg): Likewise.
3882 (push_reload): Likewise.
3883 (combine_reloads): Likewise.
3884 (find_dummy_reload): Likewise.
3885 (reload_adjust_reg_for_mode): Likewise.
3886 * reload1.c (find_reload_regs): Likewise.
3887 (forget_old_reloads_1): Likewise.
3888 (reload_reg_free_for_value_p): Likewise.
3889 (reload_adjust_reg_for_temp): Likewise.
3890 (emit_reload_insns): Likewise.
3891 (delete_output_reload): Likewise.
3892 * sel-sched.c (choose_best_reg_1): Likewise.
3893 (choose_best_pseudo_reg): Likewise.
3895 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3896 Alan Hayward <alan.hayward@arm.com>
3897 David Sherwood <david.sherwood@arm.com>
3899 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
3900 * target.def (slow_unaligned_access): New hook.
3901 * targhooks.h (default_slow_unaligned_access): Declare.
3902 * targhooks.c (default_slow_unaligned_access): New function.
3903 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
3904 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
3905 * doc/tm.texi: Regenerate.
3906 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
3907 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
3908 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
3910 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
3911 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
3913 (rs6000_slow_unaligned_access): New function.
3914 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3915 (expand_block_compare): Likewise.
3916 (expand_strn_compare): Likewise.
3917 (rs6000_rtx_costs): Likewise.
3918 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
3919 (riscv_slow_unaligned_access): Likewise.
3920 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
3921 (riscv_slow_unaligned_access_p): ...this and make static.
3922 (riscv_option_override): Update accordingly.
3923 (riscv_slow_unaligned_access): New function.
3924 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3925 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
3926 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
3927 (rs6000_slow_unaligned_access): New function.
3928 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
3929 (rs6000_rtx_costs): Likewise.
3930 * config/rs6000/rs6000-string.c (expand_block_compare)
3931 (expand_strn_compare): Use targetm.slow_unaligned_access instead
3932 of SLOW_UNALIGNED_ACCESS.
3933 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
3934 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
3935 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
3936 of SLOW_UNALIGNED_ACCESS.
3937 * expmed.c (simple_mem_bitfield_p): Likewise.
3938 * expr.c (alignment_for_piecewise_move): Likewise.
3939 (emit_group_load_1): Likewise.
3940 (emit_group_store): Likewise.
3941 (copy_blkmode_from_reg): Likewise.
3942 (emit_push_insn): Likewise.
3943 (expand_assignment): Likewise.
3944 (store_field): Likewise.
3945 (expand_expr_real_1): Likewise.
3946 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
3947 * lra-constraints.c (simplify_operand_subreg): Likewise.
3948 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
3949 * gimple-ssa-store-merging.c: Likewise in block comment at start
3951 * tree-ssa-strlen.c: Include target.h.
3952 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
3953 of SLOW_UNALIGNED_ACCESS.
3954 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
3956 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
3958 PR rtl-optimization/82185
3959 * expmed.c (emit_store_flag_int): Only test tem if it has been
3962 2017-09-12 Richard Biener <rguenther@suse.de>
3965 * match.pd ((FTYPE) N CMP CST): Fix typo.
3967 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
3969 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
3971 (mips_near_type_p): Add 'short_call' attribute as a synonym
3973 * doc/extend.texi (short_call): Document new function attribute.
3975 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3978 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
3979 assertion check that in the condition.
3980 (get_atomic_generic_size): Likewise. Before testing if parameter
3981 has pointer type, if it has array type, call for C++
3982 default_conversion to perform array-to-pointer conversion.
3984 2017-09-12 Richard Biener <rguenther@suse.de>
3986 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
3987 for operations we cannot scalarize.
3989 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
3991 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
3992 vectors heap vectors. Clean up comments.
3993 Make visited_bbs a reference.
3994 (profitable_jump_thread_path): Make GC
3995 vectors heap vectors. Clean up comments.
3997 (convert_and_register_jump_thread_path): Make GC vectors heap
3999 (check_subpath_and_update_thread_path): Same. Clean up comments.
4000 Make visited_bbs a reference.
4001 (handle_phi): Abstract common code to to
4002 register_jump_thread_path_if_profitable.
4003 Rename VAR_BB to DEF_BB.
4005 Make GC vectors heap vectors.
4006 Make visited_bbs a reference.
4007 (handle_assignment): Same.
4008 (register_jump_thread_path_if_profitable): New.
4009 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
4011 Make GC vectors heap vectors. Clean up comments.
4012 Make visited_bbs a reference.
4013 (find_jump_threads_backwards): Make visited_bbs live in the stack.
4014 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
4017 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
4020 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
4021 words of E_DImode object are reachable by xtensa_uimm8x4 access.
4023 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
4025 Revert r251800 and r251799.
4027 2017-09-11 Martin Jambor <mjambor@suse.cz>
4030 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
4031 arguments in advance.
4032 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
4033 use it to find predecessor edges.
4034 (naive_outof_ssa): Collect vector of predecessors.
4036 2017-09-08 Jason Merrill <jason@redhat.com>
4038 PR c++/70029 - ICE with ref-qualifier and -flto
4039 * langhooks.h (struct lang_hooks_for_types): Add
4040 copy_lang_qualifiers.
4041 * attribs.c (build_type_attribute_qual_variant): Use it.
4042 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
4044 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
4045 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
4047 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
4050 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
4051 (*mulsi3_sp64): New instruction.
4052 (mulsi3): New expander.
4054 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
4056 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
4058 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4060 * sancov.c: Include memmodel.h.
4062 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
4065 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
4068 2017-09-07 Carl Love <cel@us.ibm.com>
4070 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
4071 the sldi instruction.
4073 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
4075 * sancov.c: Include tm_p.h.
4077 2017-09-07 Jakub Jelinek <jakub@redhat.com>
4080 * output.h (switch_to_other_text_partition): New declaration.
4081 * varasm.c (switch_to_other_text_partition): New function.
4082 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
4083 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
4084 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
4085 to the other text partition before emitting LCL label and switch back
4086 after emitting the word after it.
4088 2017-09-07 Richard Biener <rguenther@suse.de>
4090 * passes.def (pass_split_crit_edges): Remove instance before PRE.
4091 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
4092 critical edges here, after loop init.
4093 (pass_data_pre): Remove PROP_no_crit_edges flags.
4094 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
4095 for valueization of call args to avoid leaking VN_TOP.
4096 (visit_use): Assert we do not visit default defs.
4097 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
4098 Use error_mark_node to more easily detect leaking VN_TOP.
4099 All default-defs are varying, not VN_TOP. Mark them visited.
4100 (run_scc_vn): Make code match comment.
4102 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
4104 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
4105 OPTION_MASK_FLOAT128_KEYWORD.
4106 (POWERPC_MASKS): Likewise.
4107 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
4108 support for the -mfloat128-type option, and make -mfloat128
4109 default on PowerPC Linux systems. Define or undefine
4110 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
4111 Define __float128 to be __ieee128 if IEEE 128-bit support is
4112 enabled, or undefine it.
4113 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
4114 Delete defining __FLOAT128_TYPE__.
4115 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
4116 -mfloat128-type option and make -mfloat128 default on PowerPC
4118 (TARGET_FLOAT128_TYPE): Likewise.
4119 (-mfloat128-type): Likewise.
4120 * config/rs6000/rs6000.c (rs6000_option_override_internal):
4121 Delete the -mfloat128-type option and make -mfloat128 default on
4122 PowerPC Linux systems. Always use __ieee128 to be the keyword for
4123 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
4124 128-bit floating point is enabled. Change tests from using
4125 -mfloat128-type to -mfloat128.
4126 (rs6000_mangle_type): Use the correct mangling for the __float128
4127 type even if normal long double is restricted to 64-bits.
4128 (floatn_mode): Enable the _Float128 type by default on VSX Linux
4130 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
4131 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
4132 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
4134 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
4135 documentation for -mfloat128.
4137 2017-09-06 Olivier Hainque <hainque@adacore.com>
4139 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
4141 2017-09-06 Wish Wu <wishwu007@gmail.com>
4142 Jakub Jelinek <jakub@redhat.com>
4144 * asan.c (initialize_sanitizer_builtins): Add
4145 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
4146 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
4147 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
4148 BT_FN_VOID_UINT64_PTR variables.
4149 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
4150 (BT_FN_VOID_UINT16_UINT16): Likewise.
4151 (BT_FN_VOID_UINT32_UINT32): Likewise.
4152 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
4153 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
4154 (BT_FN_VOID_UINT64_PTR): Likewise.
4155 * common.opt (flag_sanitize_coverage): New variable.
4156 (fsanitize-coverage=trace-pc): Remove.
4157 (fsanitize-coverage=): Add.
4158 * flag-types.h (enum sanitize_coverage_code): New enum.
4159 * fold-const.c (fold_range_test): Disable non-short-circuit
4160 optimization if flag_sanitize_coverage.
4161 (fold_truth_andor): Likewise.
4162 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
4163 * opts.c (COVERAGE_SANITIZER_OPT): Define.
4164 (coverage_sanitizer_opts): New array.
4165 (get_closest_sanitizer_option): Add OPTS argument, handle also
4166 OPT_fsanitize_coverage_.
4167 (parse_sanitizer_options): Adjusted to also handle
4168 OPT_fsanitize_coverage_.
4169 (common_handle_option): Add OPT_fsanitize_coverage_.
4170 * sancov.c (instrument_comparison, instrument_switch): New function.
4171 (sancov_pass): Add trace-cmp support.
4172 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
4173 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
4174 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
4175 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
4176 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
4177 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
4178 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
4179 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
4180 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
4181 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
4183 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
4185 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
4186 error. Only quit immediately if parsing is complete.
4187 (BEGIN): Initialize fatal_err and parse_done.
4188 (begin fpu, end fpu): Check number of arguments.
4189 (begin arch, end arch): Likewise.
4190 (begin cpu, end cpu): Likewise.
4191 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
4192 (optalias): Likewise.
4194 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
4196 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
4197 * config/arm/arm-isa.h: Delete. Move definitions to ...
4198 * arm-cpus.in: ... here. Use new feature and fgroup values.
4199 * config/arm/arm.c (arm_option_override): Use lower case for feature
4201 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
4202 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
4203 * config/arm/parsecpu.awk (END): Add new command 'isa'.
4205 (print_isa_bits_for): New function.
4206 (gen_isa): New function.
4207 (gen_comm_data): Use print_isa_bits_for.
4208 (define feature): New keyword.
4209 (define fgroup): New keyword.
4210 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
4211 (arm-isa.h): Add rule to generate file.
4212 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
4213 case for feature bit names.
4215 2017-09-06 Richard Biener <rguenther@suse.de>
4217 * tree-ssa-pre.c (NECESSARY): Remove.
4218 (create_expression_by_pieces): Do not touch pass-local flags.
4219 (insert_into_preds_of_block): Likewise.
4220 (do_pre_regular_insertion): Likewise.
4221 (eliminate_insert): Likewise.
4222 (eliminate_dom_walker::before_dom_children): Likewise.
4223 (fini_eliminate): Do not look at inserted_exprs.
4224 (mark_operand_necessary): Remove.
4225 (remove_dead_inserted_code): Replace with simple work-list
4226 algorithm based on inserted_exprs and SSA uses.
4227 (pass_pre::execute): Re-order fini_eliminate and
4228 remove_dead_inserted_code.
4230 2017-09-06 Olivier Hainque <hainque@adacore.com>
4232 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
4233 for VxWorks 7. Adjust surrounding comments.
4235 2017-09-06 Richard Biener <rguenther@suse.de>
4237 * gimple-ssa-strength-reduction.c
4238 (find_candidates_dom_walker::before_dom_children): Also allow
4241 2017-09-06 Richard Biener <rguenther@suse.de>
4243 PR tree-optimization/82108
4244 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
4245 for gap in the non-permutation SLP case.
4247 2017-09-06 Martin Jambor <mjambor@suse.cz>
4249 PR tree-optimization/82078
4250 * tree-sra.c (sort_and_splice_var_accesses): Move call to
4251 add_access_to_work_queue...
4252 (build_accesses_from_assign): ...here.
4253 (propagate_all_subaccesses): Make sure racc is the group
4254 representative, if there is one.
4256 2017-09-06 Jakub Jelinek <jakub@redhat.com>
4259 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
4262 2017-09-06 Richard Biener <rguenther@suse.de>
4264 * gimple-ssa-strength-reduction.c
4265 (find_candidates_dom_walker::before_doom_children): Use a
4266 type and not a mode check.
4268 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
4271 * config/arm/predicates.md (arm_general_adddi_operand): Create new
4273 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
4275 2017-09-05 Jeff Law <law@redhat.com>
4277 PR tree-optimization/64910
4278 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
4279 cases where we have 3 or more operands.
4281 2017-09-05 Jakub Jelinek <jakub@redhat.com>
4284 * omp-low.c (lower_omp_for): Recompute tree invariant if
4285 gimple_omp_for_initial/final is ADDR_EXPR.
4288 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
4289 into gimple val before gimplification fo the COND_EXPR.
4291 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
4293 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
4294 REGION_COPY argument.
4295 (thread_through_all_blocks): Remove unused argument to
4296 duplicate_thread_path.
4298 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4299 Alan Hayward <alan.hayward@arm.com>
4300 David Sherwood <david.sherwood@arm.com>
4302 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
4303 Take a scalar_mode rather than a machine_mode.
4304 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
4305 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
4306 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
4307 (aarch64_gen_adjusted_ldpstp): Likewise.
4308 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
4310 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4311 Alan Hayward <alan.hayward@arm.com>
4312 David Sherwood <david.sherwood@arm.com>
4314 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
4315 Take a scalar_int_mode instead of a machine_mode.
4316 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
4317 (aarch64_output_scalar_simd_mov_immediate): Likewise.
4318 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
4319 (aarch64_simd_attr_length_rglist): Delete.
4320 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
4321 a scalar_int_mode instead of a machine_mode.
4322 (aarch64_add_offset): Likewise.
4323 (aarch64_internal_mov_immediate): Likewise
4324 (aarch64_add_constant_internal): Likewise.
4325 (aarch64_add_constant): Likewise.
4326 (aarch64_movw_imm): Likewise.
4327 (aarch64_rtx_arith_op_extract_p): Likewise.
4328 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
4329 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
4330 Remove assert that the mode isn't a vector.
4331 (aarch64_output_scalar_simd_mov_immediate): Likewise.
4332 (aarch64_expand_mov_immediate): Update calls after above changes.
4333 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
4334 (aarch64_and_bitmask_imm): Check for scalar integer modes.
4335 (aarch64_move_imm): Likewise.
4336 (aarch64_can_const_movi_rtx_p): Likewise.
4337 (aarch64_strip_extend): Likewise.
4338 (aarch64_extr_rtx_p): Likewise.
4339 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
4340 a CONST_INT when the mode parameter is VOIDmode.
4341 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
4343 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4345 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
4346 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4347 (bitwise_type_for_mode): Update accordingly.
4349 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4351 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
4352 * stor-layout.c (mode_for_size_tree): Likewise.
4353 (mode_for_array): Update accordingly.
4354 (layout_decl): Likewise.
4355 (compute_record_mode): Likewise. Only set the mode once.
4357 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4359 * target.def (get_mask_mode): Change return type to opt_mode.
4361 * doc/tm.texi: Regenerate.
4362 * targhooks.h (default_get_mask_mode): Return an opt_mode.
4363 * targhooks.c (default_get_mask_mode): Likewise.
4364 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4365 * optabs-query.c (can_vec_mask_load_store_p): Update use of
4366 targetm.get_mask_mode.
4367 * tree.c (build_truth_vector_type): Likewise.
4369 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4371 * machmode.h (mode_for_vector): Return an opt_mode.
4372 * stor-layout.c (mode_for_vector): Likewise.
4373 (mode_for_int_vector): Update accordingly.
4374 (layout_type): Likewise.
4375 * config/i386/i386.c (emit_memmov): Likewise.
4376 (ix86_expand_set_or_movmem): Likewise.
4377 (ix86_expand_vector_init): Likewise.
4378 (ix86_get_mask_mode): Likewise.
4379 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
4381 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
4382 * expmed.c (extract_bit_field_1): Likewise.
4383 * expr.c (expand_expr_real_2): Likewise.
4384 * optabs-query.c (can_vec_perm_p): Likewise.
4385 (can_vec_mask_load_store_p): Likewise.
4386 * optabs.c (expand_vec_perm): Likewise.
4387 * targhooks.c (default_get_mask_mode): Likewise.
4388 * tree-vect-stmts.c (vectorizable_store): Likewise.
4389 (vectorizable_load): Likewise.
4390 (get_vectype_for_scalar_type_and_size): Likewise.
4392 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4394 * machmode.h (mode_for_int_vector): New function.
4395 * stor-layout.c (mode_for_int_vector): Likewise.
4396 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
4397 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
4398 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4399 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4400 (s390_expand_vcond): Likewise.
4402 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4404 * machmode.h (opt_machine_mode): New type.
4405 (opt_mode<T>): Allow construction from anything that can be
4407 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
4408 (mode_for_size): Return an opt_machine_mode.
4409 * stor-layout.c (mode_for_size): Likewise.
4410 (mode_for_size_tree): Update call accordingly.
4411 (bitwise_mode_for_mode): Likewise.
4412 (make_fract_type): Likewise.
4413 (make_accum_type): Likewise.
4414 * caller-save.c (replace_reg_with_saved_mem): Update call
4416 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4417 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4418 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4419 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4420 * expmed.c (extract_bit_field_1): Likewise.
4421 * reload.c (get_secondary_mem): Likewise.
4422 * varasm.c (assemble_integer): Likewise.
4423 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
4426 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4428 * machmode.h (decimal_float_mode_for_size): New function.
4429 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
4430 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
4431 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
4432 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
4433 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
4434 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
4436 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4438 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
4439 (get_builtin_sync_mode): Likewise.
4440 (expand_ifn_atomic_compare_exchange): Likewise.
4441 (expand_builtin_atomic_clear): Likewise.
4442 (expand_builtin_atomic_test_and_set): Likewise.
4443 (fold_builtin_atomic_always_lock_free): Likewise.
4444 * calls.c (compute_argument_addresses): Likewise.
4445 (emit_library_call_value_1): Likewise.
4446 (store_one_arg): Likewise.
4447 * combine.c (combine_instructions): Likewise.
4448 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
4449 * config/arm/arm.c (arm_function_value): Likewise.
4450 (aapcs_allocate_return_reg): Likewise.
4451 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
4452 * config/i386/i386.c (construct_container): Likewise.
4453 (ix86_gimplify_va_arg): Likewise.
4454 (ix86_expand_sse_cmp): Likewise.
4455 (emit_memmov): Likewise.
4456 (emit_memset): Likewise.
4457 (expand_small_movmem_or_setmem): Likewise.
4458 (ix86_expand_pextr): Likewise.
4459 (ix86_expand_pinsr): Likewise.
4460 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
4461 * config/microblaze/microblaze.c (microblaze_block_move_straight):
4463 * config/mips/mips.c (mips_function_value_1) Likewise.
4464 (mips_block_move_straight): Likewise.
4465 (mips_expand_ins_as_unaligned_store): Likewise.
4466 * config/powerpcspe/powerpcspe.c
4467 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4468 (rs6000_darwin64_record_arg_flush): Likewise.
4469 * config/rs6000/rs6000.c
4470 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4471 (rs6000_darwin64_record_arg_flush): Likewise.
4472 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
4473 (sparc_function_value_1): Likewise.
4474 * config/spu/spu.c (adjust_operand): Likewise.
4475 (spu_emit_branch_or_set): Likewise.
4476 (arith_immediate_p): Likewise.
4477 * emit-rtl.c (gen_lowpart_common): Likewise.
4478 * expr.c (expand_expr_real_1): Likewise.
4479 * function.c (assign_parm_setup_block): Likewise.
4480 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
4481 * reload1.c (alter_reg): Likewise.
4482 * stor-layout.c (mode_for_vector): Likewise.
4483 (layout_type): Likewise.
4485 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4487 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
4488 (spu_convert_move): Likewise.
4489 * lower-subreg.c (resolve_simple_move): Likewise.
4491 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4494 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
4495 define_insn to a define_expand.
4496 (altivec_vsum2sws_direct): New define_insn.
4497 (altivec_vsumsws): Convert from a define_insn to a define_expand.
4499 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4501 * config/arm/arm.c (arm_option_params_internal): Improve setting of
4504 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
4509 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
4510 realign stack if stack alignment needed is less than incoming
4513 2017-09-05 Marek Polacek <polacek@redhat.com>
4516 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
4519 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4521 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
4523 2017-09-05 Richard Biener <rguenther@suse.de>
4525 PR tree-optimization/82084
4526 * fold-const.c (can_native_encode_string_p): Handle wide characters.
4528 2017-09-05 Richard Biener <rguenther@suse.de>
4530 PR tree-optimization/82102
4531 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
4533 2017-09-05 Martin Liska <mliska@suse.cz>
4535 PR tree-optimization/82032
4536 * tree-cfg.c (generate_range_test): New function.
4537 * tree-cfg.h (generate_range_test): Declared here.
4538 * tree-cfgcleanup.c (convert_single_case_switch): New function.
4539 (cleanup_control_expr_graph): Use it.
4540 * tree-switch-conversion.c (try_switch_expansion): Remove
4542 (emit_case_nodes): Use generate_range_test.
4544 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
4547 * config/i386/i386.md (*<btsc><mode>_mask): Add
4548 TARGET_USE_BT to insn constraint.
4549 (*btr<mode>_mask): Ditto.
4551 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
4553 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
4554 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
4556 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4559 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
4560 TARGET_NEON and TARGET_IWMMXT.
4561 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
4562 TARGET_NEON and TARGET_IWMMXT.
4563 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
4565 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
4567 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
4568 (ix86_rewrite_tls_address): Ditto.
4569 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
4570 (ix86_rewrite_tls_address_1): Ditto.
4571 (ix86_rewrite_tls_address): Ditto.
4572 * config/i386/predicates.md (tls_address_pattern): New predicate.
4573 * config/i386/i386.md (TLS address splitter): New splitter.
4575 2017-09-04 Richard Biener <rguenther@suse.de>
4577 PR tree-optimization/82084
4578 * fold-const.h (can_native_encode_string_p): Declare.
4579 * fold-const.c (can_native_encode_string_p): Factor out from ...
4580 (native_encode_string): ... here.
4581 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
4582 vectorizing stores from constants we later cannot handle.
4584 2017-09-04 Marek Polacek <polacek@redhat.com>
4587 * doc/invoke.texi: Update -Wtautological-compare documentation.
4589 2017-09-04 Jeff Law <law@redhat.com>
4591 PR tree-optimization/64910
4592 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
4593 swap the first and last operand if the last is a constant.
4595 2017-09-04 Marek Polacek <polacek@redhat.com>
4598 * convert.c (do_narrow): When sanitizing signed integer overflows,
4599 bail out for signed types.
4600 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
4602 2017-09-04 Richard Biener <rguenther@suse.de>
4604 PR tree-optimization/82060
4605 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
4606 Move devirtualization after stmt folding and before EH/AB/noreturn
4607 cleanup to get the stmt refs canonicalized. Use a bool instead
4608 of gimple_modified_p since that doesn't work for NOPs. Schedule
4609 NOPs generated by folding for removal.
4611 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4612 Alan Hayward <alan.hayward@arm.com>
4613 David Sherwood <david.sherwood@arm.com>
4615 * coretypes.h (pad_direction): New enum.
4616 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
4617 (FUNCTION_ARG_PADDING): Likewise.
4618 * target.def (function_arg_padding): New hook.
4619 * targhooks.h (default_function_arg_padding): Declare.
4620 * targhooks.c (default_function_arg_padding): New function.
4621 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
4622 (TARGET_FUNCTION_ARG_PADDING): ...this.
4623 * doc/tm.texi: Regenerate.
4624 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
4625 instead of direction.
4626 (compute_argument_addresses): Likewise.
4627 (load_register_parameters): Likewise.
4628 (emit_library_call_value_1): Likewise.
4629 (store_one_arg): Use targetm.calls.function_arg_padding instead
4630 of FUNCTION_ARG_PADDING.
4631 (must_pass_in_stack_var_size_or_pad): Likewise.
4632 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
4633 (emit_group_store): Likewise.
4634 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
4635 instead of FUNCTION_ARG_PADDING.
4636 (emit_push_insn): Likewise, and propagate enum change throughout
4638 * function.h (direction): Delete.
4639 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
4641 * function.c (assign_parm_find_stack_rtl): Likewise.
4642 (assign_parm_setup_block_p): Likewise.
4643 (assign_parm_setup_block): Likewise.
4644 (gimplify_parameters): Likewise.
4645 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
4646 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
4648 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
4649 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4650 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
4651 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
4652 (aarch64_function_arg_padding): ...this new function.
4653 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
4654 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4655 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
4656 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4657 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
4658 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4659 (arm_pad_arg_upward): Replace with...
4660 (arm_function_arg_padding): ...this new function.
4661 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
4663 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
4664 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
4665 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4666 (ia64_hpux_function_arg_padding): Replace with...
4667 (ia64_function_arg_padding): ...this new function. Use pad_direction
4668 instead of direction. Check for TARGET_HPUX.
4669 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
4670 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4671 (iq2000_function_arg_padding): New function.
4672 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
4673 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
4674 (mips_function_arg_padding): ...this new function.
4675 (mips_pad_reg_upward): Update accordingly.
4676 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4677 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
4678 targetm.calls.function_arg_padding.
4679 (FUNCTION_ARG_PADDING): Delete.
4680 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
4681 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
4682 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4683 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
4684 (nios2_block_reg_padding): Return pad_direction instead of direction.
4685 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
4686 instead of direction.
4687 (nios2_function_arg_padding): Likewise. Make static.
4688 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4689 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
4690 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
4691 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
4692 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4693 (pa_function_arg_padding): Make static. Return pad_direction instead
4695 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
4696 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4697 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
4698 instead of direction. Use targetm.calls.function_arg_padding.
4699 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
4700 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4701 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
4702 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
4703 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
4705 (function_arg_padding): Rename to...
4706 (rs6000_function_arg_padding): ...this. Make static. Return
4707 pad_direction instead of direction.
4708 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4709 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
4710 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
4711 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
4712 instead of direction. Use targetm.calls.function_arg_padding.
4713 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
4714 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
4715 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
4716 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
4717 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4718 (function_arg_padding): Rename to...
4719 (rs6000_function_arg_padding): ...this. Make static. Return
4720 pad_direction instead of direction.
4721 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
4722 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
4723 * config/s390/s390.c (s390_function_arg_padding): New function.
4724 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4725 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
4726 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
4727 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
4728 (function_arg_padding): Rename to...
4729 (sparc_function_arg_padding): ...this. Make static. Return
4730 pad_direction instead of direction.
4731 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
4732 * config/spu/spu.c (spu_function_arg_padding): New function.
4733 (TARGET_FUNCTION_ARG_PADDING): Redefine.
4734 * system.h (FUNCTION_ARG_PADDING): Poison.
4736 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4737 Alan Hayward <alan.hayward@arm.com>
4738 David Sherwood <david.sherwood@arm.com>
4740 * target.def (modes_tieable_p): New hook.
4741 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
4742 (TARGET_MODES_TIEABLE_P): ...this.
4743 * doc/tm.texi.in: Regenerate.
4744 * hooks.h (hook_bool_mode_mode_true): Declare.
4745 * hooks.c (hook_bool_mode_mode_true): New function.
4746 * combine.c (subst): Use targetm.modes_tieable_p instead of
4748 * dse.c (find_shift_sequence): Likewise.
4749 * expmed.c (extract_low_bits): Likewise.
4750 * lower-subreg.c: Include target.h.
4751 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
4753 * rtlanal.c (rtx_cost): Likewise.
4754 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
4755 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
4756 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
4757 (TARGET_MODES_TIEABLE_P): Redefine.
4758 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
4759 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
4760 (TARGET_MODES_TIEABLE_P): Redefine.
4761 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
4762 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
4763 (arc_modes_tieable_p): New function.
4764 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
4765 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
4766 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
4767 (arm_modes_tieable_p): Make static.
4768 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
4769 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
4770 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
4771 (TARGET_MODES_TIEABLE_P): Redefine.
4772 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
4773 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
4774 (TARGET_MODES_TIEABLE_P): Redefine.
4775 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
4776 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
4777 (cr16_modes_tieable_p): New function.
4778 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
4779 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
4780 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
4781 (TRULY_NOOP_TRUNCATION): Update comment.
4782 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
4783 (TRULY_NOOP_TRUNCATION): Update comment.
4784 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
4785 (frv_modes_tieable_p): New function.
4786 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
4787 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
4788 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
4789 (TARGET_MODES_TIEABLE_P): Redefine.
4790 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
4791 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
4792 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
4793 (TARGET_MODES_TIEABLE_P): Redefine.
4794 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
4795 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
4796 (ia64_modes_tieable_p): New function.
4797 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
4798 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
4799 (iq2000_modes_tieable_p): New function.
4800 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
4801 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
4802 (lm32_modes_tieable_p): New function.
4803 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
4804 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
4805 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
4806 (TARGET_MODES_TIEABLE_P): Redefine.
4807 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
4808 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
4809 (m32r_modes_tieable_p): New function.
4810 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
4811 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
4812 (m68k_modes_tieable_p): New function.
4813 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
4814 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
4815 (mcore_modes_tieable_p): New function.
4816 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
4817 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
4819 (TARGET_MODES_TIEABLE_P): Redefine.
4820 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
4821 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
4822 * config/mips/mips.c (mips_modes_tieable_p): Make static.
4823 (TARGET_MODES_TIEABLE_P): Redefine.
4824 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
4825 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
4826 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
4827 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
4828 (mn10300_modes_tieable_p): ...this and make static.
4829 (TARGET_MODES_TIEABLE_P): Redefine.
4830 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
4831 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
4832 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
4833 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
4834 (msp430_modes_tieable_p): Make static.
4835 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
4836 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
4837 (TARGET_MODES_TIEABLE_P): Redefine.
4838 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
4839 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
4840 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
4841 (TARGET_MODES_TIEABLE_P): Redefine.
4842 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
4843 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
4844 * config/pa/pa.c (pa_modes_tieable_p): Make static.
4845 (TARGET_MODES_TIEABLE_P): Redefine.
4846 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
4847 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
4848 (pdp11_modes_tieable_p): New function.
4849 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
4850 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
4851 (rs6000_modes_tieable_p): New function.
4852 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4853 * config/powerpcspe/powerpcspe.md: Update comment.
4854 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
4855 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
4856 (TARGET_MODES_TIEABLE_P): Redefine.
4857 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
4858 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
4859 (rl78_modes_tieable_p): New function.
4860 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
4861 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
4862 (rs6000_modes_tieable_p): New function.
4863 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
4864 * config/rs6000/rs6000.md: Update comment.
4865 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
4866 * config/rx/rx.c (rx_modes_tieable_p): New function.
4867 (TARGET_MODES_TIEABLE_P): Redefine.
4868 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
4869 * config/s390/s390.c (s390_modes_tieable_p): New function.
4870 (TARGET_MODES_TIEABLE_P): Redefine.
4871 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
4872 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
4873 (sh_modes_tieable_p): New function.
4874 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
4875 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
4876 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
4877 (sparc_modes_tieable_p): Make static.
4878 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
4879 * config/spu/spu.c (spu_modes_tieable_p): New function.
4880 (TARGET_MODES_TIEABLE_P): Redefine.
4881 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
4882 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
4883 (TARGET_MODES_TIEABLE_P): Redefine.
4884 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
4885 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
4886 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
4887 * config/v850/v850.c (v850_modes_tieable_p): New function.
4888 (TARGET_MODES_TIEABLE_P): Redefine.
4889 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
4890 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
4891 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
4892 (visium_modes_tieable_p): New function.
4893 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
4894 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
4895 (xtensa_modes_tieable_p): New function.
4896 * system.h (MODES_TIEABLE_P): Poison.
4898 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
4899 Alan Hayward <alan.hayward@arm.com>
4900 David Sherwood <david.sherwood@arm.com>
4902 * target.def (hard_regno_mode_ok): New hook.
4903 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
4904 (TARGET_HARD_REGNO_MODE_OK): ...this.
4905 * doc/tm.texi.in: Regenerate.
4906 * hooks.h (hook_bool_uint_mode_true): Declare.
4907 * hooks.c (hook_bool_uint_mode_true): New function.
4908 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
4910 * genpreds.c (write_insn_preds_c): Add an include of target.h.
4911 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
4912 instead of HARD_REGNO_MODE_OK.
4913 * caller-save.c: Include target.h.
4914 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
4916 * combine.c (can_combine_p): Likewise.
4917 (combinable_i3pat): Likewise.
4918 (can_change_dest_mode): Likewise.
4919 * expr.c (init_expr_target): Likewise.
4920 (convert_move): Likewise.
4921 (convert_modes): Likewise.
4922 * ira.c (setup_prohibited_class_mode_regs): Likewise.
4923 (setup_prohibited_mode_move_regs): Likewise.
4924 * ira.h (target_ira): Likewise.
4925 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4926 * lra-constraints.c (process_alt_operands): Likewise.
4927 (split_reg): Likewise.
4928 * recog.c (peep2_find_free_register): Likewise.
4929 * ree.c (combine_reaching_defs): Likewise.
4930 * regcprop.c (maybe_mode_change): Likewise.
4931 * reginfo.c (init_reg_sets_1): Likewise.
4932 (choose_hard_reg_mode): Likewise.
4933 (simplifiable_subregs): Likewise.
4934 * regrename.c (check_new_reg_p): Likewise.
4935 * reload.c (find_valid_class): Likewise.
4936 (find_valid_class_1): Likewise.
4937 (reload_inner_reg_of_subreg): Likewise.
4938 (push_reload): Likewise.
4939 (combine_reloads): Likewise.
4940 (find_dummy_reload): Likewise.
4941 (find_reloads): Likewise.
4942 * reload1.c (find_reg): Likewise.
4943 (set_reload_reg): Likewise.
4944 (allocate_reload_reg): Likewise.
4945 (choose_reload_regs): Likewise.
4946 (reload_adjust_reg_for_temp): Likewise.
4947 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
4948 (simplify_subreg_regno): Likewise.
4949 * sel-sched.c (init_regs_for_mode): Likewise.
4950 * varasm.c (make_decl_rtl): Likewise.
4951 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
4952 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
4954 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
4955 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
4956 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4957 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
4958 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
4959 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4960 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
4961 (arc_mode_class): Delete.
4962 (HARD_REGNO_MODE_OK): Delete.
4963 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4964 (arc_hard_regno_mode_ok): Rename old array to...
4965 (arc_hard_regno_mode_ok_modes): ...this.
4966 (arc_conditional_register_usage): Update accordingly.
4967 (arc_mode_class): Make static.
4968 (arc_hard_regno_mode_ok): New function.
4969 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
4970 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
4971 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4972 (arm_hard_regno_mode_ok): Make static.
4973 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
4975 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
4976 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
4977 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
4979 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4980 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
4981 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
4982 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
4983 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
4984 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4985 * config/bfin/predicates.md (valid_reg_operand): Use
4986 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
4987 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
4988 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
4989 (TARGET_HARD_REGNO_MODE_OK): Redefine.
4990 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
4991 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
4992 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4993 (cr16_hard_regno_mode_ok): Make static and return a bool.
4994 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
4995 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
4996 (cris_hard_regno_mode_ok): New function.
4997 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
4998 (epiphany_mode_class): Delete.
4999 (HARD_REGNO_MODE_OK): Delete.
5000 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
5001 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5002 (hard_regno_mode_ok): Rename to...
5003 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
5004 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
5005 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
5007 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
5008 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
5009 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5010 (frv_hard_regno_mode_ok): Make static and return a bool.
5011 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
5013 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
5014 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
5015 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
5016 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
5018 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5019 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
5020 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
5021 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
5023 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5024 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
5025 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5026 (ia64_hard_regno_mode_ok): New function.
5027 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
5028 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5029 (iq2000_hard_regno_mode_ok): New function.
5030 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
5031 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5032 (lm32_hard_regno_mode_ok): New function.
5033 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
5034 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
5035 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
5036 instead of HARD_REGNO_MODE_OK.
5037 (m32c_hard_regno_ok): Rename to...
5038 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
5039 (m32c_cannot_change_mode_class): Update accordingly.
5040 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5041 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
5042 (m32r_mode_class): Delete.
5043 (HARD_REGNO_MODE_OK): Delete.
5044 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5045 (m32r_hard_regno_mode_ok): Rename to...
5046 (m32r_hard_regno_modes): ...this.
5047 (m32r_mode_class): Make static.
5048 (m32r_hard_regno_mode_ok): New function.
5049 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
5050 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
5051 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5052 (m68k_hard_regno_mode_ok): Make static.
5053 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
5054 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5055 (mcore_hard_regno_mode_ok): New function.
5056 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
5057 (HARD_REGNO_MODE_OK): Delete.
5058 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
5060 (microblaze_hard_regno_mode_ok_p): ...this and make static.
5061 (microblaze_hard_regno_mode_ok): New function.
5062 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5063 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
5064 (mips_hard_regno_mode_ok): Delete.
5065 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
5066 (mips_hard_regno_mode_ok_p): ...this and make static.
5067 (mips_hard_regno_mode_ok_p): Rename to...
5068 (mips_hard_regno_mode_ok_uncached): ...this.
5069 (mips_hard_regno_mode_ok): New function.
5070 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
5071 of HARD_REGNO_MODE_OK.
5072 (mips_option_override): Update after above name changes.
5073 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5074 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
5075 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
5076 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
5077 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
5078 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5079 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
5080 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
5081 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
5082 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5083 (msp430_hard_regno_mode_ok): Make static and return a bool.
5084 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
5085 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
5086 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
5088 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5089 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
5090 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
5091 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
5092 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
5093 (PA_HARD_REGNO_MODE_OK): ...this
5094 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
5095 (PA_HARD_REGNO_MODE_OK): ...this.
5096 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5097 (pa_hard_regno_mode_ok): New function.
5098 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
5099 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5100 (pdp11_hard_regno_mode_ok): New function.
5101 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
5102 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
5104 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
5106 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5107 (rs6000_hard_regno_mode_ok): Rename to...
5108 (rs6000_hard_regno_mode_ok_uncached): ...this.
5109 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5110 (rs6000_hard_regno_mode_ok): New function.
5111 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
5112 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
5113 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
5114 (riscv_hard_regno_mode_ok): ...this and make static.
5115 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5116 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
5117 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
5118 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5119 (rl78_hard_regno_mode_ok): Make static and return bool.
5120 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
5121 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
5123 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
5124 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5125 (rs6000_hard_regno_mode_ok): Rename to...
5126 (rs6000_hard_regno_mode_ok_uncached): ...this.
5127 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5128 (rs6000_hard_regno_mode_ok): New function.
5129 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
5130 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
5131 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5132 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
5133 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
5134 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
5135 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5136 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
5137 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
5138 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5139 (sh_hard_regno_mode_ok): Make static.
5140 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
5141 instead of HARD_REGNO_MODE_OK.
5142 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
5143 (sparc_mode_class): Delete.
5144 (HARD_REGNO_MODE_OK): Delete.
5145 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5146 (hard_regno_mode_classes): Make static.
5147 (sparc_mode_class): Likewise.
5148 (sparc_hard_regno_mode_ok): New function.
5149 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
5150 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
5151 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
5153 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5154 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
5155 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
5156 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
5157 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
5158 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5159 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
5160 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
5161 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5162 (visium_hard_regno_mode_ok): New function.
5163 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
5164 instead of HARD_REGNO_MODE_OK.
5165 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
5166 (HARD_REGNO_MODE_OK): Delete.
5167 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
5168 (xtensa_hard_regno_mode_ok_p): ...this and make static.
5169 (xtensa_option_override): Update accordingly.
5170 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5171 (xtensa_hard_regno_mode_ok): New function.
5172 * system.h (HARD_REGNO_MODE_OK): Poison.
5174 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5175 Alan Hayward <alan.hayward@arm.com>
5176 David Sherwood <david.sherwood@arm.com>
5178 * target.def (hard_regno_call_part_clobbered): New hook.
5179 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
5180 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
5181 * doc/tm.texi: Regenerate.
5182 * hooks.h (hook_bool_uint_mode_false): Declare.
5183 * hooks.c (hook_bool_uint_mode_false): New function.
5184 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5185 * cselib.c (cselib_process_insn): Use
5186 targetm.hard_regno_call_part_clobbered instead of
5187 HARD_REGNO_CALL_PART_CLOBBERED.
5188 * ira-conflicts.c (ira_build_conflicts): Likewise.
5189 * ira-costs.c (ira_tune_allocno_costs): Likewise.
5190 * lra-constraints.c (need_for_call_save_p): Likewise.
5191 * lra-lives.c: Include target.h.
5192 (check_pseudos_live_through_calls): Use
5193 targetm.hard_regno_call_part_clobbered instead of
5194 HARD_REGNO_CALL_PART_CLOBBERED.
5195 * regcprop.c: Include target.h.
5196 (copyprop_hardreg_forward_1): Use
5197 targetm.hard_regno_call_part_clobbered instead of
5198 HARD_REGNO_CALL_PART_CLOBBERED.
5199 * reginfo.c (choose_hard_reg_mode): Likewise.
5200 * regrename.c (check_new_reg_p): Likewise.
5201 * reload.c (find_equiv_reg): Likewise.
5202 * reload1.c (emit_reload_insns): Likewise.
5203 * sched-deps.c (deps_analyze_insn): Likewise.
5204 * sel-sched.c (init_regs_for_mode): Likewise.
5205 (mark_unavailable_hard_regs): Likewise.
5206 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
5207 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5208 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
5210 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5211 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5212 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
5214 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
5216 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5217 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5218 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
5220 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5221 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5222 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
5224 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5225 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
5227 * config/powerpcspe/powerpcspe.c
5228 (rs6000_hard_regno_call_part_clobbered): New function.
5229 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5230 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5231 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
5233 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5234 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5235 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
5237 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5238 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5239 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
5241 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5242 Alan Hayward <alan.hayward@arm.com>
5243 David Sherwood <david.sherwood@arm.com>
5245 * rtl.h (subreg_memory_offset): Declare.
5246 * emit-rtl.c (subreg_memory_offset): New function.
5247 * expmed.c (store_bit_field_1): Use it.
5248 * expr.c (undefined_operand_subword_p): Likewise.
5249 * simplify-rtx.c (simplify_subreg): Likewise.
5251 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
5253 PR rtl-optimization/57448
5256 * optabs.c (expand_atomic_load): Place compiler memory barriers if
5257 using atomic_load pattern.
5258 (expand_atomic_store): Likewise.
5260 2017-09-04 Jakub Jelinek <jakub@redhat.com>
5263 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
5264 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
5265 handling. Use replace_call_with_value with NULL instead of
5266 gsi_replace, unlink_stmt_vdef and release_defs.
5268 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
5271 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
5273 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5276 * rtl.h (emit_library_call_value_1): Declare.
5277 (emit_library_call): Replace declaration with a series of overloads.
5278 Remove the parameter count argument.
5279 (emit_library_call_value): Likewise.
5280 * calls.c (emit_library_call_value_1): Make global. Replace varargs
5281 with an "rtx_mode_t *".
5282 (emit_library_call_value): Delete.
5283 (emit_library_call): Likewise.
5284 * asan.c (asan_emit_stack_protection): Update calls accordingly.
5285 (asan_emit_allocas_unpoison): Likewise.
5286 * builtins.c (expand_builtin_powi): Likewise.
5287 (expand_asan_emit_allocas_unpoison): Likewise.
5288 * cfgexpand.c (expand_main_function): Likewise.
5289 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
5290 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
5291 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
5292 * config/arm/arm.c (arm_trampoline_init): Likewise.
5293 (arm_call_tls_get_addr): Likewise.
5294 (arm_expand_divmod_libfunc): Likewise.
5295 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
5296 (smulsi3_highpart): Likewise.
5297 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
5298 (c6x_expand_compare): Likewise.
5299 (c6x_expand_movmem): Likewise.
5300 * config/frv/frv.c (frv_trampoline_init): Likewise.
5301 * config/i386/i386.c (ix86_trampoline_init): Likewise.
5302 (ix86_expand_divmod_libfunc): Likewise.
5303 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
5304 (ia64_expand_compare): Likewise.
5305 (ia64_profile_hook): Likewise.
5306 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
5307 (nonlocal_goto): Likewise.
5308 (restore_stack_nonlocal): Likewise.
5309 * config/m32r/m32r.c (block_move_call): Likewise.
5310 (m32r_trampoline_init): Likewise.
5311 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
5312 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
5313 (m68k_call_m68k_read_tp): Likewise.
5314 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
5315 (microblaze_expand_divide): Likewise.
5316 * config/mips/mips.h (mips_args): Likewise.
5317 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
5318 (MIPS_ICACHE_SYNC): Likewise.
5319 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
5320 (nios2_trampoline_init): Likewise.
5321 * config/pa/pa.c (hppa_tls_call): Likewise.
5322 (pa_trampoline_init): Likewise.
5323 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
5324 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
5325 (expand_strn_compare): Likewise.
5326 (rs6000_generate_compare): Likewise.
5327 (rs6000_expand_float128_convert): Likewise.
5328 (output_profile_hook): Likewise.
5329 (rs6000_trampoline_init): Likewise.
5330 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
5331 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
5332 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
5333 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
5334 (rs6000_generate_compare): Likewise.
5335 (rs6000_expand_float128_convert): Likewise.
5336 (output_profile_hook): Likewise.
5337 (rs6000_trampoline_init): Likewise.
5338 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
5339 * config/sh/sh.c (sh_trampoline_init): Likewise.
5340 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
5341 (sparc_emit_float_lib_cmp): Likewise.
5342 (sparc32_initialize_trampoline): Likewise.
5343 (sparc64_initialize_trampoline): Likewise.
5344 (sparc_profile_hook): Likewise.
5345 * config/spu/spu.c (ea_load_store): Likewise.
5346 * config/spu/spu.md (floatunssidf2): Likewise.
5347 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
5348 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
5349 * config/visium/visium.c (expand_block_move_4): Likewise.
5350 (expand_block_move_2): Likewise.
5351 (expand_block_move_1): Likewise.
5352 (expand_block_set_4): Likewise.
5353 (expand_block_set_2): Likewise.
5354 (expand_block_set_1): Likewise.
5355 (visium_trampoline_init): Likewise.
5356 (visium_profile_hook): Likewise.
5357 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
5358 (xtensa_setup_frame_addresses): Likewise.
5359 (xtensa_trampoline_init): Likewise.
5360 * except.c (sjlj_emit_function_enter): Likewise.
5361 (sjlj_emit_function_exit): Likewise.
5362 * explow.c (allocate_dynamic_stack_space): Likewise.
5363 (probe_stack_range): Likewise.
5364 * expr.c (convert_mode_scalar): Likewise.
5365 * optabs.c (expand_binop): Likewise.
5366 (expand_twoval_binop_libfunc): Likewise.
5367 (expand_unop): Likewise.
5368 (prepare_cmp_insn): Likewise.
5369 (prepare_float_lib_cmp): Likewise.
5370 (expand_float): Likewise.
5371 (expand_fix): Likewise.
5372 (expand_fixed_convert): Likewise.
5373 (maybe_emit_sync_lock_test_and_set): Likewise.
5374 (expand_atomic_compare_and_swap): Likewise.
5375 (expand_mem_thread_fence): Likewise.
5376 (expand_atomic_fetch_op): Likewise.
5378 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
5380 * doc/generic.texi (OpenACC): Adjust URL.
5381 * doc/invoke.texi (C Dialect Options): Ditto.
5383 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
5385 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
5386 predicate for operand 1. Add (m,<S>) constraint.
5387 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
5388 Prevent memory operand 1 with register operand 2.
5390 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
5392 PR rtl-optimization/82024
5393 * combine.c (try_combine): If the combination result is a PARALLEL,
5394 and we only need to retain the SET in there that would be placed
5395 at I2, check that we can place that at I3 instead, before doing so.
5397 2017-09-01 Jakub Jelinek <jakub@redhat.com>
5400 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
5402 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
5403 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
5406 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
5407 Jeff Law <law@redhat.com>
5409 * varasm.c (bss_initializer_p): Do not put constants into .bss
5410 (categorize_decl_for_section): Handle bss_initializer_p returning
5411 false when DECL_INITIAL is NULL.
5413 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5416 * config/s390/s390.c (s390_can_inline_p): New function.
5418 2017-09-01 Jeff Law <law@redhat.com>
5420 PR tree-optimization/82052
5421 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
5422 Always initialize the returned slot after a hash table miss
5423 when INSERT is true.
5425 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
5427 * config/s390/s390.md (mem_signal_fence): Remove.
5428 * doc/md.texi (mem_signal_fence): Remove.
5429 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
5431 * target-insns.def (mem_signal_fence): Remove.
5433 2017-09-01 Jakub Jelinek <jakub@redhat.com>
5436 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
5439 * asan.c (create_odr_indicator): Strip name encoding from assembler
5440 name before appending it after __odr_asan_.
5442 2017-09-01 Martin Liska <mliska@suse.cz>
5444 PR tree-optimization/82059
5445 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
5446 frequency only when an edge is redirected.
5448 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5450 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
5451 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
5452 (arc_conditional_register_usage): Remove ARC600 lp_count
5454 (arc_file_start): Emit Tag_ARC_CPU_variation.
5455 (arc_can_use_doloop_p): New conditions to use ZOLs.
5456 (hwloop_fail): New function.
5457 (hwloop_optimize): Likewise.
5458 (hwloop_pattern_reg): Likewise.
5459 (arc_doloop_hooks): New struct, to be used with reorg_loops.
5460 (arc_reorg_loops): New function, calls reorg_loops.
5461 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
5462 (arc600_corereg_hazard): Remove ZOL checking, case handled by
5464 (arc_loop_hazard): Remove function, functionality moved into
5466 (arc_hazard): Remove arc_loop_hazard call.
5467 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
5468 into hwloop_optimize.
5469 (arc_label_align): Remove ZOL handling.
5470 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
5471 * config/arc/arc.md (doloop_begin): Remove pattern.
5472 (doloop_begin_i): Likewise.
5473 (doloop_end_i): Likewise.
5474 (doloop_fallback): Likewise.
5475 (doloop_fallback_m): Likewise.
5476 (doloop_end): Reimplement expand.
5477 (arc_lp): New pattern for LP instruction.
5478 (loop_end): New pattern.
5479 (loop_fail): Likewise.
5480 (decrement_and_branch_until_zero): Likewise.
5481 * config/arc/arc.opt (mlpc-width): New option.
5482 * doc/invoke.texi (mlpc-width): Document option.
5484 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5486 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
5487 (arc_ccfsm_advance): Fix checking for delay slots.
5488 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
5490 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5492 * config/arc/arc.md (movqi_insn): Add stores to save constant long
5494 (movhi_insn): Update store instruction constraint which are saving
5495 6-bit short immediates.
5496 (movsi_insn): Consider also short scaled load operations.
5497 (zero_extendhisi2_i): Use Usd constraint instead of T.
5498 (extendhisi2_i): Add q constraint.
5499 (arc_clzsi2): Add type and length attributes.
5500 (arc_ctzsi2): Likewise.
5501 * config/arc/constraints.md (Usc): Update constraint, the
5502 assembler can parse two relocations for a single instruction.
5504 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5506 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
5507 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
5509 2017-08-31 Olivier Hainque <hainque@adacore.com>
5511 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
5512 match as powerpc-wrs-vxworks*.
5514 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
5516 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
5517 register constraint for by-element operand.
5518 (aarch64_mls_elt_merge<mode>): Likewise.
5520 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5522 * config/arc/arc.c (arc_can_follow_jump): Check for short
5525 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5527 * config.gcc: Use g.opt for arc.
5528 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
5529 functionality moved to ...
5530 (legitimate_scaled_address_p): New function, ...here.
5531 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
5532 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
5533 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
5535 (arc_override_options): Handle G option.
5536 (arc_output_pic_addr_const): Correct function definition.
5537 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
5538 (arc_decl_anon_ns_mem_p): Delete.
5539 (arc_in_small_data_p): Overhaul this function to take into
5540 consideration the value given via G option.
5541 (arc_rewrite_small_data_1): Renamed and corrected old
5542 arc_rewrite_small_data function.
5543 (arc_rewrite_small_data): New function.
5544 (small_data_pattern): Don't use pic_offset_table_rtx.
5545 * config/arc/arc.h (CC1_SPEC): Recognize G option.
5546 * config/arc/simdext.md (movmisalignv2hi): Use
5547 prepare_move_operands function.
5549 (movmisalign*): Likewise.
5550 * doc/invoke.texi (ARC options): Document -G option.
5552 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5554 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
5556 * config/arc/arc.c (arc_print_operand): Output scalled address for
5557 sdata whenever is possible.
5558 (arc_in_small_data_p): Allow sdata for 64bit datum when double
5559 load/stores are available.
5560 (compact_sda_memory_operand): Check for the alignment required by
5561 code density instructions.
5562 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
5564 * config/arc/constraints.md (Usd): Update constraint.
5565 (Us0): New constraint.
5566 (Usc): Update constraint.
5568 2017-08-31 Richard Biener <rguenther@suse.de>
5571 * dwarf2out.c (dwarf2out_early_global_decl): Process each
5574 2017-08-31 Tamar Christina <tamar.christina@arm.com>
5576 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
5577 Resize type_signature.
5579 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5580 Alan Hayward <alan.hayward@arm.com>
5581 David Sherwood <david.sherwood@arm.com>
5583 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
5584 subregs whose inner modes can be stored in GPRs.
5585 (aarch64_classify_index): Likewise.
5587 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5588 Alan Hayward <alan.hayward@arm.com>
5589 David Sherwood <david.sherwood@arm.com>
5591 * config/aarch64/iterators.md (V_cmp_result): Rename to...
5592 (V_INT_EQUIV): ...this.
5593 (v_cmp_result): Rename to...
5594 (v_int_equiv): ...this.
5595 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
5596 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
5597 (copysign<mode>3): Likewise.
5598 (aarch64_simd_bsl<mode>_internal): Likewise.
5599 (aarch64_simd_bsl<mode>): Likewise.
5600 (vec_cmp<mode><mode>): Likewise.
5601 (vcond<mode><mode>): Likewise.
5602 (vcond<v_cmp_mixed><mode>): Likewise.
5603 (vcondu<mode><v_cmp_mixed>): Likewise.
5604 (aarch64_cm<optab><mode>): Likewise.
5605 (aarch64_cmtst<mode>): Likewise.
5606 (aarch64_fac<optab><mode>): Likewise.
5607 (vec_perm_const<mode>): Likewise.
5608 (vcond_mask_<mode><v_cmp_result>): Rename to...
5609 (vcond_mask_<mode><v_int_equiv>): ...this.
5610 (vec_cmp<mode><v_cmp_result>): Rename to...
5611 (vec_cmp<mode><v_int_equiv>): ...this.
5613 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
5614 Alan Hayward <alan.hayward@arm.com>
5615 David Sherwood <david.sherwood@arm.com>
5617 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
5619 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
5620 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
5621 (UNSPEC_LD4_DREG): New unspecs.
5622 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
5623 (aarch64_ld2<mode>_dreg_be): Replace with...
5624 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
5626 (aarch64_ld3<mode>_dreg_le)
5627 (aarch64_ld3<mode>_dreg_be): Replace with...
5628 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
5630 (aarch64_ld4<mode>_dreg_le)
5631 (aarch64_ld4<mode>_dreg_be): Replace with...
5632 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
5635 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5637 PR tree-optimization/81987
5638 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
5639 insert an initializer in a location not dominated by the stride
5642 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5644 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
5645 on the entire header of the finally block in the fallthru case.
5647 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
5649 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
5651 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
5653 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
5654 rs6000_emit_move_from_cr and call renamed function.
5655 (rs6000_emit_prologue): Call renamed functions.
5656 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
5657 movesi_from_cr, remove volatile CRs.
5659 2017-08-30 Jon Beniston <jon@beniston.com>
5660 Richard Biener <rguenther@suse.de>
5662 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
5663 of VECTOR_MODE_P check.
5664 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
5665 element vector types.
5667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5669 * df.h (df_read_modify_subreg_p): Remove in favor of...
5670 * rtl.h (read_modify_subreg_p): ...this new function. Take a
5671 const_rtx instead of an rtx.
5672 * cprop.c (local_cprop_find_used_regs): Update accordingly.
5673 * df-problems.c (df_word_lr_mark_ref): Likewise.
5674 * ira-lives.c (mark_pseudo_reg_live): Likewise.
5675 (mark_pseudo_reg_dead): Likewise.
5676 (mark_ref_dead): Likewise.
5677 * reginfo.c (init_subregs_of_mode): Likewise.
5678 * sched-deps.c (sched_analyze_1): Likewise.
5679 * df-scan.c (df_def_record_1): Likewise.
5680 (df_uses_record): Likewise.
5681 (df_read_modify_subreg_p): Remove in favor of...
5682 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
5683 const_rtx instead of an rtx.
5685 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5686 Alan Hayward <alan.hayward@arm.com>
5687 David Sherwood <david.sherwood@arm.com>
5689 * rtl.h (partial_subreg_p): New function.
5690 * caller-save.c (save_call_clobbered_regs): Use it.
5691 * calls.c (expand_call): Likewise.
5692 * combine.c (combinable_i3pat): Likewise.
5693 (simplify_set): Likewise.
5694 (make_extraction): Likewise.
5695 (make_compound_operation_int): Likewise.
5696 (gen_lowpart_or_truncate): Likewise.
5697 (force_to_mode): Likewise.
5698 (make_field_assignment): Likewise.
5699 (reg_truncated_to_mode): Likewise.
5700 (record_truncated_value): Likewise.
5701 (move_deaths): Likewise.
5702 * cse.c (record_jump_cond): Likewise.
5703 (cse_insn): Likewise.
5704 * cselib.c (cselib_lookup_1): Likewise.
5705 * expmed.c (extract_bit_field_using_extv): Likewise.
5706 * function.c (assign_parm_setup_reg): Likewise.
5707 * ifcvt.c (noce_convert_multiple_sets): Likewise.
5708 * ira-build.c (create_insn_allocnos): Likewise.
5709 * lra-coalesce.c (merge_pseudos): Likewise.
5710 * lra-constraints.c (match_reload): Likewise.
5711 (simplify_operand_subreg): Likewise.
5712 (curr_insn_transform): Likewise.
5713 * lra-lives.c (process_bb_lives): Likewise.
5714 * lra.c (new_insn_reg): Likewise.
5715 (lra_substitute_pseudo): Likewise.
5716 * regcprop.c (mode_change_ok): Likewise.
5717 (maybe_mode_change): Likewise.
5718 (copyprop_hardreg_forward_1): Likewise.
5719 * reload.c (push_reload): Likewise.
5720 (find_reloads): Likewise.
5721 (find_reloads_subreg_address): Likewise.
5722 * reload1.c (alter_reg): Likewise.
5723 (eliminate_regs_1): Likewise.
5724 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5726 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
5728 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
5729 back to if statements, including unpack.
5731 2017-08-30 Martin Liska <mliska@suse.cz>
5734 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
5736 (func_checker::compare_asm_inputs_outputs): ... this function.
5737 (func_checker::compare_gimple_asm): Use the function to compare
5738 also ASM constrains.
5739 * ipa-icf-gimple.h: Rename the function.
5741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5742 Alan Hayward <alan.hayward@arm.com>
5743 David Sherwood <david.sherwood@arm.com>
5745 * coretypes.h (complex_mode): New type.
5746 * gdbhooks.py (build_pretty_printer): Handle it.
5747 * machmode.h (complex_mode): New class.
5748 (complex_mode::includes_p): New function.
5749 (is_complex_int_mode): Likewise.
5750 (is_complex_float_mode): Likewise.
5751 * genmodes.c (get_mode_class): Handle complex mode classes.
5752 * function.c (expand_function_end): Use is_complex_int_mode.
5754 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5755 Alan Hayward <alan.hayward@arm.com>
5756 David Sherwood <david.sherwood@arm.com>
5758 * coretypes.h (scalar_mode_pod): New typedef.
5759 * gdbhooks.py (build_pretty_printer): Handle it.
5760 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
5761 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
5762 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
5763 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
5766 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5767 Alan Hayward <alan.hayward@arm.com>
5768 David Sherwood <david.sherwood@arm.com>
5770 * machmode.h (mode_for_vector): Take a scalar_mode instead
5772 * stor-layout.c (mode_for_vector): Likewise.
5773 * explow.c (promote_mode): Use as_a <scalar_mode>.
5774 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
5776 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5777 Alan Hayward <alan.hayward@arm.com>
5778 David Sherwood <david.sherwood@arm.com>
5780 * target.def (preferred_simd_mode): Take a scalar_mode
5781 instead of a machine_mode.
5782 * targhooks.h (default_preferred_simd_mode): Likewise.
5783 * targhooks.c (default_preferred_simd_mode): Likewise.
5784 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
5785 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
5786 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
5787 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
5788 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
5789 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
5790 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
5791 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
5793 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
5794 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
5795 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
5796 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
5797 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
5798 * doc/tm.texi: Regenerate.
5799 * optabs-query.c (can_vec_mask_load_store_p): Return false for
5802 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5803 Alan Hayward <alan.hayward@arm.com>
5804 David Sherwood <david.sherwood@arm.com>
5806 * target.def (scalar_mode_supported_p): Take a scalar_mode
5807 instead of a machine_mode.
5808 * targhooks.h (default_scalar_mode_supported_p): Likewise.
5809 * targhooks.c (default_scalar_mode_supported_p): Likewise.
5810 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
5811 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
5812 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
5813 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
5814 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
5815 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
5816 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
5817 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
5818 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
5819 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
5820 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
5821 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
5823 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
5824 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
5825 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
5826 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
5827 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
5829 * doc/tm.texi: Regenerate.
5831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5832 Alan Hayward <alan.hayward@arm.com>
5833 David Sherwood <david.sherwood@arm.com>
5835 * coretypes.h (opt_scalar_mode): New typedef.
5836 * gdbhooks.py (build_pretty_printers): Handle it.
5837 * machmode.h (mode_iterator::get_2xwider): Add overload for
5839 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
5841 * expr.c (convert_mode_scalar): Likewise.
5842 * omp-low.c (omp_clause_aligned_alignment): Likewise.
5843 * optabs.c (expand_float): Likewise.
5844 (expand_fix): Likewise.
5845 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5847 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5848 Alan Hayward <alan.hayward@arm.com>
5849 David Sherwood <david.sherwood@arm.com>
5851 * optabs.c (expand_float): Explicitly check for scalars before
5852 using a branching expansion.
5853 (expand_fix): Likewise.
5855 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5856 Alan Hayward <alan.hayward@arm.com>
5857 David Sherwood <david.sherwood@arm.com>
5859 * expr.c (convert_mode): Split scalar handling out into...
5860 (convert_mode_scalar): ...this new function. Treat the modes
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 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
5869 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
5871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5872 Alan Hayward <alan.hayward@arm.com>
5873 David Sherwood <david.sherwood@arm.com>
5875 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
5876 rather than a machine_mode.
5877 (fixed_from_string): Likewise.
5878 (fixed_convert): Likewise.
5879 (fixed_convert_from_int): Likewise.
5880 (fixed_convert_from_real): Likewise.
5881 (real_convert_from_fixed): Likewise.
5882 * fixed-value.c (fixed_from_double_int): Likewise.
5883 (fixed_from_string): Likewise.
5884 (fixed_convert): Likewise.
5885 (fixed_convert_from_int): Likewise.
5886 (fixed_convert_from_real): Likewise.
5887 (real_convert_from_fixed): Likewise.
5888 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
5890 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5891 Alan Hayward <alan.hayward@arm.com>
5892 David Sherwood <david.sherwood@arm.com>
5894 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
5895 of separate mode class checks. Do not allow vector modes here.
5896 (immed_wide_int_const): Use as_a <scalar_mode>.
5897 * explow.c (trunc_int_for_mode): Likewise.
5898 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
5899 (wi::shwi): Likewise.
5900 (wi::min_value): Likewise.
5901 (wi::max_value): Likewise.
5902 * dwarf2out.c (loc_descriptor): Likewise.
5903 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
5906 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5907 Alan Hayward <alan.hayward@arm.com>
5908 David Sherwood <david.sherwood@arm.com>
5910 * tree.h (SCALAR_TYPE_MODE): New macro.
5911 * expr.c (expand_expr_addr_expr_1): Use it.
5912 (expand_expr_real_2): Likewise.
5913 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
5914 (fold_convert_const_fixed_from_int): Likewise.
5915 (fold_convert_const_fixed_from_real): Likewise.
5916 (native_encode_fixed): Likewise
5917 (native_encode_complex): Likewise
5918 (native_encode_vector): Likewise.
5919 (native_interpret_fixed): Likewise.
5920 (native_interpret_real): Likewise.
5921 (native_interpret_complex): Likewise.
5922 (native_interpret_vector): Likewise.
5923 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
5924 (simd_clone_adjust_argument_types): Likewise.
5925 (simd_clone_init_simd_arrays): Likewise.
5926 (simd_clone_adjust): Likewise.
5927 * stor-layout.c (layout_type): Likewise.
5928 * tree.c (build_minus_one_cst): Likewise.
5929 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
5930 * tree-inline.c (estimate_move_cost): Likewise.
5931 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
5932 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
5933 (vectorizable_reduction): Likewise.
5934 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
5935 (vect_recog_mixed_size_cond_pattern): Likewise.
5936 (check_bool_pattern): Likewise.
5937 (adjust_bool_pattern): Likewise.
5938 (search_type_for_mask_1): Likewise.
5939 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
5940 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
5941 (vectorizable_load): Likewise.
5942 (vectorizable_store): Likewise.
5943 * ubsan.c (ubsan_encode_value): Likewise.
5944 * varasm.c (output_constant): Likewise.
5946 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5947 Alan Hayward <alan.hayward@arm.com>
5948 David Sherwood <david.sherwood@arm.com>
5950 * coretypes.h (scalar_mode): New class.
5951 * machmode.h (scalar_mode): Likewise.
5952 (scalar_mode::includes_p): New function.
5953 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
5954 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
5955 * genmodes.c (get_mode_class): Handle remaining scalar modes.
5956 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
5957 * expmed.c (store_bit_field_1): Likewise.
5958 (extract_bit_field_1): Likewise.
5959 * expr.c (write_complex_part): Likewise.
5960 (read_complex_part): Likewise.
5961 (emit_move_complex_push): Likewise.
5962 (expand_expr_real_2): Likewise.
5963 * function.c (assign_parm_setup_reg): Likewise.
5964 (assign_parms_unsplit_complex): Likewise.
5965 * optabs.c (expand_binop): Likewise.
5966 * rtlanal.c (subreg_get_info): Likewise.
5967 * simplify-rtx.c (simplify_immed_subreg): Likewise.
5968 * varasm.c (output_constant_pool_2): Likewise.
5970 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5971 Alan Hayward <alan.hayward@arm.com>
5972 David Sherwood <david.sherwood@arm.com>
5974 * expmed.c (extract_high_half): Use scalar_int_mode and remove
5976 (expmed_mult_highpart_optab): Likewise.
5977 (expmed_mult_highpart): Likewise.
5979 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
5980 Alan Hayward <alan.hayward@arm.com>
5981 David Sherwood <david.sherwood@arm.com>
5983 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
5984 instead of a machine_mode.
5985 (builtin_memset_read_str): Likewise.
5986 * builtins.c (c_readstr): Likewise.
5987 (builtin_memcpy_read_str): Likewise.
5988 (builtin_strncpy_read_str): Likewise.
5989 (builtin_memset_read_str): Likewise.
5990 (builtin_memset_gen_str): Likewise.
5991 (expand_builtin_signbit): Use scalar_int_mode for local variables.
5992 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
5993 instead of a machine_mode.
5994 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
5996 (make_extraction): Likewise.
5997 (try_widen_shift_mode): Take and return scalar_int_modes instead
5999 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
6000 a scalar_int_mode instead of a machine_mode.
6001 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
6002 (avr_addr_space_pointer_mode): Likewise.
6003 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
6004 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
6005 (msp430_unwind_word_mode): Likewise.
6006 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
6007 (spu_addr_space_pointer_mode): Likewise.
6008 (spu_addr_space_address_mode): Likewise.
6009 (spu_libgcc_cmp_return_mode): Likewise.
6010 (spu_libgcc_shift_count_mode): Likewise.
6011 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
6012 (rl78_addr_space_pointer_mode): Likewise.
6013 (fl78_unwind_word_mode): Likewise.
6014 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
6016 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
6017 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
6018 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
6019 (mips_valid_pointer_mode): Likewise.
6020 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
6021 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
6022 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
6024 (ft32_addr_space_address_mode): Likewise.
6025 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
6026 scalar_int_mode instead of a machine_mode.
6027 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
6029 (m32c_addr_space_address_mode): Likewise.
6030 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
6031 (rs6000_eh_return_filter_mode): Likewise.
6032 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
6033 (rs6000_eh_return_filter_mode): Likewise.
6034 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
6035 (s390_libgcc_shift_count_mode): Likewise.
6036 (s390_unwind_word_mode): Likewise.
6037 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
6039 * target.def (mode_rep_extended): Likewise.
6040 (valid_pointer_mode): Likewise.
6041 (addr_space.valid_pointer_mode): Likewise.
6042 (eh_return_filter_mode): Return a scalar_int_mode rather than
6044 (libgcc_cmp_return_mode): Likewise.
6045 (libgcc_shift_count_mode): Likewise.
6046 (unwind_word_mode): Likewise.
6047 (addr_space.pointer_mode): Likewise.
6048 (addr_space.address_mode): Likewise.
6049 * doc/tm.texi: Regenerate.
6050 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
6052 (do_jump): Use scalar_int_mode for local variables.
6053 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
6054 rather than a machine_mode.
6055 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
6056 (scompare_loc_descriptor_wide): Likewise.
6057 (scompare_loc_descriptor_narrow): Likewise.
6058 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
6060 * except.c (sjlj_emit_dispatch_table): Likewise.
6061 (expand_builtin_eh_copy_values): Likewise.
6062 * explow.c (convert_memory_address_addr_space_1): Likewise.
6063 Take a scalar_int_mode rather than a machine_mode.
6064 (convert_memory_address_addr_space): Take a scalar_int_mode rather
6065 than a machine_mode.
6066 (memory_address_addr_space): Use scalar_int_mode for local variables.
6067 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
6068 rather than a machine_mode.
6069 * expmed.c (mask_rtx): Likewise.
6070 (init_expmed_one_conv): Likewise.
6071 (expand_mult_highpart_adjust): Likewise.
6072 (extract_high_half): Likewise.
6073 (expmed_mult_highpart_optab): Likewise.
6074 (expmed_mult_highpart): Likewise.
6075 (expand_smod_pow2): Likewise.
6076 (expand_sdiv_pow2): Likewise.
6077 (emit_store_flag_int): Likewise.
6078 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
6080 (extract_low_bits): Likewise.
6081 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
6083 * expr.c (pieces_addr::adjust): Likewise.
6084 (can_store_by_pieces): Likewise.
6085 (store_by_pieces): Likewise.
6086 (clear_by_pieces_1): Likewise.
6087 (expand_expr_addr_expr_1): Likewise.
6088 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
6089 (expand_expr_real_1): Likewise.
6090 (try_casesi): Likewise.
6091 * final.c (shorten_branches): Likewise.
6092 * fold-const.c (fold_convert_const_int_from_fixed): Change the
6093 type of "mode" to machine_mode.
6094 * internal-fn.c (expand_arith_overflow_result_store): Take a
6095 scalar_int_mode rather than a machine_mode.
6096 (expand_mul_overflow): Use scalar_int_mode for local variables.
6097 * loop-doloop.c (doloop_modify): Likewise.
6098 (doloop_optimize): Likewise.
6099 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
6100 than a machine_mode.
6101 (expand_doubleword_shift_condmove): Likewise.
6102 (expand_doubleword_shift): Likewise.
6103 (expand_doubleword_clz): Likewise.
6104 (expand_doubleword_popcount): Likewise.
6105 (expand_doubleword_parity): Likewise.
6106 (expand_absneg_bit): Use scalar_int_mode for local variables.
6107 (prepare_float_lib_cmp): Likewise.
6108 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
6109 rather than a machine_mode.
6110 (convert_memory_address_addr_space): Likewise.
6111 (get_mode_bounds): Likewise.
6112 (get_address_mode): Return a scalar_int_mode rather than a
6114 * rtlanal.c (get_address_mode): Likewise.
6115 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
6116 than a machine_mode.
6117 * targhooks.c (default_mode_rep_extended): Likewise.
6118 (default_valid_pointer_mode): Likewise.
6119 (default_addr_space_valid_pointer_mode): Likewise.
6120 (default_eh_return_filter_mode): Return a scalar_int_mode rather
6121 than a machine_mode.
6122 (default_libgcc_cmp_return_mode): Likewise.
6123 (default_libgcc_shift_count_mode): Likewise.
6124 (default_unwind_word_mode): Likewise.
6125 (default_addr_space_pointer_mode): Likewise.
6126 (default_addr_space_address_mode): Likewise.
6127 * targhooks.h (default_eh_return_filter_mode): Likewise.
6128 (default_libgcc_cmp_return_mode): Likewise.
6129 (default_libgcc_shift_count_mode): Likewise.
6130 (default_unwind_word_mode): Likewise.
6131 (default_addr_space_pointer_mode): Likewise.
6132 (default_addr_space_address_mode): Likewise.
6133 (default_mode_rep_extended): Take a scalar_int_mode rather than
6135 (default_valid_pointer_mode): Likewise.
6136 (default_addr_space_valid_pointer_mode): Likewise.
6137 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
6139 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
6140 rather than a machine_mode.
6141 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
6142 for local variables.
6143 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6144 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
6145 than a machine_mode.
6147 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6148 Alan Hayward <alan.hayward@arm.com>
6149 David Sherwood <david.sherwood@arm.com>
6151 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
6152 the mode argument to scalar_int_mode.
6153 (do_jump_by_parts_zero_rtx): Likewise.
6154 (do_jump_by_parts_equality_rtx): Likewise.
6155 (do_jump_by_parts_greater): Take a mode argument.
6156 (do_jump_by_parts_equality): Likewise.
6157 (do_jump_1): Update calls accordingly.
6159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6160 Alan Hayward <alan.hayward@arm.com>
6161 David Sherwood <david.sherwood@arm.com>
6163 * is-a.h (safe_dyn_cast): New function.
6164 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
6165 (jump_table_for_label): Likewise.
6166 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
6167 instead of an rtx_insn *.
6168 (shorten_branches): Use dyn_cast instead of LABEL_P and
6169 JUMP_TABLE_DATA_P. Use jump_table_for_label and
6170 rtx_jump_table_data::get_data_mode.
6171 (final_scan_insn): Likewise.
6173 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6174 Alan Hayward <alan.hayward@arm.com>
6175 David Sherwood <david.sherwood@arm.com>
6177 * combine.c (try_combine): Use is_a <scalar_int_mode> when
6178 trying to combine a full-register integer set with a subreg
6181 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6182 Alan Hayward <alan.hayward@arm.com>
6183 David Sherwood <david.sherwood@arm.com>
6185 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
6186 that is always either address_mode or pointer_mode.
6188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6189 Alan Hayward <alan.hayward@arm.com>
6190 David Sherwood <david.sherwood@arm.com>
6192 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
6193 when the two are known to be equal.
6195 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6196 Alan Hayward <alan.hayward@arm.com>
6197 David Sherwood <david.sherwood@arm.com>
6199 * simplify-rtx.c (simplify_const_unary_operation): Use
6200 is_a <scalar_int_mode> instead of checking for a nonzero
6201 precision. Forcibly convert op_mode to a scalar_int_mode
6202 in that case. More clearly differentiate the operand and
6203 result modes and use the former when deciding what the value
6204 of a count-bits operation should be. Use is_int_mode instead
6205 of checking for a MODE_INT. Remove redundant check for whether
6206 this mode has a zero precision.
6208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6209 Alan Hayward <alan.hayward@arm.com>
6210 David Sherwood <david.sherwood@arm.com>
6212 * optabs.c (widen_leading): Change the type of the mode argument
6213 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
6214 (widen_bswap): Likewise.
6215 (expand_parity): Likewise.
6216 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
6217 (expand_ffs): Likewise.
6218 (epand_unop): Check for scalar integer modes before calling the
6221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6222 Alan Hayward <alan.hayward@arm.com>
6223 David Sherwood <david.sherwood@arm.com>
6225 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
6227 (expand_expr_real_1): Update call accordingly.
6229 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6230 Alan Hayward <alan.hayward@arm.com>
6231 David Sherwood <david.sherwood@arm.com>
6233 * expmed.c (store_bit_field_using_insv): Add op0_mode and
6234 value_mode arguments. Use scalar_int_mode internally.
6235 (store_bit_field_1): Rename the new integer mode from imode
6236 to op0_mode and use it instead of GET_MODE (op0). Update calls
6237 to store_split_bit_field, store_bit_field_using_insv and
6238 store_fixed_bit_field.
6239 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
6240 Use scalar_int_mode internally. Use a bit count rather than a mode
6241 when calculating the largest bit size for get_best_mode.
6242 Update calls to store_split_bit_field and store_fixed_bit_field_1.
6243 (store_fixed_bit_field_1): Add mode and value_mode arguments.
6244 Remove assertion that OP0 has a scalar integer mode.
6245 (store_split_bit_field): Add op0_mode and value_mode arguments.
6246 Update calls to extract_fixed_bit_field.
6247 (extract_bit_field_using_extv): Add an op0_mode argument.
6248 Use scalar_int_mode internally.
6249 (extract_bit_field_1): Rename the new integer mode from imode to
6250 op0_mode and use it instead of GET_MODE (op0). Update calls to
6251 extract_split_bit_field, extract_bit_field_using_extv and
6252 extract_fixed_bit_field.
6253 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
6254 to extract_split_bit_field and extract_fixed_bit_field_1.
6255 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
6256 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
6258 (extract_split_bit_field): Add an op0_mode argument. Update call
6259 to extract_fixed_bit_field.
6261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6262 Alan Hayward <alan.hayward@arm.com>
6263 David Sherwood <david.sherwood@arm.com>
6265 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
6266 * explow.c (hard_function_value): Likewise.
6267 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
6268 convert_to_mode call outside the loop.
6269 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
6270 for the mode iterator. Require the mode specified by max_pieces
6272 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
6274 (copy_blkmode_to_reg): Likewise.
6275 (set_storage_via_setmem): Likewise.
6276 * optabs.c (prepare_cmp_insn): Likewise.
6277 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6278 * stor-layout.c (finish_bitfield_representative): Likewise.
6280 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6281 Alan Hayward <alan.hayward@arm.com>
6282 David Sherwood <david.sherwood@arm.com>
6284 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
6285 * expr.c (convert_move): Use them.
6286 (convert_modes): Likewise.
6287 (store_expr_with_bounds): Likewise.
6289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6290 Alan Hayward <alan.hayward@arm.com>
6291 David Sherwood <david.sherwood@arm.com>
6293 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
6294 parameter for the mode of "x". Remove the "known_x", "known_mode"
6295 and "known_ret" arguments. Change the type of the mode argument
6297 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
6298 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
6299 (reg_num_sign_bit_copies_for_combine): Likewise.
6300 * rtlanal.c (nonzero_bits1): Likewise.
6301 (num_sign_bit_copies1): Likewise.
6302 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
6303 (reg_num_sign_bit_copies_general): Likewise.
6304 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
6305 (reg_nonzero_bits_general): Likewise.
6307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6308 Alan Hayward <alan.hayward@arm.com>
6309 David Sherwood <david.sherwood@arm.com>
6311 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
6312 than in subroutines. Return 1 for non-integer modes.
6313 (cached_num_sign_bit_copies): Change the type of the mode parameter
6315 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
6316 classes. Handle CONST_INT_P first and then check whether X also
6317 has a scalar integer mode. Check the same thing for inner registers
6318 of a SUBREG and for values that are being extended or truncated.
6320 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6321 Alan Hayward <alan.hayward@arm.com>
6322 David Sherwood <david.sherwood@arm.com>
6324 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
6325 in subroutines. Return the mode mask for non-integer modes.
6326 (cached_nonzero_bits): Change the type of the mode parameter
6328 (nonzero_bits1): Likewise. Remove early exit for other mode
6329 classes. Handle CONST_INT_P first and then check whether X
6330 also has a scalar integer mode.
6332 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6333 Alan Hayward <alan.hayward@arm.com>
6334 David Sherwood <david.sherwood@arm.com>
6336 * expr.c (widest_int_mode_for_size): Make the comment match the code.
6337 Return a scalar_int_mode and assert that the size is greater than
6339 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
6340 (op_by_pieces_d::op_by_pieces_d): Likewise.
6341 (op_by_pieces_d::run): Likewise.
6342 (can_store_by_pieces): Likewise.
6344 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6345 Alan Hayward <alan.hayward@arm.com>
6346 David Sherwood <david.sherwood@arm.com>
6348 * combine.c (extract_left_shift): Add a mode argument and update
6350 (make_compound_operation_int): Change the type of the mode parameter
6351 to scalar_int_mode and update the call to extract_left_shift.
6353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6354 Alan Hayward <alan.hayward@arm.com>
6355 David Sherwood <david.sherwood@arm.com>
6357 * combine.c (simplify_and_const_int): Change the type of the mode
6358 parameter to scalar_int_mode.
6359 (simplify_and_const_int_1): Likewise. Update recursive call.
6361 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6362 Alan Hayward <alan.hayward@arm.com>
6363 David Sherwood <david.sherwood@arm.com>
6365 * combine.c (simplify_compare_const): Check that the mode is a
6366 scalar_int_mode (rather than VOIDmode) before testing its
6368 (simplify_comparison): Move COMPARISON_P handling out of the
6369 loop and restrict the latter part of the loop to scalar_int_modes.
6370 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
6371 and when considering SUBREG_REGs. Use is_int_mode instead of
6372 checking GET_MODE_CLASS against MODE_INT.
6374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6375 Alan Hayward <alan.hayward@arm.com>
6376 David Sherwood <david.sherwood@arm.com>
6378 * combine.c (try_widen_shift_mode): Move check for equal modes to...
6379 (simplify_shift_const_1): ...here. Use scalar_int_mode for
6380 shift_unit_mode and for modes involved in scalar shifts.
6382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6383 Alan Hayward <alan.hayward@arm.com>
6384 David Sherwood <david.sherwood@arm.com>
6386 * combine.c (force_int_to_mode): New function, split out from...
6387 (force_to_mode): ...here. Keep xmode up-to-date and use it
6388 instead of GET_MODE (x).
6390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6391 Alan Hayward <alan.hayward@arm.com>
6392 David Sherwood <david.sherwood@arm.com>
6394 * optabs-query.h (extraction_insn::struct_mode): Change type to
6395 opt_scalar_int_mode and update comment.
6396 (extraction_insn::field_mode): Change type to scalar_int_mode.
6397 (extraction_insn::pos_mode): Likewise.
6398 * combine.c (make_extraction): Update accordingly.
6399 * optabs-query.c (get_traditional_extraction_insn): Likewise.
6400 (get_optab_extraction_insn): Likewise.
6401 * recog.c (simplify_while_replacing): Likewise.
6402 * expmed.c (narrow_bit_field_mem): Change the type of the mode
6403 parameter to opt_scalar_int_mode.
6405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6406 Alan Hayward <alan.hayward@arm.com>
6407 David Sherwood <david.sherwood@arm.com>
6409 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
6410 to a scalar_int_mode instead of a machine_mode.
6411 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
6412 (get_best_mode): Return a boolean and use a pointer argument to store
6413 the selected mode. Replace the limit mode parameter with a bit limit.
6414 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
6415 for the values returned by bit_field_mode_iterator::next_mode.
6416 (store_bit_field): Update call to get_best_mode.
6417 (store_fixed_bit_field): Likewise.
6418 (extract_fixed_bit_field): Likewise.
6419 * expr.c (optimize_bitfield_assignment_op): Likewise.
6420 * fold-const.c (optimize_bit_field_compare): Likewise.
6421 (fold_truth_andor_1): Likewise.
6422 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
6423 Update for new type of m_mode.
6424 (get_best_mode): As above.
6426 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6427 Alan Hayward <alan.hayward@arm.com>
6428 David Sherwood <david.sherwood@arm.com>
6430 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
6431 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
6432 (store_bit_field): Check is_a <scalar_int_mode> before calling
6433 strict_volatile_bitfield_p.
6434 (extract_bit_field): Likewise.
6436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6437 Alan Hayward <alan.hayward@arm.com>
6438 David Sherwood <david.sherwood@arm.com>
6440 * target.def (cstore_mode): Return a scalar_int_mode.
6441 * doc/tm.texi: Regenerate.
6442 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
6443 * targhooks.h (default_cstore_mode): Likewise.
6444 * targhooks.c (default_cstore_mode): Likewise, using a forced
6446 * expmed.c (emit_cstore): Expect the target of the cstore to be
6449 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6450 Alan Hayward <alan.hayward@arm.com>
6451 David Sherwood <david.sherwood@arm.com>
6453 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
6455 (niter_desc): Likewise mode.
6456 (iv_analyze): Add a mode parameter.
6458 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
6459 and change its type to scalar_int_mode.
6460 * loop-iv.c: Update commentary at head of file.
6461 (iv_constant): Pass the mode paraeter before the rtx it describes
6462 and change its type to scalar_int_mode. Remove VOIDmode handling.
6463 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
6464 (iv_extend): Likewise.
6465 (shorten_into_mode): Likewise.
6466 (iv_add): Use scalar_int_mode.
6467 (iv_mult): Likewise.
6468 (iv_shift): Likewise.
6469 (canonicalize_iv_subregs): Likewise.
6470 (get_biv_step_1): Pass the outer_mode parameter before the rtx
6471 it describes and change its mode to scalar_int_mode. Also change
6472 the type of the returned inner_mode to scalar_int_mode.
6473 (get_biv_step): Likewise, turning outer_mode from a pointer
6474 into a direct parameter. Update call to get_biv_step_1.
6475 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
6476 iv_constant and get_biv_step.
6477 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
6478 and change its type to scalar_int_mode. Don't initialise iv->mode
6479 to VOIDmode and remove later checks for its still being VOIDmode.
6480 Update calls to iv_analyze_op and iv_analyze_expr. Check
6481 is_a <scalar_int_mode> when changing the mode under consideration.
6482 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
6483 Update call to iv_analyze_expr.
6484 (iv_analyze_op): Add a mode parameter. Reject subregs whose
6485 inner register is not also a scalar_int_mode. Update call to
6487 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
6488 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
6489 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
6490 separate mode class checks. Update calls to iv_analyze. Remove
6491 fix-up of VOIDmodes after iv_analyze_biv.
6492 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
6493 don't have a scalar_int_mode. Update call to biv_p.
6495 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6496 Alan Hayward <alan.hayward@arm.com>
6497 David Sherwood <david.sherwood@arm.com>
6499 * cfgexpand.c (convert_debug_memory_address): Use
6500 as_a <scalar_int_mode>.
6501 * combine.c (expand_compound_operation): Likewise.
6502 (make_extraction): Likewise.
6503 (change_zero_ext): Likewise.
6504 (simplify_comparison): Likewise.
6505 * cse.c (cse_insn): Likewise.
6506 * dwarf2out.c (minmax_loc_descriptor): Likewise.
6507 (mem_loc_descriptor): Likewise.
6508 (loc_descriptor): Likewise.
6509 * expmed.c (init_expmed_one_mode): Likewise.
6510 (synth_mult): Likewise.
6511 (emit_store_flag_1): Likewise.
6512 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
6513 of a comparison with size.
6514 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
6515 (reduce_to_bit_field_precision): Likewise.
6516 * function.c (expand_function_end): Likewise.
6517 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
6518 * loop-doloop.c (doloop_modify): Likewise.
6519 * optabs.c (expand_binop): Likewise.
6520 (expand_unop): Likewise.
6521 (expand_copysign_absneg): Likewise.
6522 (prepare_cmp_insn): Likewise.
6523 (maybe_legitimize_operand): Likewise.
6524 * recog.c (const_scalar_int_operand): Likewise.
6525 * rtlanal.c (get_address_mode): Likewise.
6526 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6527 (simplify_cond_clz_ctz): Likewise.
6528 * tree-nested.c (get_nl_goto_field): Likewise.
6529 * tree.c (build_vector_type_for_mode): Likewise.
6530 * var-tracking.c (use_narrower_mode): Likewise.
6532 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6533 Alan Hayward <alan.hayward@arm.com>
6534 David Sherwood <david.sherwood@arm.com>
6536 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
6537 * builtins.c (expand_builtin_signbit): Use it.
6538 * cfgexpand.c (expand_debug_expr): Likewise.
6539 * dojump.c (do_jump): Likewise.
6540 (do_compare_and_jump): Likewise.
6541 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
6542 * expmed.c (make_tree): Likewise.
6543 * expr.c (expand_expr_real_2): Likewise.
6544 (expand_expr_real_1): Likewise.
6545 (try_casesi): Likewise.
6546 * fold-const-call.c (fold_const_call_ss): Likewise.
6547 * fold-const.c (unextend): Likewise.
6548 (extract_muldiv_1): Likewise.
6549 (fold_single_bit_test): Likewise.
6550 (native_encode_int): Likewise.
6551 (native_encode_string): Likewise.
6552 (native_interpret_int): Likewise.
6553 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
6554 * internal-fn.c (expand_addsub_overflow): Likewise.
6555 (expand_neg_overflow): Likewise.
6556 (expand_mul_overflow): Likewise.
6557 (expand_arith_overflow): Likewise.
6558 * match.pd: Likewise.
6559 * stor-layout.c (layout_type): Likewise.
6560 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
6561 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
6562 * tree-ssanames.c (get_range_info): Likewise.
6563 * tree-switch-conversion.c (array_value_type) Likewise.
6564 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
6565 (vect_recog_divmod_pattern): Likewise.
6566 (vect_recog_mixed_size_cond_pattern): Likewise.
6567 * tree-vrp.c (extract_range_basic): Likewise.
6568 (simplify_float_conversion_using_ranges): Likewise.
6569 * tree.c (int_fits_type_p): Likewise.
6570 * ubsan.c (instrument_bool_enum_load): Likewise.
6571 * varasm.c (mergeable_string_section): Likewise.
6572 (narrowing_initializer_constant_valid_p): Likewise.
6573 (output_constant): Likewise.
6575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6576 Alan Hayward <alan.hayward@arm.com>
6577 David Sherwood <david.sherwood@arm.com>
6579 * machmode.h (NARROWEST_INT_MODE): New macro.
6580 * expr.c (alignment_for_piecewise_move): Use it instead of
6581 GET_CLASS_NARROWEST_MODE (MODE_INT).
6582 (push_block): Likewise.
6583 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
6585 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6588 Alan Hayward <alan.hayward@arm.com>
6589 David Sherwood <david.sherwood@arm.com>
6591 * postreload.c (move2add_valid_value_p): Change the type of the
6592 mode parameter to scalar_int_mode.
6593 (move2add_use_add2_insn): Add a mode parameter and use it instead
6595 (move2add_use_add3_insn): Likewise.
6596 (reload_cse_move2add): Update accordingly.
6598 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6599 Alan Hayward <alan.hayward@arm.com>
6600 David Sherwood <david.sherwood@arm.com>
6602 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
6604 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
6605 * optabs.c (expand_unop): Likewise.
6607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6608 Alan Hayward <alan.hayward@arm.com>
6609 David Sherwood <david.sherwood@arm.com>
6611 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
6612 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
6613 (popcount_loc_descriptor): Likewise.
6614 (bswap_loc_descriptor): Likewise.
6615 (rotate_loc_descriptor): Likewise.
6616 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
6617 calling the functions above.
6619 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6620 Alan Hayward <alan.hayward@arm.com>
6621 David Sherwood <david.sherwood@arm.com>
6623 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
6625 (try_combine): Likewise.
6626 (simplify_if_then_else): Likewise.
6627 * cse.c (cse_insn): Likewise.
6628 * dwarf2out.c (mem_loc_descriptor): Likewise.
6629 * emit-rtl.c (gen_lowpart_common): Likewise.
6630 * simplify-rtx.c (simplify_truncation): Likewise.
6631 (simplify_binary_operation_1): Likewise.
6632 (simplify_const_relational_operation): Likewise.
6633 (simplify_ternary_operation): Likewise.
6634 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
6636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6637 Alan Hayward <alan.hayward@arm.com>
6638 David Sherwood <david.sherwood@arm.com>
6640 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
6641 * reload.c (push_reload): Likewise.
6642 (find_reloads): Likewise.
6644 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6645 Alan Hayward <alan.hayward@arm.com>
6646 David Sherwood <david.sherwood@arm.com>
6648 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
6649 (make_compound_operation_int): Likewise.
6650 (change_zero_ext): Likewise.
6651 * expr.c (convert_move): Likewise.
6652 (convert_modes): Likewise.
6653 * fwprop.c (forward_propagate_subreg): Likewise.
6654 * loop-iv.c (get_biv_step_1): Likewise.
6655 * optabs.c (widen_operand): Likewise.
6656 * postreload.c (move2add_valid_value_p): Likewise.
6657 * recog.c (simplify_while_replacing): Likewise.
6658 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6659 (simplify_binary_operation_1): Likewise. Remove redundant
6660 mode equality check.
6662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6663 Alan Hayward <alan.hayward@arm.com>
6664 David Sherwood <david.sherwood@arm.com>
6666 * combine.c (combine_simplify_rtx): Add checks for
6667 is_a <scalar_int_mode>.
6668 (simplify_if_then_else): Likewise.
6669 (make_field_assignment): Likewise.
6670 (simplify_comparison): Likewise.
6671 * ifcvt.c (noce_try_bitop): Likewise.
6672 * loop-invariant.c (canonicalize_address_mult): Likewise.
6673 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6676 Alan Hayward <alan.hayward@arm.com>
6677 David Sherwood <david.sherwood@arm.com>
6679 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
6680 is_a <scalar_int_mode> instead of != BLKmode.
6682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6683 Alan Hayward <alan.hayward@arm.com>
6684 David Sherwood <david.sherwood@arm.com>
6686 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
6687 instead of != VOIDmode.
6688 * combine.c (if_then_else_cond): Likewise.
6689 (change_zero_ext): Likewise.
6690 * dwarf2out.c (mem_loc_descriptor): Likewise.
6691 (loc_descriptor): Likewise.
6692 * rtlanal.c (canonicalize_condition): Likewise.
6693 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
6695 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6696 Alan Hayward <alan.hayward@arm.com>
6697 David Sherwood <david.sherwood@arm.com>
6699 * simplify-rtx.c (simplify_binary_operation_1): Use
6700 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
6702 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6703 Alan Hayward <alan.hayward@arm.com>
6704 David Sherwood <david.sherwood@arm.com>
6706 * wide-int.h (int_traits<unsigned char>) New class.
6707 (int_traits<unsigned short>) Likewise.
6708 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
6709 Use GET_MODE_UNIT_PRECISION and remove redundant test for
6711 * combine.c (set_nonzero_bits_and_sign_copies): Use
6712 is_a <scalar_int_mode>.
6713 (find_split_point): Likewise.
6714 (combine_simplify_rtx): Likewise.
6715 (simplify_logical): Likewise.
6716 (expand_compound_operation): Likewise.
6717 (expand_field_assignment): Likewise.
6718 (make_compound_operation): Likewise.
6719 (extended_count): Likewise.
6720 (change_zero_ext): Likewise.
6721 (simplify_comparison): Likewise.
6722 * dwarf2out.c (scompare_loc_descriptor): Likewise.
6723 (ucompare_loc_descriptor): Likewise.
6724 (minmax_loc_descriptor): Likewise.
6725 (mem_loc_descriptor): Likewise.
6726 (loc_descriptor): Likewise.
6727 * expmed.c (init_expmed_one_mode): Likewise.
6728 * lra-constraints.c (lra_constraint_offset): Likewise.
6729 * optabs.c (prepare_libcall_arg): Likewise.
6730 * postreload.c (move2add_note_store): Likewise.
6731 * reload.c (operands_match_p): Likewise.
6732 * rtl.h (load_extend_op): Likewise.
6733 * rtlhooks.c (gen_lowpart_general): Likewise.
6734 * simplify-rtx.c (simplify_truncation): Likewise.
6735 (simplify_unary_operation_1): Likewise.
6736 (simplify_binary_operation_1): Likewise.
6737 (simplify_const_binary_operation): Likewise.
6738 (simplify_const_relational_operation): Likewise.
6739 (simplify_subreg): Likewise.
6740 * stor-layout.c (bitwise_mode_for_mode): Likewise.
6741 * var-tracking.c (adjust_mems): Likewise.
6742 (prepare_call_arguments): Likewise.
6744 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6745 Alan Hayward <alan.hayward@arm.com>
6746 David Sherwood <david.sherwood@arm.com>
6748 * machmode.h (is_int_mode): New fuction.
6749 * combine.c (find_split_point): Use it.
6750 (combine_simplify_rtx): Likewise.
6751 (simplify_if_then_else): Likewise.
6752 (simplify_set): Likewise.
6753 (simplify_shift_const_1): Likewise.
6754 (simplify_comparison): Likewise.
6755 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
6756 * cse.c (notreg_cost): Likewise.
6757 (cse_insn): Likewise.
6758 * cselib.c (cselib_lookup_1): Likewise.
6759 * dojump.c (do_jump_1): Likewise.
6760 (do_compare_rtx_and_jump): Likewise.
6761 * dse.c (get_call_args): Likewise.
6762 * dwarf2out.c (rtl_for_decl_init): Likewise.
6763 (native_encode_initializer): Likewise.
6764 * expmed.c (emit_store_flag_1): Likewise.
6765 (emit_store_flag): Likewise.
6766 * expr.c (convert_modes): Likewise.
6767 (store_field): Likewise.
6768 (expand_expr_real_1): Likewise.
6769 * fold-const.c (fold_read_from_constant_string): Likewise.
6770 * gimple-ssa-sprintf.c (get_format_string): Likewise.
6771 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
6772 * optabs.c (expand_binop): Likewise.
6773 (expand_unop): Likewise.
6774 (expand_abs_nojump): Likewise.
6775 (expand_one_cmpl_abs_nojump): Likewise.
6776 * simplify-rtx.c (mode_signbit_p): Likewise.
6777 (val_signbit_p): Likewise.
6778 (val_signbit_known_set_p): Likewise.
6779 (val_signbit_known_clear_p): Likewise.
6780 (simplify_relational_operation_1): Likewise.
6781 * tree.c (vector_type_mode): Likewise.
6783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6784 Alan Hayward <alan.hayward@arm.com>
6785 David Sherwood <david.sherwood@arm.com>
6787 * machmode.h (smallest_mode_for_size): Fix formatting.
6788 (smallest_int_mode_for_size): New function.
6789 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
6790 instead of smallest_mode_for_size.
6791 * combine.c (make_extraction): Likewise.
6792 * config/arc/arc.c (arc_expand_movmem): Likewise.
6793 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
6794 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
6795 * config/s390/s390.c (s390_expand_insv): Likewise.
6796 * config/sparc/sparc.c (assign_int_registers): Likewise.
6797 * config/spu/spu.c (spu_function_value): Likewise.
6798 (spu_function_arg): Likewise.
6799 * coverage.c (get_gcov_type): Likewise.
6800 (get_gcov_unsigned_t): Likewise.
6801 * dse.c (find_shift_sequence): Likewise.
6802 * expmed.c (store_bit_field_1): Likewise.
6803 * expr.c (convert_move): Likewise.
6804 (store_field): Likewise.
6805 * internal-fn.c (expand_arith_overflow): Likewise.
6806 * optabs-query.c (get_best_extraction_insn): Likewise.
6807 * optabs.c (expand_twoval_binop_libfunc): Likewise.
6808 * stor-layout.c (layout_type): Likewise.
6809 (initialize_sizetypes): Likewise.
6810 * targhooks.c (default_get_mask_mode): Likewise.
6811 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
6813 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6814 Alan Hayward <alan.hayward@arm.com>
6815 David Sherwood <david.sherwood@arm.com>
6817 * machmode.h (opt_mode::else_blk): New function.
6818 (int_mode_for_mode): Declare.
6819 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
6820 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
6822 * cfgexpand.c (expand_debug_expr): Likewise.
6823 * combine.c (gen_lowpart_or_truncate): Likewise.
6824 (gen_lowpart_for_combine): Likewise.
6825 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
6826 * config/avr/avr.c (avr_to_int_mode): Likewise.
6827 (avr_out_plus_1): Likewise.
6828 (avr_out_plus): Likewise.
6829 (avr_out_round): Likewise.
6830 * config/i386/i386.c (ix86_split_to_parts): Likewise.
6831 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
6832 (s390_expand_vcond): Likewise.
6833 * config/spu/spu.c (spu_split_immediate): Likewise.
6834 (spu_expand_mov): Likewise.
6835 * dse.c (get_stored_val): Likewise.
6836 * expmed.c (store_bit_field_1): Likewise.
6837 (convert_extracted_bit_field): Use int_mode_for_mode instead of
6839 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
6840 (extract_low_bits): Likewise.
6841 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
6842 handling rather than repeating the check.
6843 (emit_group_store): Likewise.
6844 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
6845 * optabs.c (expand_absneg_bit): Likewise.
6846 (expand_copysign_absneg): Likewise.
6847 (expand_copysign_bit): Likewise.
6848 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
6849 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
6850 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
6851 * var-tracking.c (prepare_call_arguments): Likewise.
6852 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
6853 int_mode_for_mode instead of mode_for_size.
6854 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
6856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6857 Alan Hayward <alan.hayward@arm.com>
6858 David Sherwood <david.sherwood@arm.com>
6860 * machmode.h (int_mode_for_size): New function.
6861 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
6862 instead of mode_for_size.
6863 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
6865 * combine.c (expand_field_assignment): Use int_mode_for_size
6866 instead of mode_for_size.
6867 (make_extraction): Likewise.
6868 (simplify_shift_const_1): Likewise.
6869 (simplify_comparison): Likewise.
6870 * dojump.c (do_jump): Likewise.
6871 * dwarf2out.c (mem_loc_descriptor): Likewise.
6872 * emit-rtl.c (init_derived_machine_modes): Likewise.
6873 * expmed.c (flip_storage_order): Likewise.
6874 (convert_extracted_bit_field): Likewise.
6875 * expr.c (copy_blkmode_from_reg): Likewise.
6876 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
6877 * internal-fn.c (expand_mul_overflow): Likewise.
6878 * lower-subreg.c (simple_move): Likewise.
6879 * optabs-libfuncs.c (init_optabs): Likewise.
6880 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6881 * tree.c (vector_type_mode): Likewise.
6882 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
6883 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
6884 * tree-vect-generic.c (expand_vector_parallel): Likewise.
6885 * tree-vect-stmts.c (vectorizable_load): Likewise.
6886 (vectorizable_store): Likewise.
6888 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6889 Alan Hayward <alan.hayward@arm.com>
6890 David Sherwood <david.sherwood@arm.com>
6892 * coretypes.h (pod_mode): New type.
6893 (scalar_int_mode_pod): New typedef.
6894 * machmode.h (pod_mode): New class.
6895 (int_n_data_t::m): Change type to scalar_int_mode_pod.
6896 * genmodes.c (emit_mode_int_n): Update accordingly.
6897 * lower-subreg.h (target_lower_subreg): Change type to
6898 scalar_int_mode_pod.
6899 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
6900 scalar_int_mode_pod.
6902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6903 Alan Hayward <alan.hayward@arm.com>
6904 David Sherwood <david.sherwood@arm.com>
6906 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
6907 machine_mode to scalar_int_mode.
6908 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
6909 (rs6000_option_override_internal): Remove cast to int.
6910 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
6911 machine_mode to scalar_int_mode.
6912 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
6913 (rs6000_option_override_internal): Remove cast to int.
6914 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
6915 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
6917 * config/s390/s390.c (s390_expand_builtin): Likewise.
6918 * coretypes.h (scalar_int_mode): New type.
6919 (opt_scalar_int_mode): New typedef.
6920 * machmode.h (scalar_int_mode): New class.
6921 (scalar_int_mode::includes_p): New function.
6922 (byte_mode): Change type to scalar_int_mode.
6923 (word_mode): Likewise.
6924 (ptr_mode): Likewise.
6925 * emit-rtl.c (byte_mode): Likewise.
6926 (word_mode): Likewise.
6927 (ptr_mode): Likewise.
6928 (init_derived_machine_modes): Update accordingly.
6929 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
6930 and MODE_PARTIAL_INT.
6931 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
6932 opt_scalar_int_mode.
6934 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6935 Alan Hayward <alan.hayward@arm.com>
6936 David Sherwood <david.sherwood@arm.com>
6938 * target.def (libgcc_floating_mode_supported_p): Take a
6940 * doc/tm.texi: Regenerate.
6941 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
6943 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
6944 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
6947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6948 Alan Hayward <alan.hayward@arm.com>
6949 David Sherwood <david.sherwood@arm.com>
6951 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
6952 * doc/tm.texi: Regenerate.
6953 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
6954 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
6955 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
6956 * targhooks.h (default_floatn_mode): Likewise.
6957 * targhooks.c (default_floatn_mode): Likewise.
6958 * tree.c (build_common_tree_nodes): Update accordingly.
6960 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6961 Alan Hayward <alan.hayward@arm.com>
6962 David Sherwood <david.sherwood@arm.com>
6964 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
6965 (mode_iterator::iterate_p): Likewise.
6966 (mode_iterator::get_wider): Likewise.
6967 * expr.c (init_expr_target): Use opt_scalar_float_mode.
6969 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6970 Alan Hayward <alan.hayward@arm.com>
6971 David Sherwood <david.sherwood@arm.com>
6973 * coretypes.h (opt_scalar_float_mode): New typedef.
6974 * machmode.h (float_mode_for_size): New function.
6975 * emit-rtl.c (double_mode): Delete.
6976 (init_emit_once): Use float_mode_for_size.
6977 * stor-layout.c (layout_type): Likewise.
6978 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
6980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6981 Alan Hayward <alan.hayward@arm.com>
6982 David Sherwood <david.sherwood@arm.com>
6984 * output.h (assemble_real): Take a scalar_float_mode.
6985 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
6986 * config/arm/arm.md (consttable_4): Likewise.
6987 (consttable_8): Likewise.
6988 (consttable_16): Likewise.
6989 * config/mips/mips.md (consttable_float): Likewise.
6990 * config/s390/s390.c (s390_output_pool_entry): Likewise.
6991 * varasm.c (assemble_real): Take a scalar_float_mode.
6992 (output_constant_pool_2): Update accordingly.
6993 (output_constant): Likewise.
6995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6996 Alan Hayward <alan.hayward@arm.com>
6997 David Sherwood <david.sherwood@arm.com>
6999 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
7000 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
7001 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
7002 (native_encode_real): Likewise.
7003 (native_interpret_real): Likewise.
7004 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
7005 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7007 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7008 Alan Hayward <alan.hayward@arm.com>
7009 David Sherwood <david.sherwood@arm.com>
7011 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
7012 <scalar_float_mode>. Simplify.
7013 (gen_extend_conv_libfunc): Likewise.
7015 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7016 Alan Hayward <alan.hayward@arm.com>
7017 David Sherwood <david.sherwood@arm.com>
7019 * coretypes.h (scalar_float_mode): New type.
7020 * machmode.h (mode_traits::from_int): Use machine_mode if
7021 USE_ENUM_MODES is defined.
7022 (is_a): New function.
7024 (dyn_cast): Likewise.
7025 (scalar_float_mode): New class.
7026 (scalar_float_mode::includes_p): New function.
7027 (is_float_mode): Likewise.
7028 * gdbhooks.py (MachineModePrinter): New class.
7029 (build_pretty_printer): Use it for scalar_float_mode.
7030 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
7031 (format_helper::format_helper): Turn into a template.
7032 * genmodes.c (get_mode_class): New function.
7033 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
7034 or machine_mode if none.
7035 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
7036 as_a <scalar_float_mode>.
7037 * dwarf2out.c (mem_loc_descriptor): Likewise.
7038 (insert_float): Likewise.
7039 (add_const_value_attribute): Likewise.
7040 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7041 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
7042 (expand_unop): Update accordingly.
7043 (expand_abs_nojump): Likewise.
7044 (expand_copysign_absneg): Take a scalar_float_mode.
7045 (expand_copysign_bit): Likewise.
7046 (expand_copysign): Update accordingly.
7048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7049 Alan Hayward <alan.hayward@arm.com>
7050 David Sherwood <david.sherwood@arm.com>
7052 * coretypes.h (opt_mode): New class.
7053 * machmode.h (opt_mode): Likewise.
7054 (opt_mode::else_void): New function.
7055 (opt_mode::require): Likewise.
7056 (opt_mode::exists): Likewise.
7057 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
7058 (GET_MODE_2XWIDER_MODE): Likewise.
7059 (mode_iterator::get_wider): Update accordingly.
7060 (mode_iterator::get_2xwider): Likewise.
7061 (mode_iterator::get_known_wider): Likewise, turning into a template.
7062 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
7063 forcing a wider mode to exist.
7064 * config/cr16/cr16.h (LONG_REG_P): Likewise.
7065 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7066 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
7067 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
7068 * lower-subreg.c (init_lower_subreg): Likewise.
7069 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
7070 on the final iteration.
7071 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
7072 a wider mode exists before asking for a move pattern.
7073 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
7074 forcing a wider mode to exist.
7075 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
7076 returning false if no such mode exists.
7077 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
7078 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
7079 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
7080 Avoid checking for a MODE_INT if we already know the mode is not a
7082 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
7083 forcing a wider mode to exist.
7084 (expmed_mult_highpart_optab): Likewise.
7085 (expmed_mult_highpart): Likewise.
7086 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
7088 * lto-streamer-in.c (lto_input_mode_table): Likewise.
7089 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
7090 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
7091 * internal-fn.c (expand_mul_overflow): Update use of
7092 GET_MODE_2XWIDER_MODE.
7093 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7094 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
7095 GET_MODE_WIDER_MODE.
7096 (convert_plusminus_to_widen): Likewise.
7097 * tree-switch-conversion.c (array_value_type): Likewise.
7098 * var-tracking.c (emit_note_insn_var_location): Likewise.
7099 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7100 Return false inside rather than outside the loop if no wider mode
7102 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
7103 and GET_MODE_2XWIDER_MODE
7104 (can_compare_p): Use else_void.
7105 * gdbhooks.py (OptMachineModePrinter): New class.
7106 (build_pretty_printer): Use it for opt_mode.
7108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7109 Alan Hayward <alan.hayward@arm.com>
7110 David Sherwood <david.sherwood@arm.com>
7112 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
7113 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
7115 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7116 Alan Hayward <alan.hayward@arm.com>
7117 David Sherwood <david.sherwood@arm.com>
7119 * machmode.h (mode_traits): New structure.
7120 (get_narrowest_mode): New function.
7121 (mode_iterator::start): Likewise.
7122 (mode_iterator::iterate_p): Likewise.
7123 (mode_iterator::get_wider): Likewise.
7124 (mode_iterator::get_known_wider): Likewise.
7125 (mode_iterator::get_2xwider): Likewise.
7126 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
7127 (FOR_EACH_MODE): Likewise.
7128 (FOR_EACH_MODE_FROM): Likewise.
7129 (FOR_EACH_MODE_UNTIL): Likewise.
7130 (FOR_EACH_WIDER_MODE): Likewise.
7131 (FOR_EACH_2XWIDER_MODE): Likewise.
7132 * builtins.c (expand_builtin_strlen): Use new mode iterators.
7133 * combine.c (simplify_comparison): Likewise
7134 * config/i386/i386.c (type_natural_mode): Likewise.
7135 * cse.c (cse_insn): Likewise.
7136 * dse.c (find_shift_sequence): Likewise.
7137 * emit-rtl.c (init_derived_machine_modes): Likewise.
7138 (init_emit_once): Likewise.
7139 * explow.c (hard_function_value): Likewise.
7140 * expmed.c (extract_fixed_bit_field_1): Likewise.
7141 (extract_bit_field_1): Likewise.
7142 (expand_divmod): Likewise.
7143 (emit_store_flag_1): Likewise.
7144 * expr.c (init_expr_target): Likewise.
7145 (convert_move): Likewise.
7146 (alignment_for_piecewise_move): Likewise.
7147 (widest_int_mode_for_size): Likewise.
7148 (emit_block_move_via_movmem): Likewise.
7149 (copy_blkmode_to_reg): Likewise.
7150 (set_storage_via_setmem): Likewise.
7151 (compress_float_constant): Likewise.
7152 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7153 * optabs-query.c (get_best_extraction_insn): Likewise.
7154 * optabs.c (expand_binop): Likewise.
7155 (expand_twoval_unop): Likewise.
7156 (expand_twoval_binop): Likewise.
7157 (widen_leading): Likewise.
7158 (widen_bswap): Likewise.
7159 (expand_parity): Likewise.
7160 (expand_unop): Likewise.
7161 (prepare_cmp_insn): Likewise.
7162 (prepare_float_lib_cmp): Likewise.
7163 (expand_float): Likewise.
7164 (expand_fix): Likewise.
7165 (expand_sfix_optab): Likewise.
7166 * postreload.c (move2add_use_add2_insn): Likewise.
7167 * reg-stack.c (reg_to_stack): Likewise.
7168 * reginfo.c (choose_hard_reg_mode): Likewise.
7169 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7170 * stor-layout.c (mode_for_size): Likewise.
7171 (smallest_mode_for_size): Likewise.
7172 (mode_for_vector): Likewise.
7173 (finish_bitfield_representative): Likewise.
7174 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
7175 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
7176 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7177 * var-tracking.c (prepare_call_arguments): Likewise.
7179 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7180 Alan Hayward <alan.hayward@arm.com>
7181 David Sherwood <david.sherwood@arm.com>
7183 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
7184 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
7185 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
7186 * machmode.h (mode_size): Move earlier in file.
7187 (mode_precision): Likewise.
7188 (mode_inner): Likewise.
7189 (mode_nunits): Likewise.
7190 (mode_unit_size): Likewise.
7191 (unit_unit_precision): Likewise.
7192 (mode_wider): Likewise.
7193 (mode_2xwider): Likewise.
7194 (machine_mode): New class.
7195 (mode_to_bytes): New function.
7196 (mode_to_bits): Likewise.
7197 (mode_to_precision): Likewise.
7198 (mode_to_inner): Likewise.
7199 (mode_to_unit_size): Likewise.
7200 (mode_to_unit_precision): Likewise.
7201 (mode_to_nunits): Likewise.
7202 (GET_MODE_SIZE): Use mode_to_bytes.
7203 (GET_MODE_BITSIZE): Use mode_to_bits.
7204 (GET_MODE_PRECISION): Use mode_to_precision.
7205 (GET_MODE_INNER): Use mode_to_inner.
7206 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
7207 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
7208 (GET_MODE_NUNITS): Use mode_to_nunits.
7209 * system.h (ALWAYS_INLINE): New macro.
7210 * config/powerpcspe/powerpcspe-c.c
7211 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
7212 int for arg1_mode and arg2_mode.
7214 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7215 Alan Hayward <alan.hayward@arm.com>
7216 David Sherwood <david.sherwood@arm.com>
7218 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
7219 Prefix mode names with E_ in case statements.
7220 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7221 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
7222 (aarch64_split_simd_move): Likewise.
7223 (aarch64_gen_storewb_pair): Likewise.
7224 (aarch64_gen_loadwb_pair): Likewise.
7225 (aarch64_gen_store_pair): Likewise.
7226 (aarch64_gen_load_pair): Likewise.
7227 (aarch64_get_condition_code_1): Likewise.
7228 (aarch64_constant_pool_reload_icode): Likewise.
7229 (get_rsqrte_type): Likewise.
7230 (get_rsqrts_type): Likewise.
7231 (get_recpe_type): Likewise.
7232 (get_recps_type): Likewise.
7233 (aarch64_gimplify_va_arg_expr): Likewise.
7234 (aarch64_simd_container_mode): Likewise.
7235 (aarch64_emit_load_exclusive): Likewise.
7236 (aarch64_emit_store_exclusive): Likewise.
7237 (aarch64_expand_compare_and_swap): Likewise.
7238 (aarch64_gen_atomic_cas): Likewise.
7239 (aarch64_emit_bic): Likewise.
7240 (aarch64_emit_atomic_swap): Likewise.
7241 (aarch64_emit_atomic_load_op): Likewise.
7242 (aarch64_evpc_trn): Likewise.
7243 (aarch64_evpc_uzp): Likewise.
7244 (aarch64_evpc_zip): Likewise.
7245 (aarch64_evpc_ext): Likewise.
7246 (aarch64_evpc_rev): Likewise.
7247 (aarch64_evpc_dup): Likewise.
7248 (aarch64_gen_ccmp_first): Likewise.
7249 (aarch64_gen_ccmp_next): Likewise.
7250 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7251 (alpha_emit_xfloating_libcall): Likewise.
7252 (emit_insxl): Likewise.
7253 (alpha_arg_type): Likewise.
7254 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
7255 (arc_preferred_simd_mode): Likewise.
7256 (arc_secondary_reload): Likewise.
7257 (get_arc_condition_code): Likewise.
7258 (arc_print_operand): Likewise.
7259 (arc_legitimate_constant_p): Likewise.
7260 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7261 * config/arc/arc.md (casesi_load): Likewise.
7262 (casesi_compact_jump): Likewise.
7263 * config/arc/predicates.md (proper_comparison_operator): Likewise.
7264 (cc_use_register): Likewise.
7265 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7266 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
7267 (arm_init_iwmmxt_builtins): Likewise.
7268 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
7269 (neon_expand_vector_init): Likewise.
7270 (arm_attr_length_move_neon): Likewise.
7271 (maybe_get_arm_condition_code): Likewise.
7272 (arm_emit_vector_const): Likewise.
7273 (arm_preferred_simd_mode): Likewise.
7274 (arm_output_iwmmxt_tinsr): Likewise.
7275 (thumb1_output_casesi): Likewise.
7276 (thumb2_output_casesi): Likewise.
7277 (arm_emit_load_exclusive): Likewise.
7278 (arm_emit_store_exclusive): Likewise.
7279 (arm_expand_compare_and_swap): Likewise.
7280 (arm_evpc_neon_vuzp): Likewise.
7281 (arm_evpc_neon_vzip): Likewise.
7282 (arm_evpc_neon_vrev): Likewise.
7283 (arm_evpc_neon_vtrn): Likewise.
7284 (arm_evpc_neon_vext): Likewise.
7285 (arm_validize_comparison): Likewise.
7286 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
7287 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
7288 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
7289 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
7290 (c6x_preferred_simd_mode): Likewise.
7291 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
7292 (epiphany_rtx_costs): Likewise.
7293 * config/epiphany/predicates.md (proper_comparison_operator):
7295 * config/frv/frv.c (condexec_memory_operand): Likewise.
7296 (frv_emit_move): Likewise.
7297 (output_move_single): Likewise.
7298 (output_condmove_single): Likewise.
7299 (frv_hard_regno_mode_ok): Likewise.
7300 (frv_matching_accg_mode): Likewise.
7301 * config/h8300/h8300.c (split_adds_subs): Likewise.
7302 (h8300_rtx_costs): Likewise.
7303 (h8300_print_operand): Likewise.
7304 (compute_mov_length): Likewise.
7305 (output_logical_op): Likewise.
7306 (compute_logical_op_length): Likewise.
7307 (compute_logical_op_cc): Likewise.
7308 (h8300_shift_needs_scratch_p): Likewise.
7309 (output_a_shift): Likewise.
7310 (compute_a_shift_length): Likewise.
7311 (compute_a_shift_cc): Likewise.
7312 (expand_a_rotate): Likewise.
7313 (output_a_rotate): Likewise.
7314 * config/i386/i386.c (classify_argument): Likewise.
7315 (function_arg_advance_32): Likewise.
7316 (function_arg_32): Likewise.
7317 (function_arg_64): Likewise.
7318 (function_value_64): Likewise.
7319 (ix86_gimplify_va_arg): Likewise.
7320 (ix86_legitimate_constant_p): Likewise.
7321 (put_condition_code): Likewise.
7322 (split_double_mode): Likewise.
7323 (ix86_avx256_split_vector_move_misalign): Likewise.
7324 (ix86_expand_vector_logical_operator): Likewise.
7325 (ix86_split_idivmod): Likewise.
7326 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
7327 (ix86_build_const_vector): Likewise.
7328 (ix86_build_signbit_mask): Likewise.
7329 (ix86_match_ccmode): Likewise.
7330 (ix86_cc_modes_compatible): Likewise.
7331 (ix86_expand_branch): Likewise.
7332 (ix86_expand_sse_cmp): Likewise.
7333 (ix86_expand_sse_movcc): Likewise.
7334 (ix86_expand_int_sse_cmp): Likewise.
7335 (ix86_expand_vec_perm_vpermi2): Likewise.
7336 (ix86_expand_vec_perm): Likewise.
7337 (ix86_expand_sse_unpack): Likewise.
7338 (ix86_expand_int_addcc): Likewise.
7339 (ix86_split_to_parts): Likewise.
7340 (ix86_vectorize_builtin_gather): Likewise.
7341 (ix86_vectorize_builtin_scatter): Likewise.
7342 (avx_vpermilp_parallel): Likewise.
7343 (inline_memory_move_cost): Likewise.
7344 (ix86_tieable_integer_mode_p): Likewise.
7345 (x86_maybe_negate_const_int): Likewise.
7346 (ix86_expand_vector_init_duplicate): Likewise.
7347 (ix86_expand_vector_init_one_nonzero): Likewise.
7348 (ix86_expand_vector_init_one_var): Likewise.
7349 (ix86_expand_vector_init_concat): Likewise.
7350 (ix86_expand_vector_init_interleave): Likewise.
7351 (ix86_expand_vector_init_general): Likewise.
7352 (ix86_expand_vector_set): Likewise.
7353 (ix86_expand_vector_extract): Likewise.
7354 (emit_reduc_half): Likewise.
7355 (ix86_emit_i387_round): Likewise.
7356 (ix86_mangle_type): Likewise.
7357 (ix86_expand_round_sse4): Likewise.
7358 (expand_vec_perm_blend): Likewise.
7359 (canonicalize_vector_int_perm): Likewise.
7360 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
7361 (expand_vec_perm_1): Likewise.
7362 (expand_vec_perm_interleave3): Likewise.
7363 (expand_vec_perm_even_odd_pack): Likewise.
7364 (expand_vec_perm_even_odd_1): Likewise.
7365 (expand_vec_perm_broadcast_1): Likewise.
7366 (ix86_vectorize_vec_perm_const_ok): Likewise.
7367 (ix86_expand_vecop_qihi): Likewise.
7368 (ix86_expand_mul_widen_hilo): Likewise.
7369 (ix86_expand_sse2_abs): Likewise.
7370 (ix86_expand_pextr): Likewise.
7371 (ix86_expand_pinsr): Likewise.
7372 (ix86_preferred_simd_mode): Likewise.
7373 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
7374 * config/i386/sse.md (*andnot<mode>3): Likewise.
7375 (<mask_codefor><code><mode>3<mask_name>): Likewise.
7376 (*<code><mode>3): Likewise.
7377 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
7378 (ia64_expand_atomic_op): Likewise.
7379 (ia64_arg_type): Likewise.
7380 (ia64_mode_to_int): Likewise.
7381 (ia64_scalar_mode_supported_p): Likewise.
7382 (ia64_vector_mode_supported_p): Likewise.
7383 (expand_vec_perm_broadcast): Likewise.
7384 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
7385 (iq2000_function_arg_advance): Likewise.
7386 (iq2000_function_arg): Likewise.
7387 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
7388 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
7389 (m68k_libcall_value): Likewise.
7390 (m68k_function_value): Likewise.
7391 (sched_attr_op_type): Likewise.
7392 * config/mcore/mcore.c (mcore_output_move): Likewise.
7393 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
7395 (microblaze_function_arg): Likewise.
7396 * config/mips/mips.c (mips16_build_call_stub): Likewise.
7397 (mips_print_operand): Likewise.
7398 (mips_mode_ok_for_mov_fmt_p): Likewise.
7399 (mips_vector_mode_supported_p): Likewise.
7400 (mips_preferred_simd_mode): Likewise.
7401 (mips_expand_vpc_loongson_even_odd): Likewise.
7402 (mips_expand_vec_unpack): Likewise.
7403 (mips_expand_vi_broadcast): Likewise.
7404 (mips_expand_vector_init): Likewise.
7405 (mips_expand_vec_reduc): Likewise.
7406 (mips_expand_msa_cmp): Likewise.
7407 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
7408 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
7409 (cc_flags_for_mode): Likewise.
7410 * config/msp430/msp430.c (msp430_print_operand): Likewise.
7411 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
7412 (nds32_output_casesi_pc_relative): Likewise.
7413 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7414 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
7415 (nvptx_gen_unpack): Likewise.
7416 (nvptx_gen_pack): Likewise.
7417 (nvptx_gen_shuffle): Likewise.
7418 (nvptx_gen_wcast): Likewise.
7419 (nvptx_preferred_simd_mode): Likewise.
7420 * config/pa/pa.c (pa_secondary_reload): Likewise.
7421 * config/pa/predicates.md (base14_operand): Likewise.
7422 * config/powerpcspe/powerpcspe-c.c
7423 (altivec_resolve_overloaded_builtin): Likewise.
7424 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
7426 (rs6000_preferred_simd_mode): Likewise.
7427 (output_vec_const_move): Likewise.
7428 (rs6000_expand_vector_extract): Likewise.
7429 (rs6000_split_vec_extract_var): Likewise.
7430 (reg_offset_addressing_ok_p): Likewise.
7431 (rs6000_legitimate_offset_address_p): Likewise.
7432 (rs6000_legitimize_address): Likewise.
7433 (rs6000_emit_set_const): Likewise.
7434 (rs6000_const_vec): Likewise.
7435 (rs6000_emit_move): Likewise.
7436 (spe_build_register_parallel): Likewise.
7437 (rs6000_darwin64_record_arg_recurse): Likewise.
7438 (swap_selector_for_mode): Likewise.
7439 (spe_init_builtins): Likewise.
7440 (paired_init_builtins): Likewise.
7441 (altivec_init_builtins): Likewise.
7442 (do_load_for_compare): Likewise.
7443 (rs6000_generate_compare): Likewise.
7444 (rs6000_expand_float128_convert): Likewise.
7445 (emit_load_locked): Likewise.
7446 (emit_store_conditional): Likewise.
7447 (rs6000_output_function_epilogue): Likewise.
7448 (rs6000_handle_altivec_attribute): Likewise.
7449 (rs6000_function_value): Likewise.
7450 (emit_fusion_gpr_load): Likewise.
7451 (emit_fusion_p9_load): Likewise.
7452 (emit_fusion_p9_store): Likewise.
7453 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
7454 (fusion_gpr_mem_load): Likewise.
7455 (fusion_addis_mem_combo_load): Likewise.
7456 (fusion_addis_mem_combo_store): Likewise.
7457 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
7458 (fusion_gpr_mem_load): Likewise.
7459 (fusion_addis_mem_combo_load): Likewise.
7460 (fusion_addis_mem_combo_store): Likewise.
7461 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7463 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
7464 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7465 (rs6000_preferred_simd_mode): Likewise.
7466 (output_vec_const_move): Likewise.
7467 (rs6000_expand_vector_extract): Likewise.
7468 (rs6000_split_vec_extract_var): Likewise.
7469 (reg_offset_addressing_ok_p): Likewise.
7470 (rs6000_legitimate_offset_address_p): Likewise.
7471 (rs6000_legitimize_address): Likewise.
7472 (rs6000_emit_set_const): Likewise.
7473 (rs6000_const_vec): Likewise.
7474 (rs6000_emit_move): Likewise.
7475 (rs6000_darwin64_record_arg_recurse): Likewise.
7476 (swap_selector_for_mode): Likewise.
7477 (paired_init_builtins): Likewise.
7478 (altivec_init_builtins): Likewise.
7479 (rs6000_expand_float128_convert): Likewise.
7480 (emit_load_locked): Likewise.
7481 (emit_store_conditional): Likewise.
7482 (rs6000_output_function_epilogue): Likewise.
7483 (rs6000_handle_altivec_attribute): Likewise.
7484 (rs6000_function_value): Likewise.
7485 (emit_fusion_gpr_load): Likewise.
7486 (emit_fusion_p9_load): Likewise.
7487 (emit_fusion_p9_store): Likewise.
7488 * config/rx/rx.c (rx_gen_move_template): Likewise.
7489 (flags_from_mode): Likewise.
7490 * config/s390/predicates.md (s390_alc_comparison): Likewise.
7491 (s390_slb_comparison): Likewise.
7492 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
7493 (s390_vector_mode_supported_p): Likewise.
7494 (s390_cc_modes_compatible): Likewise.
7495 (s390_match_ccmode_set): Likewise.
7496 (s390_canonicalize_comparison): Likewise.
7497 (s390_emit_compare_and_swap): Likewise.
7498 (s390_branch_condition_mask): Likewise.
7499 (s390_rtx_costs): Likewise.
7500 (s390_secondary_reload): Likewise.
7501 (__SECONDARY_RELOAD_CASE): Likewise.
7502 (s390_expand_cs): Likewise.
7503 (s390_preferred_simd_mode): Likewise.
7504 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
7505 * config/sh/sh.c (sh_print_operand): Likewise.
7506 (dump_table): Likewise.
7507 (sh_secondary_reload): Likewise.
7508 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7509 * config/sh/sh.md (casesi_worker_1): Likewise.
7510 (casesi_worker_2): Likewise.
7511 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
7512 (fcc_comparison_operator): Likewise.
7513 * config/sparc/sparc.c (sparc_expand_move): Likewise.
7514 (emit_soft_tfmode_cvt): Likewise.
7515 (sparc_preferred_simd_mode): Likewise.
7516 (output_cbranch): Likewise.
7517 (sparc_print_operand): Likewise.
7518 (sparc_expand_vec_perm_bmask): Likewise.
7519 (vector_init_bshuffle): Likewise.
7520 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7521 (spu_vector_mode_supported_p): Likewise.
7522 (spu_expand_insv): Likewise.
7523 (spu_emit_branch_or_set): Likewise.
7524 (spu_handle_vector_attribute): Likewise.
7525 (spu_builtin_splats): Likewise.
7526 (spu_builtin_extract): Likewise.
7527 (spu_builtin_promote): Likewise.
7528 (spu_expand_sign_extend): Likewise.
7529 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
7530 (tilegx_simd_int): Likewise.
7531 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
7532 (tilepro_simd_int): Likewise.
7533 * config/v850/v850.c (const_double_split): Likewise.
7534 (v850_print_operand): Likewise.
7535 (ep_memory_offset): Likewise.
7536 * config/vax/vax.c (vax_rtx_costs): Likewise.
7537 (vax_output_int_move): Likewise.
7538 (vax_output_int_add): Likewise.
7539 (vax_output_int_subtract): Likewise.
7540 * config/visium/predicates.md (visium_branch_operator): Likewise.
7541 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
7542 (visium_print_operand_address): Likewise.
7543 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7544 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
7545 (xtensa_expand_conditional_branch): Likewise.
7546 (xtensa_copy_incoming_a7): Likewise.
7547 (xtensa_output_literal): Likewise.
7548 * dfp.c (decimal_real_maxval): Likewise.
7549 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
7551 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7552 Alan Hayward <alan.hayward@arm.com>
7553 David Sherwood <david.sherwood@arm.com>
7555 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
7556 (mode_nunits_inline): Likewise.
7557 (mode_inner_inline): Likewise.
7558 (mode_unit_size_inline): Likewise.
7559 (mode_unit_precision_inline): Likewise.
7560 (emit_insn_modes_h): Likewise. Also emit a #define of the
7562 (emit_mode_wider): Add an E_ prefix to mode names.
7563 (emit_mode_complex): Likewise.
7564 (emit_mode_inner): Likewise.
7565 (emit_mode_adjustments): Likewise.
7566 (emit_mode_int_n): Likewise.
7567 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
7568 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
7569 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
7570 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
7571 (CRC32_BUILTIN, ENTRY): Likewise.
7572 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
7573 (aarch64_pop_regs): Likewise.
7574 (aarch64_process_components): Likewise.
7575 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
7576 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
7577 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
7578 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
7579 * config/arm/arm.c (arm_init_libfuncs): Likewise.
7580 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
7582 * config/i386/i386-builtin.def (pcmpestr): Likewise.
7583 (pcmpistr): Likewise.
7584 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
7585 * config/mmix/mmix.c (mmix_output_condition): Likewise.
7586 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
7588 * config/rl78/rl78.c (mduc_regs): Likewise.
7589 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7590 (htm_expand_builtin): Likewise.
7591 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
7592 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
7593 * config/xtensa/xtensa.c (print_operand): Likewise.
7594 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
7595 (NUM_MODE_VECTOR_INT): Likewise.
7596 * genoutput.c (null_operand): Likewise.
7597 (output_operand_data): Likewise.
7598 * genrecog.c (print_parameter_value): Likewise.
7599 * lra.c (debug_operand_data): Likewise.
7601 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7603 * dwarf2out.c (scompare_loc_descriptor_wide)
7604 (scompare_loc_descriptor_narrow): New functions, split out from...
7605 (scompare_loc_descriptor): ...here.
7606 * expmed.c (emit_store_flag_int): New function, split out from...
7607 (emit_store_flag): ...here.
7609 2017-08-30 Richard Biener <rguenther@suse.de>
7611 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
7612 (dwarf2out_early_finish): Move setting of AT_pubnames from
7613 early debug output to early finish.
7615 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7617 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
7618 and -mdata-region to the assembler.
7620 2017-08-30 Richard Biener <rguenther@suse.de>
7622 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
7624 (gen_subprogram_die): Add DW_AT_object_pointer only early.
7625 (dwarf2out_early_global_decl): Only generate a DIE for the
7626 abstract origin if it doesn't already exist or is a declaration DIE.
7627 (resolve_addr): Do not add the linkage name twice when
7628 generating a stub DIE for the DW_TAG_GNU_call_site target.
7630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7632 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7633 Use machine_mode rather than int for arg1_mode.
7635 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
7638 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
7639 that the second argument of the built-in functions to unpack
7640 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
7641 switch statement instead a lot of if statements.
7642 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
7643 Allow 64-bit values to be in Altivec registers as well as
7644 traditional floating point registers.
7645 (pack<mode>, FMOVE128_VSX iterator): Likewise.
7647 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
7649 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
7650 MAX_REGS_PER_ADDRESS == 1.
7652 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
7654 * config/i386/i386.opt (flag_fentry): Do not init to -1.
7655 * config/i386/i386.c (ix86_option_override_internal): Simplify
7656 setting of opts->x_flag_entry.
7658 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7659 Jakub Jelinek <jakub@redhat.com>
7660 Richard Biener <rguenther@suse.de>
7662 PR tree-optimization/81503
7663 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
7664 folded constant fits in the target type; reorder tests for clarity.
7666 2017-08-29 Martin Liska <mliska@suse.cz>
7668 * passes.def: Include pass_lower_switch.
7669 * stmt.c (dump_case_nodes): Remove and move to
7670 tree-switch-conversion.
7671 (case_values_threshold): Likewise.
7672 (expand_switch_as_decision_tree_p): Likewise.
7673 (emit_case_decision_tree): Likewise.
7674 (expand_case): Likewise.
7675 (balance_case_nodes): Likewise.
7676 (node_has_low_bound): Likewise.
7677 (node_has_high_bound): Likewise.
7678 (node_is_bounded): Likewise.
7679 (emit_case_nodes): Likewise.
7680 (struct simple_case_node): New struct.
7681 (add_case_node): Remove.
7682 (emit_case_dispatch_table): Use vector instead of case_list.
7683 (reset_out_edges_aux): Remove.
7684 (compute_cases_per_edge): Likewise.
7685 (expand_case): Build list of simple_case_node.
7686 (expand_sjlj_dispatch_table): Use it.
7687 * tree-switch-conversion.c (struct case_node): Moved from
7688 stmt.c and adjusted.
7689 (emit_case_nodes): Likewise.
7690 (node_has_low_bound): Likewise.
7691 (node_has_high_bound): Likewise.
7692 (node_is_bounded): Likewise.
7693 (case_values_threshold): Likewise.
7694 (reset_out_edges_aux): Likewise.
7695 (compute_cases_per_edge): Likewise.
7696 (add_case_node): Likewise.
7697 (dump_case_nodes): Likewise.
7698 (balance_case_nodes): Likewise.
7699 (expand_switch_as_decision_tree_p): Likewise.
7700 (emit_jump): Likewise.
7701 (emit_case_decision_tree): Likewise.
7702 (try_switch_expansion): Likewise.
7703 (do_jump_if_equal): Likewise.
7704 (emit_cmp_and_jump_insns): Likewise.
7705 (fix_phi_operands_for_edge): New function.
7706 (record_phi_operand_mapping): Likewise.
7707 (class pass_lower_switch): New pass.
7708 (pass_lower_switch::execute): New function.
7709 (make_pass_lower_switch): Likewise.
7710 (conditional_probability):
7711 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
7712 * tree-pass.h: Add make_pass_lower_switch.
7714 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
7717 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
7720 2017-08-29 Richard Biener <rguenther@suse.de>
7722 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
7723 we do not add a DW_AT_inline attribute twice.
7724 (gen_subprogram_die): Remove code setting DW_AT_inline on
7725 DECL_ABSTRACT_P nodes.
7727 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
7729 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
7730 calls to internal functions.
7731 (gimplify_modify_expr): Likewise.
7732 * tree-call-cdce.c (use_internal_fn): Likewise.
7733 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
7734 (convert_to_divmod): Set the nothrow flag.
7735 * tree-if-conv.c (predicate_mem_writes): Likewise.
7736 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7737 (vectorizable_call): Likewise.
7738 (vectorizable_store): Likewise.
7739 (vectorizable_load): Likewise.
7740 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
7741 (vect_recog_mask_conversion_pattern): Likewise.
7743 2017-08-29 Martin Liska <mliska@suse.cz>
7746 * gcc.c (driver_handle_option): Add new argument.
7747 * opts-common.c (handle_option): Pass
7748 target_option_override_hook.
7749 * opts-global.c (lang_handle_option): Add new option.
7750 (set_default_handlers): Add new argument.
7751 (decode_options): Likewise.
7752 * opts.c (target_handle_option): Likewise.
7753 (common_handle_option): Call target_option_override_hook.
7754 * opts.h (struct cl_option_handler_func): Add hook for
7755 target option override.
7756 (struct cl_option_handlers): Likewise.
7757 (set_default_handlers): Add new argument.
7758 (decode_options): Likewise.
7759 (common_handle_option): Likewise.
7760 (target_handle_option): Likewise.
7761 * toplev.c (toplev::main): Pass targetm.target_option.override
7764 2017-08-29 Richard Biener <rguenther@suse.de>
7765 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7767 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
7768 life to the active subtree.
7770 2017-08-28 Jeff Law <law@redhat.com>
7772 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
7773 derive_equivalences.
7774 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
7776 (edge_info::derive_equivalences): New private member function
7778 * tree-ssa-dom.c (class edge_info): Changed from a struct
7779 to a class. Add ctor/dtor, methods and data members.
7780 (edge_info::edge_info): Renamed from allocate_edge_info.
7781 Initialize additional members.
7782 (edge_info::~edge_info): New.
7783 (free_dom_edge_info): Delete the edge info.
7784 (record_edge_info): Use new class & associated member functions.
7785 Tighten forms for testing for edge equivalences.
7786 (record_temporary_equivalences): Iterate over the simple
7787 equivalences rather than assuming there's only one per edge.
7788 (cprop_into_successor_phis): Iterate over the simple
7789 equivalences rather than assuming there's only one per edge.
7790 (optimize_stmt): Use operand_equal_p rather than pointer
7791 equality for mini-DSE code.
7793 2017-08-28 Nathan Sidwell <nathan@acm.org>
7795 * gcc.c (execute): Fold SIGPIPE handling into switch
7796 statement. Adjust internal error message.
7798 2017-08-28 Richard Biener <rguenther@suse.de>
7801 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
7802 Do nothing for removed DIEs.
7804 2017-08-28 Richard Biener <rguenther@suse.de>
7806 PR tree-optimization/81977
7807 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
7810 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
7813 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
7814 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
7815 using targetm.gen_mem_thread_fence.
7817 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
7820 * config/i386/i386.md (*<btsc><mode>): Change operand 2
7821 predicate to register_operand. Reorder operands.
7822 (*btr<mode>): Ditto.
7823 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
7824 (*btr<mode>_mask): Ditto.
7826 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
7828 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
7829 * config/rs6000/xmmintrin.h: New file.
7830 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
7832 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7835 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
7836 parameter and_insn and return it.
7837 (recombine_lvx_pattern): Insert a copy to ensure availability of
7838 the base register of the copied masking operation at the point of
7839 the instruction replacement.
7840 (recombine_stvx_pattern): Likewise.
7842 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
7844 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
7845 undocumented switches.
7846 (-mpower9-dform-vector): Likewise.
7847 (-mpower9-dform): Likewise.
7848 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
7849 comments to delete references to -mpower9-dform* switches.
7850 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
7851 Delete reference to -mpower9-dform* switches, test for
7852 -mpower9-vector instead.
7853 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
7854 (OTHER_P9_VECTOR_MASKS): Likewise.
7855 (POWERPC_MASKS): Likewise.
7856 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
7857 tests against -mpower9-dform* to -mpower9-vector. Delete code
7858 that checked for -mpower9-dform* consistancy with other options.
7859 Add test for -mpower9-misc to enable other power9 switches.
7860 (rs6000_init_hard_regno_mode_ok): Likewise.
7861 (rs6000_option_override_internal): Likewise.
7862 (rs6000_emit_prologue): Likewise.
7863 (rs6000_emit_epilogue): Likewise.
7864 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
7865 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
7866 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
7868 (emit_fusion_p9_store): Likewise.
7869 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
7870 resetting these macros if the assembler does not support ISA 3.0
7872 (TARGET_P9_DFORM_VECTOR): Likewise.
7873 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
7874 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
7876 2017-08-25 Alan Modra <amodra@gmail.com>
7878 PR rtl-optimization/81747
7879 * cse.c (cse_extended_basic_block): Don't attempt to record
7880 equivalences for degenerate conditional branches that branch
7881 to their fall-through.
7883 2017-08-24 Martin Sebor <msebor@redhat.com>
7886 * gimple-fold.c (size_must_be_zero_p): New function.
7887 (gimple_fold_builtin_memory_op): Call it.
7889 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
7891 * config/rs6000/mm_malloc.h: New file.
7893 2017-08-24 Bin Cheng <bin.cheng@arm.com>
7895 PR tree-optimization/81913
7896 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
7897 analysis when either IVs in condition can wrap.
7899 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
7901 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
7902 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
7904 2017-08-24 Richard Biener <rguenther@suse.de>
7907 * targhooks.c (default_target_can_inline_p): Properly
7908 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
7909 is present and always compare.
7910 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
7911 infer -mfpmath=sse from TARGET_SSE_P.
7912 (ix86_can_inline_p): Properly use target_option_default_node when
7913 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
7915 2017-08-24 Richard Biener <rguenther@suse.de>
7918 * dwarf2out.c (output_die): Handle flag_generate_offload like
7920 (output_comp_unit): Likewise.
7921 (gen_array_type_die): Likewise.
7922 (dwarf2out_early_finish): Likewise.
7923 (note_variable_value_in_expr): Likewise.
7924 (dwarf2out_finish): Likewise. Adjust assert.
7925 * cgraphunit.c (symbol_table::compile): Move setting of
7926 flag_generate_offload earlier ...
7927 (symbol_table::finalize_compilation_unit): ... here, before
7928 early debug finalization.
7930 2017-08-24 Richard Biener <rguenther@suse.de>
7932 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
7933 and ipa-fnsummary.h.
7934 (ix86_can_inline_p): When ix86_fpmath flags do not match
7935 check whether the callee uses FP math at all.
7937 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
7940 * tree-ssanames.c (get_nonzero_bits): Use element_precision
7941 instead of TYPE_PRECISION.
7943 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
7944 Alan Hayward <alan.hayward@arm.com>
7945 David Sherwood <david.sherwood@arm.com>
7947 * combine.c (make_extraction): Use subreg_offset_from_lsb.
7949 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
7951 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
7953 (ix86_frame::stack_realign_allocate): New field.
7954 (struct machine_frame_state): Modify comments.
7955 (machine_frame_state::sp_realigned_fp_end): New field.
7956 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
7958 (sp_valid_at): Add assertion to assure no attempt to access invalid
7959 offset of a realigned stack.
7960 (fp_valid_at): Likewise.
7961 (choose_baseaddr): Modify comments.
7962 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
7963 ix86_expand_prologue.
7964 (ix86_expand_prologue): Modify stack realignment and allocation.
7965 (ix86_expand_epilogue): Modify comments.
7966 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
7967 avx2_runtime, avx512f, and avx512f_runtime.
7969 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
7971 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
7972 (mstackrealign): Do not init to -1.
7973 * config/i386/i386.c (ix86_option_override_internal):
7974 Check opts_set, not opts when setting default value of
7975 opts->x_ix86_force_align_arg_pointer.
7977 2017-08-23 Richard Biener <rguenther@suse.de>
7979 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
7980 lang_hooks.decl_printable_name.
7981 * print-rtl-function.c (print_rtx_function): Likewise.
7982 * tree-pretty-print.c (dump_function_header): Likewise.
7984 2017-08-23 Richard Biener <rguenther@suse.de>
7987 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
7988 -g0 at compile-time.
7990 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7993 * doc/sourcebuild.texi (Other hardware attributes):
7996 2017-08-23 Tamar Christina <tamar.christina@arm.com>
7999 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
8000 Fix single-use check.
8002 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
8004 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
8006 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8008 * doc/install.texi: Modify to add more details on running selected
8011 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8013 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
8014 is combined with -mabi=ms.
8015 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
8018 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8020 PR tree-optimization/81488
8021 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
8022 and cached_basis fields.
8023 (MAX_SPREAD): New constant.
8024 (alloc_cand_and_find_basis): Initialize new fields.
8025 (clear_visited): New function.
8026 (create_phi_basis_1): Rename from create_phi_basis, set visited
8027 and cached_basis fields.
8028 (create_phi_basis): New wrapper function.
8029 (phi_add_costs_1): Rename from phi_add_costs, add spread
8030 parameter, set visited field, short-circuit when limits reached.
8031 (phi_add_costs): New wrapper function.
8032 (record_phi_increments_1): Rename from record_phi_increments, set
8034 (record_phi_increments): New wrapper function.
8035 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
8036 (phi_incr_cost): New wrapper function.
8037 (all_phi_incrs_profitable_1): Rename from
8038 all_phi_incrs_profitable, set visited field.
8039 (all_phi_incrs_profitable): New wrapper function.
8041 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
8042 Alan Hayward <alan.hayward@arm.com>
8043 David Sherwood <david.sherwood@arm.com>
8045 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
8046 that takes the outer and inner modes.
8047 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
8048 comparison as the canonical test for a paradoxical subreg.
8049 * combine.c (simplify_set): Use paradoxical_subreg_p.
8050 (make_extraction): Likewise.
8051 (force_to_mode): Likewise.
8052 (rtx_equal_for_field_assignment_p): Likewise.
8053 (gen_lowpart_for_combine): Likewise.
8054 (simplify_comparison): Likewise.
8055 * cse.c (equiv_constant): Likewise.
8056 * expmed.c (store_bit_field_1): Likewise.
8057 * final.c (alter_subreg): Likewise.
8058 * fwprop.c (propagate_rtx): Likewise.
8059 (forward_propagate_subreg): Likewise.
8060 * ira-conflicts.c (ira_build_conflicts): Likewise.
8061 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
8062 * lra-constraints.c (curr_insn_transform): Likewise.
8063 (split_reg): Likewise.
8064 * lra-eliminations.c (move_plus_up): Likewise.
8065 (lra_eliminate_regs_1): Likewise.
8066 * recog.c (general_operand): Likewise.
8067 * ree.c (combine_reaching_defs): Likewise.
8068 * reload.c (push_reload): Likewise.
8069 (find_reloads): Likewise.
8070 * reload1.c (elimination_effects): Likewise.
8071 (compute_reload_subreg_offset): Likewise.
8072 (choose_reload_regs): Likewise.
8073 * rtlanal.c (subreg_lsb_1): Likewise.
8074 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8075 (simplify_subreg): Likewise.
8076 * var-tracking.c (track_loc_p): Likewise.
8077 * emit-rtl.c (byte_lowpart_offset): Likewise.
8078 (paradoxical_subreg_p): Delete out-of-line definition.
8080 2017-08-22 Jeff Law <law@redhat.com>
8082 PR tree-optimization/81741
8083 PR tree-optimization/71947
8084 * tree-ssa-dom.c: Include tree-inline.h.
8085 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
8086 equivalences if one is more expensive to compute than the other.
8087 * tree-ssa-scopedtables.h (class const_or_copies): Make
8088 record_const_or_copy_raw method private.
8089 (class avail_exprs_stack): New method simplify_binary_operation.
8090 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
8091 avail_exprs_stack::simplify_binary_operation as needed.
8092 (avail_exprs_stack::simplify_binary_operation): New function.
8094 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
8096 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
8097 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
8098 (DOT_SYMBOLS): Likewise.
8099 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
8100 (RELOCATABLE_NEEDS_FIXUP): Likewise.
8101 (RS6000_ABI_NAME): Likewise.
8102 (TARGET_CMODEL): Likewise.
8103 (TOC_SECTION_ASM_OP): Likewise.
8104 (SET_CMODEL): New macro.
8105 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
8107 2017-08-22 Richard Biener <rguenther@suse.de>
8109 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
8110 to allow for free-lang-data replacements similar to verify_type_variant.
8112 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
8113 Alan Hayward <alan.hayward@arm.com>
8114 David Sherwood <david.sherwood@arm.com>
8116 * config/aarch64/aarch64.md (casesi): Use DImode rather than
8117 VOIDmode for the LABEL_REF.
8119 2017-08-22 Richard Biener <rguenther@suse.de>
8121 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
8123 2017-08-22 Richard Biener <rguenther@suse.de>
8125 * common.opt (feliminate-dwarf2-dups): Ignore.
8126 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
8127 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
8128 same_die_p_wrap, compute_section_prefix,
8129 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
8130 (comdat_symbol_id, comdat_symbol_number): Likewise.
8131 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
8133 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
8134 (output_die): Mark unreachable path unreachable.
8135 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
8136 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
8137 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
8138 (dwarf2out_early_finish): Likewise.
8140 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
8142 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
8144 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
8147 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
8148 not VAR_P. Filter attribute warnings with OPT_Wattributes.
8149 (avr_attribute_table) <io, io_low, address>: Initialize
8150 .decl_required with true.
8152 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
8154 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
8155 undocumented debugging options.
8156 (-mvsx-scalar-double): Likewise.
8157 (-mallow-df-permute): Likewise.
8158 (-mvectorize-builtins): Likewise.
8159 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8160 (rs6000_builtin_vectorized_function): Likewise.
8161 (rs6000_builtin_md_vectorized_function): Likewise.
8162 (rs6000_opt_vars): Likewise.
8164 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
8167 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
8168 (*btrq_imm): Rename from *btrq.
8169 (*btcq_imm): Rename from *btcq.
8170 (btsc): New code attribute.
8171 (*<btsc><mode>): New insn pattern.
8172 (*btr<mode>): Ditto.
8173 (*<btsc><mode>_mask): New insn_and_split pattern.
8174 (*btr<mode>_mask): Ditto.
8176 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8177 Alan Hayward <alan.hayward@arm.com>
8178 David Sherwood <david.sherwood@arm.com>
8180 * function.c (pad_below): Simplify padding calculation.
8182 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8183 Alan Hayward <alan.hayward@arm.com>
8184 David Sherwood <david.sherwood@arm.com>
8186 * target.def (function_prologue): Remove frame size argument.
8187 (function_epilogue): Likewise.
8188 * doc/tm.texi: Regenerate.
8189 * final.c (final_start_function): Update call to function_prologue.
8190 (final_end_function): Update call to function_epilogue.
8191 (default_function_pro_epilogue): Remove frame size argument.
8192 * output.h (default_function_pro_epilogue): Likewise.
8193 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
8194 (arm_output_function_prologue): Likewise.
8195 * config/frv/frv.c (frv_function_prologue): Likewise.
8196 (frv_function_epilogue): Likewise.
8197 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
8198 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
8199 (ia64_output_function_epilogue): Likewise.
8200 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
8201 (m32r_output_function_epilogue): Likewise.
8202 * config/microblaze/microblaze.c (microblaze_function_prologue)
8203 (microblaze_function_epilogue): Likewise.
8204 * config/mips/mips.c (mips_output_function_prologue): Likewise.
8205 (mips_output_function_epilogue): Likewise.
8206 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
8207 (mmix_target_asm_function_epilogue): Likewise.
8208 * config/msp430/msp430.c (msp430_start_function): Likewise.
8209 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
8210 (nds32_asm_function_epilogue): Likewise.
8211 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
8212 * config/pa/pa.c (pa_output_function_prologue): Likewise.
8213 (pa_output_function_epilogue): Likewise.
8214 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
8215 (rs6000_output_function_epilogue): Likewise.
8216 * config/rl78/rl78.c (rl78_start_function): Likewise.
8217 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
8218 (rs6000_output_function_epilogue): Likewise.
8219 * config/rx/rx.c (rx_output_function_prologue): Likewise.
8220 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
8221 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
8222 (sparc_asm_function_epilogue): Likewise.
8224 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8226 * tree.h (type_has_mode_precision_p): New function.
8227 * convert.c (convert_to_integer_1): Use it.
8228 * expr.c (expand_expr_real_2): Likewise.
8229 (expand_expr_real_1): Likewise.
8230 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
8231 * match.pd: Likewise.
8232 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
8233 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
8234 * tree-tailcall.c (process_assignment): Likewise.
8235 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8236 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
8237 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
8238 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8239 (vectorizable_assignment): Likewise.
8240 (vectorizable_shift): Likewise.
8241 (vectorizable_operation): Likewise.
8242 * tree-vrp.c (register_edge_assert_for_2): Likewise.
8244 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
8246 * match.pd: Add pow (C, x) simplification.
8248 2017-08-21 Richard Biener <rguenther@suse.de>
8250 PR tree-optimization/81900
8251 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
8252 for blocks with abnormal predecessors.
8253 (compute_antic): Do not set visited flag prematurely.
8255 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
8258 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
8260 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8262 * stor-layout.h (vector_type_mode): Move to...
8263 * tree.h (vector_type_mode): ...here.
8264 * stor-layout.c (vector_type_mode): Move to...
8265 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
8267 2017-08-21 Richard Biener <rguenther@suse.de>
8269 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
8270 register_external_die hooks.
8271 (debug_false_tree_charstarstar_uhwistar): Declare.
8272 (debug_nothing_tree_charstar_uhwi): Likewise.
8273 * debug.c (do_nothing_debug_hooks): Adjust.
8274 (debug_false_tree_charstarstar_uhwistar): New do nothing.
8275 (debug_nothing_tree_charstar_uhwi): Likewise.
8276 * dbxout.c (dbx_debug_hooks): Adjust.
8277 (xcoff_debug_hooks): Likewise.
8278 * sdbout.c (sdb_debug_hooks): Likewise.
8279 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
8280 * dwarf2out.c (macinfo_label_base): New global.
8281 (dwarf2out_register_external_die): New function for the
8282 register_external_die hook.
8283 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
8284 (dwarf2_debug_hooks): Use them.
8285 (dwarf2_lineno_debug_hooks): Adjust.
8286 (struct die_struct): Add with_offset flag.
8287 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
8288 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
8289 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
8290 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
8291 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
8292 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
8293 defining section names for the early LTO debug variants.
8294 (reset_indirect_string): New helper.
8295 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
8296 (print_dw_val): Add support for offsetted symbol references.
8297 (get_ultimate_context): Split out from is_cxx.
8298 (is_cxx): Use get_ultimate_context.
8299 (is_fortran): Add decl overload.
8300 (compute_comp_unit_symbol): Split out worker from
8301 compute_section_prefix.
8302 (compute_section_prefix): Call compute_comp_unit_symbol and
8303 set comdat_type_p here.
8304 (output_die): Skip DIE symbol output for the LTO added one.
8305 Handle DIE symbol references with offset.
8306 (output_comp_unit): Guard section name mangling properly.
8307 For LTO debug sections emit a symbol at the section beginning
8308 which we use to refer to its DIEs.
8309 (add_abstract_origin_attribute): For DIEs registered via
8310 dwarf2out_register_external_die directly refer to the early
8311 DIE rather than indirectly through the shadow one we created.
8312 Remove obsolete call to dwarf2out_abstract_function for
8313 non-function/block origins.
8314 (gen_array_type_die): When generating early LTO debug do
8315 not emit DW_AT_string_length.
8316 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
8317 late when in LTO. As suggested place a gcc_unreachable for
8318 the DECL_ABSTRACT_P case.
8319 (gen_subprogram_die): Avoid another specification DIE
8320 for early built declarations/definitions for the late LTO case.
8321 (gen_variable_die): Add type references for late duplicated VLA dies
8323 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
8324 we have the abstract instance already.
8325 (process_scope_var): Adjust decl DIE contexts in LTO which
8326 first puts them in limbo.
8327 (gen_decl_die): Do not generate type DIEs late apart from
8328 types for VLAs or for decls we do not yet have a DIE. Do not
8329 call dwarf2out_abstract_function late.
8330 (dwarf2out_early_global_decl): Make sure to create DIEs
8331 for abstract instances of a decl first.
8332 (dwarf2out_late_global_decl): Adjust comment.
8333 (output_macinfo_op): With multiple macro sections use
8334 macinfo_label_base to distinguish labels.
8335 (output_macinfo): Likewise. Update macinfo_label_base.
8336 Pass in the line info label.
8337 (note_variable_value_in_expr): When generating LTO resolve
8338 all variable values here by generating DIEs as needed.
8339 (init_sections_and_labels): Add early LTO debug flag parameter
8340 and generate different sections and names if set. Add generation
8341 counter for the labels so we can have multiple of them.
8342 (reset_dies): Helper to allow DIEs to be output multiple times.
8343 (dwarf2out_finish): When outputting DIEs to the fat part of an
8344 LTO object first reset DIEs.
8345 (dwarf2out_early_finish): Output early DIEs when generating LTO.
8346 (modified_type_die): Check for decl_ultimate_origin being self
8348 (gen_type_die_with_usage): Likewise.
8349 (gen_typedef_die): Allow decl_ultimate_origin being self.
8350 (set_decl_abstract_flags): Remove.
8351 (set_block_abstract_flags): Likewise.
8352 (dwarf2out_abstract_function): Treat the early generated DIEs
8353 as the abstract copy and only add DW_AT_inline and
8354 DW_AT_artificial here and call set_decl_origin_self.
8355 If the DIE has an abstract origin don't do anything.
8356 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
8357 if we have none yet (Go fails to build one, PR78628).
8358 (variably_modified_type_p): Prevent endless recursion for Ada
8359 cyclic pointer types.
8360 * lto-streamer-in.c: Include debug.h.
8361 (dref_queue): New global.
8362 (lto_read_tree_1): Stream in DIE references.
8363 (lto_input_tree): Register DIE references.
8364 (input_function): Stream DECL_DEBUG_ARGS.
8365 * lto-streamer-out.c: Include debug.h.
8366 (lto_write_tree_1): Output DIE references.
8367 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
8368 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
8369 (output_function): Stream DECL_DEBUG_ARGS.
8370 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
8371 Stream DECL_ABSTRACT_ORIGIN.
8372 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
8373 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
8374 DECL_CONTEXT for file-scope decls.
8375 * lto-streamer.h (struct dref_entry): Declare.
8376 (dref_queue): Likewise.
8377 * cfgexpand.c (pass_expand::execute): Do not call the
8378 outlining_inline_function hook here.
8379 * lto-wrapper.c (debug_obj): New global.
8380 (tool_cleanup): Unlink it if required.
8381 (debug_objcopy): New function.
8382 (run_gcc): Handle early debug sections in the IL files by
8383 extracting them to separate files, partially linkin them and
8384 feeding the result back as result to the linker.
8385 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
8386 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
8387 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
8388 sections into a separate segment.
8389 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
8391 (darwin_asm_dwarf_section): Likewise.
8392 (darwin_asm_output_dwarf_offset): Likewise.
8393 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
8395 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8396 Alan Hayward <alan.hayward@arm.com>
8397 David Sherwood <david.sherwood@arm.com>
8399 * read-md.h (md_reader::record_potential_iterator_use): Replace
8400 pointer argument with an rtx and an index.
8401 * read-rtl.c (iterator_group::apply_iterator): Likewise.
8402 (apply_mode_iterator): Likewise.
8403 (apply_code_iterator): Likewise.
8404 (apply_int_iterator): Likewise.
8405 (apply_subst_iterator): Likewise.
8406 (record_iterator_use): Likewise.
8407 (record_attribute_use): Likewise.
8408 (md_reader::record_potential_iterator_use): Likewise. Update calls
8409 to record_iterator_use and apply_iterator.
8410 (iterator_use): Replace ptr with x and index.
8411 (attribute_use): Likewise.
8412 (apply_attribute_uses): Update calls to apply_iterator.
8413 (apply_iterators): Likewise. Update initialization of iterator_use.
8414 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
8415 and record_potential_iterator_use.
8416 (rtx_reader::read_rtx_operand): Likewise.
8418 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8419 Alan Hayward <alan.hayward@arm.com>
8420 David Sherwood <david.sherwood@arm.com>
8422 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
8425 2017-08-21 Richard Biener <rguenther@suse.de>
8428 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
8429 at struct end conservatively when comparing common bases.
8431 2017-08-21 Richard Biener <rguenther@suse.de>
8433 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
8434 (mem_ref_in_stmt): Remove.
8435 (determine_max_movement): Use ref index to get at the reference.
8436 (invariantness_dom_walker::before_dom_children): Deal with
8437 lim data already initialized.
8438 (gather_mem_refs_stmt): Initialize lim data and record ref index.
8440 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
8442 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
8443 (TARGET_ISA_ROUND): Ditto.
8444 (TARGET_ROUND): Ditto.
8445 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
8446 * config/i386/i386.md: Ditto.
8447 * config/i386/sse.md: Ditto.
8448 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
8449 with OPTION_MASK_ISA_SSE4_1.
8451 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
8454 * doc/extend.texi (x86 Built-in Functions): Correct the name of
8455 __builtin_ia32_lzcnt_u16.
8457 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
8460 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
8461 (rs6000_set_current_function): Rewrite function to use it.
8463 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
8466 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
8467 and TYPE_WARN_IF_NOT_ALIGN.
8468 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
8469 (handle_warn_if_not_align): New.
8470 (place_union_field): Call handle_warn_if_not_align.
8471 (place_field): Call handle_warn_if_not_align.
8472 Copy TYPE_WARN_IF_NOT_ALIGN.
8473 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
8474 (layout_type): Likewise.
8475 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
8477 (tree_decl_common): Add warn_if_not_align.
8478 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
8479 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
8480 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
8481 (DECL_WARN_IF_NOT_ALIGN): Likewise.
8482 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
8483 * doc/extend.texi: Document warn_if_not_aligned attribute.
8484 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
8486 2017-08-17 Martin Liska <mliska@suse.cz>
8489 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
8490 (get_data_dependence): Use it as pointer type.
8491 (distribute_loop): Likewise.
8493 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8495 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
8496 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
8497 (p8_vmrgow_<mode>_direct): New define_insn.
8498 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
8499 handle endianness for vmrgew and vmrgow permute patterns.
8501 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
8503 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
8504 * config/rs6000/rs6000-cpus.def: Remove comment.
8505 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
8506 (POWERPC_MASKS): Likewise.
8507 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
8508 use of TARGET_VSX_TIMODE.
8509 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
8510 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
8511 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
8512 (rs6000_option_override_internal): Remove dead code.
8513 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
8514 (rs6000_legitimize_reload_address): Likewise.
8515 (rs6000_legitimate_address_p): Likewise.
8516 (rs6000_opt_masks): Delete "vsx-timode".
8517 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
8518 from function comment.
8519 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
8520 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
8521 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
8523 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
8524 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
8525 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
8528 (define_peephole2): Likewise.
8530 2017-08-17 Martin Sebor <msebor@redhat.com>
8533 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
8534 past the end of an array.
8535 (test_pp_format): Add test cases.
8537 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
8539 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
8540 missing ECF_NOTHROW flags.
8542 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
8545 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
8546 operands residing in integer registers.
8547 (*vsx_le_perm_load_<mode>): Likewise.
8548 (*vsx_le_perm_store_<mode>): Likewise.
8549 (define_peephole2): Add peepholes to optimize the above.
8551 2017-08-17 Marek Polacek <polacek@redhat.com>
8554 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
8555 to mimic what shorten_compare did. Change the return type to bool.
8556 (fold_cond_expr_with_comparison): Update call to
8557 operand_equal_for_comparison_p.
8558 (fold_ternary_loc): Likewise.
8560 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
8562 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
8564 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
8566 2017-08-17 Richard Biener <rguenther@suse.de>
8568 * tree-ssa-structalias.c (solve_graph): When propagating
8569 to successors update the graphs succ edges and avoid duplicate work.
8571 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
8574 * config/i386/i386.c (ix86_option_override_internal): Save target
8575 specific options after ix86_stack_protector_guard_reg was changed.
8577 2017-08-17 Richard Biener <rguenther@suse.de>
8579 PR tree-optimization/81827
8580 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
8581 (new_var_info): Initialize it conservatively.
8582 (get_call_vi): Mark register vars.
8583 (new_scalar_tmp_constraint_exp): Likewise.
8584 (handle_rhs_call): Likewise.
8585 (handle_const_call): Likewise.
8586 (create_function_info_for): Likewise.
8587 (solve_constraints): Sort varinfos to separate register from
8588 non-register vars to pack points-to solution bitmaps during
8591 2017-08-17 Marek Polacek <polacek@redhat.com>
8593 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
8595 2017-08-17 Richard Biener <rguenther@suse.de>
8597 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
8598 to true when overflow is undefined and we saturated the result.
8600 2017-08-17 Alan Modra <amodra@gmail.com>
8603 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
8604 Don't use store multiple if only one reg needs saving.
8605 (interesting_frame_related_regno): New function.
8606 (rs6000_frame_related): Don't emit frame info for regs that
8608 (rs6000_emit_epilogue): Likewise.
8610 2017-08-16 Nathan Sidwell <nathan@acm.org>
8612 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
8613 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
8614 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
8615 * tree.c (free_lang_data_in_type): Use else-if chain. Always
8616 clear TYPE_LANG_1. Remove obsolete member-function stripping.
8617 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
8618 (verify_type): Adjust for TYPE_BINFO move.
8619 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
8620 process TYPE_BINFO directly.
8621 (hash_tree): Likewise.
8622 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
8624 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
8627 2017-08-16 David Malcolm <dmalcolm@redhat.com>
8629 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
8631 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
8634 * config/i386/i386.md (*anddi_1_btr): Change predicates of
8635 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
8636 Add ix86_binary_operator_ok to insn constraint.
8637 (*iordi_1_bts): Ditto.
8638 (*xordi_1_btc): Ditto.
8639 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
8640 Update corresponding peephole2 pattern.
8644 2017-08-16 Bin Cheng <bin.cheng@arm.com>
8646 PR tree-optimization/81832
8647 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
8648 copy loop header which has IFN_LOOP_DIST_ALIAS call.
8650 2017-08-16 Marek Polacek <polacek@redhat.com>
8653 * fold-const.c (fold_indirect_ref_1): Restore original behavior
8654 regarding size_zero_node.
8656 2017-08-16 Martin Liska <mliska@suse.cz>
8659 * config.gcc: Respect previously set extra_objs in case
8662 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
8664 PR tree-optimization/81835
8665 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
8666 the phi SSA_NAME. Check that the condition in a COND_EXPR does
8667 not depend on the phi.
8669 2017-08-16 Alan Modra <amodra@gmail.com>
8671 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
8674 2017-08-16 Alan Modra <amodra@gmail.com>
8676 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
8677 (save_reg_p): ..into this. Update all callers.
8678 (first_reg_to_save): Simplify.
8680 2017-08-16 Alan Modra <amodra@gmail.com>
8682 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
8685 2017-08-15 Joseph Myers <joseph@codesourcery.com>
8689 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
8690 constant count if that count is less than 32.
8692 2017-08-15 Nathan Sidwell <nathan@acm.org>
8694 * gcc.c (execute): Emit friendlier message if inferior is killed
8695 by an external cause.
8697 2017-08-15 Richard Biener <rguenther@suse.de>
8699 PR tree-optimization/81790
8700 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
8701 CONSTRUCTORs from simplifying and VN.
8703 2017-08-14 Martin Sebor <msebor@redhat.com>
8705 * builtin-attrs.def: Add comments.
8707 2017-08-14 Martin Sebor <msebor@redhat.com>
8710 * doc/extend.texi (attribute nonstring): Document new attribute.
8712 2017-08-14 Martin Sebor <msebor@redhat.com>
8715 * tree-diagnostic.c (default_tree_printer): Handle %G.
8716 * gimple-pretty-print.h (percent_G_format): Declare new function.
8717 * gimple-pretty-print.c (percent_G_format): Define.
8718 * tree-pretty-print.c (percent_K_format): Add argument.
8720 2017-08-14 Martin Sebor <msebor@redhat.com>
8722 PR translation/79998
8723 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
8724 Remove a stray space.
8726 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
8729 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
8730 (*iordi_1_bts): Ditto.
8731 (*xordi_1_btc): Ditto.
8733 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8736 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
8737 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8739 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
8740 quoted strings, and make more translator-friendly.
8741 (darwin_rs6000_override_options): Likewise.
8742 (rs6000_option_override_internal): Likewise.
8743 (rs6000_return_in_memory): Fix overlong line.
8744 (init_cmulative_args): Use quoted strings, and make more
8745 translator-friendly.
8746 (rs6000_pass_by_reference): Fix overlong line.
8747 (def_builtin): Use quoted strings.
8748 (altivec_expand_predicate_builtin): Use quoted strings, and make
8749 more translator-friendly.
8750 (htm_expand_builtin): Use quoted strings.
8751 (cpu_expand_builtin): Use quoted strings, and make more
8752 translator-friendly.
8753 (altivec_expand_builtin): Likewise.
8754 (paired_expand_predicate_builtin): Likewise.
8755 (rs6000_invalid_builtin): Likewise.
8756 (builtin_function_type): Use quoted strings.
8757 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
8758 more translator-friendly.
8759 (rs6000_trampoline_init): Likewise.
8760 (rs6000_handle_altivec_attribute): Likewise.
8761 (rs6000_inner_target_options): Use quoted strings.
8762 (rs6000_disable_incompatible_switches): Likewise.
8763 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
8764 strings, and make more translator-friendly.
8765 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
8767 2017-08-14 Bin Cheng <bin.cheng@arm.com>
8769 PR tree-optimization/81799
8770 * tree-loop-distribution.c (version_loop_by_alias_check): Force
8771 cond_expr to simple gimple operand.
8773 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
8776 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
8778 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
8782 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
8783 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
8784 TARGET_GASISR_PROLOGUES.
8785 * config/avr/avr.c (avr_option_override): Same.
8786 (avr_pass_pre_proep::execute): Same.
8788 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
8791 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
8792 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
8794 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
8796 * config/i386/i386.md (*load_tp_<mode>): Redefine as
8797 define_insn_and_split. Split to a memory load from 0 in
8798 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
8799 using PTR mode iterator.
8800 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
8801 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
8802 (*add_tp_<mode>): Redefine as define_insn_and_split.
8803 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
8804 address space. Merge with *add_tp_x32 using PTR mode iterator.
8805 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
8806 Split to an add with a memory load from 0 in
8807 DEFAULT_TLS_SEG_REG address space.
8809 2017-08-12 Andrew Pinski <apinski@cavium.com>
8811 * config/aarch64/aarch64-option-extensions.def (rdma):
8812 Fix feature string to what Linux prints out in /proc/cpuinfo.
8814 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
8817 * dwarf2out.c (modified_type_die): For C typedef types that have
8818 an ultimate origin, process the ultimate origin instead of the
8820 (gen_typedef_die): Assert that input DECLs have no ultimate
8822 (gen_type_die_with_usage): For typedef variants that have an
8823 ultimate origin, just call gen_decl_die on the original DECL.
8824 (process_scope_var): Avoid creating DIEs for local typedefs and
8825 concrete static variables.
8827 2017-08-12 Alan Modra <amodra@gmail.com>
8831 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
8832 match gnu-user.h startfile.
8833 (ENDFILE_LINUX_SPEC): Similarly.
8835 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
8838 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
8840 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
8842 2017-08-11 Tamar Christina <tamar.christina@arm.com>
8843 * config/aarch64/aarch64.md (mov<mode>): Change.
8844 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
8845 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
8846 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
8848 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
8850 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
8851 for storage order barriers.
8853 2017-08-11 Martin Liska <mliska@suse.cz>
8856 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
8857 variables of void type.
8859 2017-08-11 Martin Liska <mliska@suse.cz>
8861 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
8862 TARGET_SUPPORTS_ALIASES.
8863 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
8864 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
8865 (optimize_weakref): Likewise.
8866 * symtab.c (symtab_node::noninterposable_alias): Likewise.
8867 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
8868 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
8870 2017-08-11 Martin Liska <mliska@suse.cz>
8873 * config/i386/i386.c (make_resolver_func): Do complete
8874 refactoring of the function.
8876 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8879 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
8880 * config/i386/i386.c (ix86_stack_protect_guard): Use
8881 ix86_stack_protect_guard_symbol_str to generate varible declaration.
8882 * doc/invoke.texi (x86 Options): Document
8883 -mstack-protector-guard-symbol= option.
8885 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
8887 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
8888 * config/i386/i386.c (ix86_split_stack_guard): New function.
8889 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
8890 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
8891 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
8892 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
8893 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
8894 (split_stack_space_check): Call ix86_split_stack_guard.
8896 2017-08-10 Martin Sebor <msebor@redhat.com>
8898 * print-tree.c (print_node): Print location using the established
8900 Replace spaces with colons.
8901 (debug_raw, debug): Ditto.
8903 2017-08-10 Martin Sebor <msebor@redhat.com>
8906 * pretty-print.c (pp_format): Correct the handling of %s precision.
8908 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
8911 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
8913 (ix86_finalize_stack_frame_flags): This. Also clear
8914 frame_pointer_needed if -fno-omit-frame-pointer is used without
8916 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
8917 with ix86_finalize_stack_frame_flags.
8918 (ix86_expand_epilogue): Likewise.
8919 (ix86_expand_split_stack_prologue): Likewise.
8920 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
8922 2017-08-10 Martin Liska <mliska@suse.cz>
8925 * c-attribs.c (handle_target_attribute):
8926 Report warning for an empty string argument of target attribute.
8928 2017-08-09 Jakub Jelinek <jakub@redhat.com>
8931 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
8933 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
8934 or DECL_NONLOCAL labels.
8935 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
8936 or DECL_NONLOCAL labels here.
8938 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
8940 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
8941 to indicate when early gimple folding has been disabled.
8942 (rs6000_gimple_fold_builtin): Add debug content.
8943 (rs6000_invalid_builtin): Fix whitespace.
8944 (rs6000_expand_builtin): Fix whitespace.
8945 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
8947 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
8950 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
8951 SAVE_MULTIPLE if not all the registers that saves, should be saved.
8953 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
8955 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
8956 (qdf24xx): Likewise.
8957 * config/aarch64/aarch64.md: Include falkor.md.
8958 * config/aarch64/falkor.md: New.
8960 2017-08-09 Marek Polacek <polacek@redhat.com>
8963 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
8964 * diagnostic.c (emit_diagnostic): Add a comment.
8965 (emit_diagnostic_valist): New function.
8967 2017-08-09 Marek Polacek <polacek@redhat.com>
8970 * input.c (make_location): New overload.
8971 * input.h (make_location): Declare.
8973 2017-08-08 Alan Modra <amodra@gmail.com>
8974 H.J. Lu <hongjiu.lu@intel.com>
8977 * gcc.c (NO_PIE_SPEC): Delete.
8978 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
8980 (LINK_PIE_SPEC): ..to here.
8981 (LINK_COMMAND_SPEC): Support -no-pie.
8982 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
8983 chain of crtbegin*.o selection, update for PIE_SPEC changes and
8985 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
8986 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
8987 (ENDFILE_CRTEND_SPEC): Similarly.
8989 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
8992 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
8993 (mstack-protector-guard-offset=): Ditto.
8994 * config/i386/i386.c (ix86_option_override): Handle
8995 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
8997 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
8998 ix86_stack_protect_guard_offset variables.
8999 (TARGET_STACK_PROTECT_GUARD): Always define.
9000 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
9001 and -mstack-protector-guard-offset= options.
9003 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9005 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
9006 boundary case for the last candidate.
9008 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9010 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
9011 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
9013 2017-08-08 Tamar Christina <tamar.christina@arm.com>
9016 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
9017 * config/aarch64/aarch64-builtins.c
9018 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
9019 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
9020 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
9022 2017-08-08 Tamar Christina <tamar.christina@arm.com>
9023 Andrew Pinski <pinskia@gmail.com>
9026 * internal-fn.def (XORSIGN): New.
9027 * optabs.def (xorsign_optab): New.
9028 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
9029 (convert_expand_mult_copysign): New.
9030 (pass_optimize_widening_mul::execute): Call
9031 convert_expand_mult_copysign.
9033 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9035 PR tree-optimization/81354
9036 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
9037 Insert on edges rather than explicitly creating landing pads.
9038 (analyze_candidates_and_replace): Commit edge inserts.
9040 2017-08-08 Richard Biener <rguenther@suse.de>
9043 * tree-ssa-loop-niter.c: Include tree-dfa.h.
9044 (expand_simple_operations): Also look through ADDR_EXPRs with
9045 MEM_REF bases treating them as POINTER_PLUS_EXPR.
9047 2017-08-08 Richard Biener <rguenther@suse.de>
9049 PR tree-optimization/81723
9050 * tree-vect-slp.c (struct bst_traits): New hash traits.
9051 (bst_fail): New global.
9052 (vect_build_slp_tree_2): New worker, split out from ...
9053 (vect_build_slp_tree): ... this now wrapping it with using
9054 bst_fail set to cache SLP tree build fails. Properly handle
9056 (vect_analyze_slp_instance): Allocate and free bst_fail.
9058 2017-08-08 Martin Liska <mliska@suse.cz>
9061 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
9062 LABEL_DECLs that can be from a different function.
9064 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9066 PR tree-optimization/81744
9067 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
9068 loop's number of iterations.
9070 2017-08-08 Martin Liska <mliska@suse.cz>
9072 * asan.c: Include header files.
9073 * attribs.c (build_decl_attribute_variant): New function moved
9075 (build_type_attribute_qual_variant): Likewise.
9076 (cmp_attrib_identifiers): Likewise.
9077 (simple_cst_list_equal): Likewise.
9078 (omp_declare_simd_clauses_equal): Likewise.
9079 (attribute_value_equal): Likewise.
9080 (comp_type_attributes): Likewise.
9081 (build_type_attribute_variant): Likewise.
9082 (lookup_ident_attribute): Likewise.
9083 (remove_attribute): Likewise.
9084 (merge_attributes): Likewise.
9085 (merge_type_attributes): Likewise.
9086 (merge_decl_attributes): Likewise.
9087 (merge_dllimport_decl_attributes): Likewise.
9088 (handle_dll_attribute): Likewise.
9089 (attribute_list_equal): Likewise.
9090 (attribute_list_contained): Likewise.
9091 * attribs.h (lookup_attribute): New function moved from tree.[ch].
9092 (lookup_attribute_by_prefix): Likewise.
9093 * bb-reorder.c: Include header files.
9094 * builtins.c: Likewise.
9095 * calls.c: Likewise.
9096 * cfgexpand.c: Likewise.
9097 * cgraph.c: Likewise.
9098 * cgraphunit.c: Likewise.
9099 * convert.c: Likewise.
9100 * dwarf2out.c: Likewise.
9101 * final.c: Likewise.
9102 * fold-const.c: Likewise.
9103 * function.c: Likewise.
9104 * gimple-expr.c: Likewise.
9105 * gimple-fold.c: Likewise.
9106 * gimple-pretty-print.c: Likewise.
9107 * gimple.c: Likewise.
9108 * gimplify.c: Likewise.
9109 * hsa-common.c: Likewise.
9110 * hsa-gen.c: Likewise.
9111 * internal-fn.c: Likewise.
9112 * ipa-chkp.c: Likewise.
9113 * ipa-cp.c: Likewise.
9114 * ipa-devirt.c: Likewise.
9115 * ipa-fnsummary.c: Likewise.
9116 * ipa-inline.c: Likewise.
9117 * ipa-visibility.c: Likewise.
9119 * lto-cgraph.c: Likewise.
9120 * omp-expand.c: Likewise.
9121 * omp-general.c: Likewise.
9122 * omp-low.c: Likewise.
9123 * omp-offload.c: Likewise.
9124 * omp-simd-clone.c: Likewise.
9125 * opts-global.c: Likewise.
9126 * passes.c: Likewise.
9127 * predict.c: Likewise.
9128 * sancov.c: Likewise.
9129 * sanopt.c: Likewise.
9130 * symtab.c: Likewise.
9131 * toplev.c: Likewise.
9132 * trans-mem.c: Likewise.
9133 * tree-chkp.c: Likewise.
9134 * tree-eh.c: Likewise.
9135 * tree-into-ssa.c: Likewise.
9136 * tree-object-size.c: Likewise.
9137 * tree-parloops.c: Likewise.
9138 * tree-profile.c: Likewise.
9139 * tree-ssa-ccp.c: Likewise.
9140 * tree-ssa-live.c: Likewise.
9141 * tree-ssa-loop.c: Likewise.
9142 * tree-ssa-sccvn.c: Likewise.
9143 * tree-ssa-structalias.c: Likewise.
9144 * tree-ssa.c: Likewise.
9145 * tree-streamer-in.c: Likewise.
9146 * tree-vectorizer.c: Likewise.
9147 * tree-vrp.c: Likewise.
9149 * ubsan.c: Likewise.
9150 * varasm.c: Likewise.
9151 * varpool.c: Likewise.
9152 * tree.c: Remove functions moved to attribs.[ch].
9154 * config/aarch64/aarch64.c: Add attrs.h header file.
9155 * config/alpha/alpha.c: Likewise.
9156 * config/arc/arc.c: Likewise.
9157 * config/arm/arm.c: Likewise.
9158 * config/avr/avr.c: Likewise.
9159 * config/bfin/bfin.c: Likewise.
9160 * config/c6x/c6x.c: Likewise.
9161 * config/cr16/cr16.c: Likewise.
9162 * config/cris/cris.c: Likewise.
9163 * config/darwin.c: Likewise.
9164 * config/epiphany/epiphany.c: Likewise.
9165 * config/fr30/fr30.c: Likewise.
9166 * config/frv/frv.c: Likewise.
9167 * config/ft32/ft32.c: Likewise.
9168 * config/h8300/h8300.c: Likewise.
9169 * config/i386/winnt.c: Likewise.
9170 * config/ia64/ia64.c: Likewise.
9171 * config/iq2000/iq2000.c: Likewise.
9172 * config/lm32/lm32.c: Likewise.
9173 * config/m32c/m32c.c: Likewise.
9174 * config/m32r/m32r.c: Likewise.
9175 * config/m68k/m68k.c: Likewise.
9176 * config/mcore/mcore.c: Likewise.
9177 * config/microblaze/microblaze.c: Likewise.
9178 * config/mips/mips.c: Likewise.
9179 * config/mmix/mmix.c: Likewise.
9180 * config/mn10300/mn10300.c: Likewise.
9181 * config/moxie/moxie.c: Likewise.
9182 * config/msp430/msp430.c: Likewise.
9183 * config/nds32/nds32-isr.c: Likewise.
9184 * config/nds32/nds32.c: Likewise.
9185 * config/nios2/nios2.c: Likewise.
9186 * config/nvptx/nvptx.c: Likewise.
9187 * config/pa/pa.c: Likewise.
9188 * config/pdp11/pdp11.c: Likewise.
9189 * config/powerpcspe/powerpcspe.c: Likewise.
9190 * config/riscv/riscv.c: Likewise.
9191 * config/rl78/rl78.c: Likewise.
9192 * config/rx/rx.c: Likewise.
9193 * config/s390/s390.c: Likewise.
9194 * config/sh/sh.c: Likewise.
9195 * config/sol2.c: Likewise.
9196 * config/sparc/sparc.c: Likewise.
9197 * config/spu/spu.c: Likewise.
9198 * config/stormy16/stormy16.c: Likewise.
9199 * config/tilegx/tilegx.c: Likewise.
9200 * config/tilepro/tilepro.c: Likewise.
9201 * config/v850/v850.c: Likewise.
9202 * config/vax/vax.c: Likewise.
9203 * config/visium/visium.c: Likewise.
9204 * config/xtensa/xtensa.c: Likewise.
9206 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9209 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
9210 constraints since the -mupper-regs-* switches have been
9212 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
9213 into a vector from a double word element that was extracted from
9214 another vector, and eliminate extra XXPERMDI instructions.
9215 (vsx_concat_<mode>_2): Likewise.
9216 (vsx_concat_<mode>_3): Likewise.
9217 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
9218 concat to allow optimizing inserts from previous extracts.
9220 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
9222 * config/i386/i386.c (ix86_stack_protect_guard): Generate
9223 memory reference to a SSP offset in TLS address space.
9224 (ix86_print_operand) <case '@'>: Remove.
9225 (ix86_print_operand_punct_valid_p): Remove '@' code.
9226 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
9228 (stack_tls_protect_set_<mode>): Remove.
9229 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
9230 (stack_tls_protect_test_<mode>): Remove.
9231 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
9233 2017-08-07 Olivier Hainque <hainque@adacore.com>
9236 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
9238 2017-08-07 Douglas Rupp <rupp@adacore.com>
9240 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
9241 variable was referenced as multidir in command.
9243 2017-08-07 Jakub Jelinek <jakub@redhat.com>
9246 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
9249 2017-08-07 Martin Liska <mliska@suse.cz>
9251 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
9252 * config/rl78/rl78.c: Add include of attribs.h.
9253 * config/sh/sh.c: Likewise.
9254 * config/v850/v850.c: Likewise.
9256 2017-08-07 Tom de Vries <tom@codesourcery.com>
9259 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
9261 2017-08-07 Martin Liska <mliska@suse.cz>
9263 * config/mips/mips.c: Include attribs.h.
9265 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
9268 * doc/invoke.texi: Document change in behvaior for -Ofast for
9271 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
9273 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
9275 (aarch64_pop_regs): Likewise.
9276 (aarch64_gen_load_pair): Likewise.
9277 (aarch64_save_callee_saves): Likewise.
9278 (aarch64_restore_callee_saves): Likewise.
9280 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9282 * config/i386/i386.c: Revert the last change.
9284 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9287 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
9289 (ix86_finalize_stack_frame_flags): This. Also clear
9290 frame_pointer_needed if -fno-omit-frame-pointer is used without
9292 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
9293 with ix86_finalize_stack_frame_flags.
9294 (ix86_expand_epilogue): Likewise.
9295 (ix86_expand_split_stack_prologue): Likewise.
9297 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9300 * config/i386/i386.c (get_builtin_code_for_version): Set priority
9301 to P_AES for Westmere.
9303 2017-08-07 Jonathan Yong <10walls@gmail.com>
9305 * config/i386/mingw.opt (fset-stack-executable): Removed.
9306 * config/i386/cygming.opt (fset-stack-executable): Moved
9308 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
9310 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
9312 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
9314 2017-08-07 Marek Polacek <polacek@redhat.com>
9317 * fold-const.c (fold_indirect_ref_1): Check type_domain.
9319 2017-08-07 Martin Liska <mliska@suse.cz>
9321 * attribs.h (canonicalize_attr_name): New function.
9322 (cmp_attribs): Move from c-format.c and adjusted.
9323 (is_attribute_p): Moved from tree.h.
9324 * tree-inline.c: Add new includes.
9325 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
9326 (private_is_attribute_p): Remove.
9327 (private_lookup_attribute): Likewise.
9328 (private_lookup_attribute_by_prefix): Simplify.
9329 (remove_attribute): Use is_attribute_p.
9330 * tree.h: Remove removed declarations.
9332 2017-08-07 Jakub Jelinek <jakub@redhat.com>
9335 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
9336 instead of computing it in the function. Formatting fix.
9337 (expand_case): Don't rely on default_edge being the first edge,
9338 clear it if removing it, pass default_edge to
9339 emit_case_dispatch_table.
9340 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
9343 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
9345 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
9346 insn in the function, emit NOP after the insn.
9348 2017-08-06 Tom de Vries <tom@codesourcery.com>
9350 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
9353 2017-08-06 Tom de Vries <tom@codesourcery.com>
9355 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
9358 2017-08-04 Yury Gribov <tetra2005@gmail.com>
9360 PR tree-optimization/57371
9361 * match.pd: New pattern.
9363 2017-08-04 Marek Polacek <polacek@redhat.com>
9366 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
9367 perform the computation in offset_int.
9369 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9371 PR tree-optimization/81136
9372 * tree-vectorizer.h: Include tree-hash-traits.h.
9373 (vec_base_alignments): New typedef.
9374 (vec_info): Add a base_alignments field.
9375 (vect_record_base_alignments): Declare.
9376 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
9378 (DR_IS_CONDITIONAL_IN_STMT): New macro.
9379 (create_data_ref): Add an is_conditional_in_stmt argument.
9380 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
9381 the is_conditional_in_stmt field.
9382 (data_ref_loc): Add an is_conditional_in_stmt field.
9383 (get_references_in_stmt): Set the is_conditional_in_stmt field.
9384 (find_data_references_in_stmt): Update call to create_data_ref.
9385 (graphite_find_data_references_in_stmt): Likewise.
9386 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
9387 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
9388 (vect_record_base_alignment): New function.
9389 (vect_record_base_alignments): Likewise.
9390 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
9391 for nested statements even if we fail to compute a misalignment.
9392 Use pooled base alignments for unconditional references.
9393 (vect_find_same_alignment_drs): Compare base addresses instead
9395 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
9396 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
9398 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9400 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
9401 Add an explicit name for the enum. Use auto_vec for slp_instances
9403 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
9405 (_bb_vec_info): Add a constructor and destructor.
9406 (vinfo_for_stmt): Return NULL for uids of -1 as well.
9407 (destroy_loop_vec_info): Delete.
9408 (vect_destroy_datarefs): Likewise.
9409 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
9410 (vec_info::vec_info): New function.
9411 (vec_info::~vec_info): Likewise.
9412 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
9413 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
9414 destroy_loop_vec_info.
9415 * tree-vect-loop.c (new_loop_vec_info): Replace with...
9416 (_loop_vec_info::_loop_vec_info): ...this.
9417 (destroy_loop_vec_info): Replace with...
9418 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
9419 the stmt_vec_infos. Leave handling of vec_info information to its
9420 destructor. Remove explicit vector releases.
9421 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
9422 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
9423 * tree-vect-slp.c (new_bb_vec_info): Replace with...
9424 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
9425 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
9426 (destroy_bb_vec_info): Replace with...
9427 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
9428 information to its destructor.
9429 (vect_slp_analyze_bb_1): Use new and delete instead of
9430 new_bb_vec_info and destroy_bb_vec_info.
9431 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
9434 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9436 * tree-data-ref.h (subscript): Add access_fn field.
9437 (data_dependence_relation): Add could_be_independent_p.
9438 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
9439 (same_access_functions): Move to tree-data-ref.c.
9440 * tree-data-ref.c (ref_contains_union_access_p): New function.
9441 (access_fn_component_p): Likewise.
9442 (access_fn_components_comparable_p): Likewise.
9443 (dr_analyze_indices): Add a reference to access_fn_component_p.
9444 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
9446 (constant_access_functions): Likewise.
9447 (add_other_self_distances): Likewise.
9448 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
9449 (initialize_data_dependence_relation): Use XCNEW and remove
9450 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
9451 of access functions that have the same type. Allow the
9452 subsequence to end with different bases in some circumstances.
9453 Record the chosen access functions in SUB_ACCESS_FN.
9454 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
9455 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
9456 (subscript_dependence_tester_1): Likewise dra and drb.
9457 (build_classic_dist_vector): Update calls accordingly.
9458 (subscript_dependence_tester): Likewise.
9459 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
9460 DDR_COULD_BE_INDEPENDENT_P.
9461 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
9462 comp_alias_ddrs instead of may_alias_ddrs.
9463 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
9465 (vect_analyze_data_ref_dependence): Use it if
9466 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
9467 distance vectors if that fails.
9468 (dependence_distance_ge_vf): New function.
9469 (vect_prune_runtime_alias_test_list): Use it. Don't clear
9470 LOOP_VINFO_MAY_ALIAS_DDRS.
9472 2017-08-04 Richard Biener <rguenther@suse.de>
9475 * fold-const.c (fold_binary_loc): Properly restrict
9476 minus_var0 && minus_var1 case when associating undefined overflow
9479 2017-08-04 Tom de Vries <tom@codesourcery.com>
9481 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
9483 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9485 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9486 Don't start diagnostic messages with a capital letter.
9487 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9489 (rs6000_invalid_builtin): Likewise.
9490 (rs6000_trampoline_init): Likewise.
9492 2017-08-03 Jakub Jelinek <jakub@redhat.com>
9495 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
9496 after setting changeable df flags.
9498 2017-08-03 Richard Biener <rguenther@suse.de>
9500 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
9501 up if the use is in USE - X.
9503 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
9505 * toplev.c (dumpfile.h): New include.
9506 (internal_error_reentered): New static function. Use it...
9507 (internal_error_function): ...here to handle reentered internal_error.
9509 2017-08-03 Richard Biener <rguenther@suse.de>
9512 * fold-const.c (split_tree): Add minus_var and minus_con
9513 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
9514 here but always use minus_*.
9515 (associate_trees): Assert we never associate with MINUS_EXPR
9516 and NULL first operand. Do not recurse for PLUS_EXPR operands
9517 when associating as MINUS_EXPR either.
9518 (fold_binary_loc): Track minus_var and minus_con.
9520 2017-08-03 Tom de Vries <tom@codesourcery.com>
9523 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
9524 ACCEL_COMPILER, apply finish_options on
9525 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
9527 2017-08-03 Tom de Vries <tom@codesourcery.com>
9530 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
9531 function_entry_patch_area_size > 0.
9533 2017-08-03 Jakub Jelinek <jakub@redhat.com>
9536 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
9537 instead of 10??LU, perform unit multiplication in wide_int,
9538 don't change alloc_object_size_limit if the limit is larger
9541 PR tree-optimization/81655
9542 PR tree-optimization/81588
9543 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
9544 the case when ranges[i].low and high are 1 for unsigned type with
9548 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
9549 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
9551 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
9553 * tree-vrp.h: Add include guard.
9555 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
9558 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
9559 (ud2): New insn pattern.
9560 * config/i386/i386.c (ix86_expand_epilogue):
9561 For naked functions, generate ud2 instead of trap insn.
9563 2017-08-02 Marek Polacek <polacek@redhat.com>
9566 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
9568 2017-08-02 Tom de Vries <tom@codesourcery.com>
9569 Cesar Philippidis <cesar@codesourcery.com>
9571 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
9572 Add missing edge probabilities.
9574 2017-08-02 Tamar Christina <tamar.christina@arm.com>
9576 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
9579 2017-08-02 Jakub Jelinek <jakub@redhat.com>
9582 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
9583 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
9584 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
9585 prologue_seq sequences - if any.
9587 2017-08-02 Richard Biener <rguenther@suse.de>
9589 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
9590 via vectors if supported, integer extracts via punning if supported
9591 or otherwise vector extracts.
9593 2017-08-02 Richard Biener <rguenther@suse.de>
9595 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
9597 (bitmap_insert_into_set): ... this.
9599 2017-08-02 Richard Biener <rguenther@suse.de>
9601 PR tree-optimization/81633
9603 2015-08-17 Alan Hayward <alan.hayward@arm.com>
9605 PR tree-optimization/71752
9606 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
9608 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
9610 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
9611 (machine_function::call_ms2sysv_pad_out): Remove field.
9612 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
9613 (ix86_compute_frame_layout): Likewise.
9615 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
9618 * config/i386/i386.c (ix86_set_func_type): Disallow naked
9619 attribute with interrupt attribute.
9621 2017-08-01 Andrew Pinski <apinski@cavium.com>
9623 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
9624 BIT_INSERT_EXPR's operand 1
9625 to see if the types precision matches.
9627 2017-08-01 Martin Liska <mliska@suse.cz>
9630 * builtins.c (expand_builtin_memcpy_args): Remove.
9631 (expand_builtin_memcpy): Call newly added function
9632 expand_builtin_memory_copy_args.
9633 (expand_builtin_memcpy_with_bounds): Likewise.
9634 (expand_builtin_mempcpy): Remove last argument.
9635 (expand_builtin_mempcpy_with_bounds): Likewise.
9636 (expand_builtin_memory_copy_args): New function created from
9637 expand_builtin_mempcpy_args with small modifications.
9638 (expand_builtin_mempcpy_args): Remove.
9639 (expand_builtin_stpcpy): Remove unused argument.
9640 (expand_builtin): Likewise.
9641 (expand_builtin_with_bounds): Likewise.
9643 2017-08-01 Martin Liska <mliska@suse.cz>
9646 Make mempcpy more optimal (PR middle-end/70140).
9648 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9651 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9652 __builtin_vec_cmpne verify both arguments are compatible vectors
9653 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
9654 verify arg1_type is a pointer or array type. For __builtin_vec_st,
9655 move computation of aligned to after checking the argument types.
9659 * config/rs6000/vsx.md (vextract_fp_from_shorth,
9660 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
9663 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
9664 Doug Rupp <rupp@adacore.com>
9665 Olivier Hainque <hainque@adacore.com>
9667 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
9668 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
9670 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
9671 for TARGET_OS_CPP_BUILTIN.
9672 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
9673 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
9675 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
9676 passing required abi options to the assembler for EABI configurations.
9677 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
9678 of .text.hot and .text.unlikely sections for kernel modules when
9679 using ARM style exceptions.
9680 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
9681 options. Add EXTRA_CC1_SPEC.
9682 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
9684 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
9686 (ARM_TARGET2_DWARF_FORMAT): Define.
9687 * config/arm/t-vxworks: Adjust multilib control to removal of the
9688 Diab command line options.
9690 2017-08-01 Martin Liska <mliska@suse.cz>
9692 PR gcov-profile/81561
9693 * gcov.c (unblock): Make unblocking safe as we need to preserve
9694 index correspondence of blocks and block_lists.
9696 2017-08-01 Richard Biener <rguenther@suse.de>
9698 PR tree-optimization/81181
9699 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
9700 (compute_antic): ... end of iteration here.
9702 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
9704 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
9705 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
9706 (ftree-slp-vectorize): Likewise.
9707 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
9708 can no longer be set independent of flag_tree_loop_vectorize.
9709 * omp-general.c (emp_max_vf): Likewise.
9710 * opts.c (enable_fdo_optimizations): Remove references to
9711 flag_tree_vectorize, these are now implicit.
9712 (common_handle_option): Remove handling for OPT_ftree_vectorize,
9713 and leave it for the options machinery.
9715 2017-08-01 Martin Liska <mliska@suse.cz>
9718 * builtins.c (expand_builtin_memcpy_args): Remove.
9719 (expand_builtin_memcpy): Call newly added function
9720 expand_builtin_memory_copy_args.
9721 (expand_builtin_memcpy_with_bounds): Likewise.
9722 (expand_builtin_mempcpy): Remove last argument.
9723 (expand_builtin_mempcpy_with_bounds): Likewise.
9724 (expand_builtin_memory_copy_args): New function created from
9725 expand_builtin_mempcpy_args with small modifications.
9726 (expand_builtin_mempcpy_args): Remove.
9727 (expand_builtin_stpcpy): Remove unused argument.
9728 (expand_builtin): Likewise.
9729 (expand_builtin_with_bounds): Likewise.
9731 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9734 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
9735 print "ds:" only for immediates in generic address space.
9737 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
9740 * config/i386/i386.c (ix86_funciton_naked): New prototype.
9741 (ix86_function_ok_for_sibcall): Return false for naked functions.
9743 2017-08-01 Richard Biener <rguenther@suse.de>
9745 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
9746 (compute_antic): Seed worklist with exit block predecessors.
9747 * cfganal.c (dfs_find_deadend): For a cycle return the source
9748 of the edge closing it.
9750 2017-08-01 Tamar Christina <tamar.christina@arm.com>
9752 * config/aarch64/aarch64.c
9753 (aarch64_can_const_movi_rtx_p): Move 0 check.
9755 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9757 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
9758 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
9760 * match.pd: Ditto in address comparison pattern.
9762 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9764 PR tree-optimization/81627
9765 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
9766 closed ssa form for store-store chain.
9768 2017-08-01 Bin Cheng <bin.cheng@arm.com>
9770 PR tree-optimization/81620
9771 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
9772 for store-store chain.
9774 2017-08-01 Jakub Jelinek <jakub@redhat.com>
9776 PR tree-optimization/81588
9777 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
9778 ranges[i].in_p, invert comparison code ccode. For >/>=,
9779 swap rhs1 and rhs2 and comparison code unconditionally,
9780 for </<= don't do that. Don't swap rhs1/rhs2 again if
9781 ranges[i].in_p, instead invert comparison code ccode if
9782 opcode or oe->rank is BIT_IOR_EXPR.
9785 * optabs.def (vec_extract_optab, vec_init_optab): Change from
9786 a direct optab to conversion optab.
9787 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
9788 with GET_MODE_INNER as last argument instead of optab_handler.
9789 * expmed.c (extract_bit_field_1): Likewise. Use vector from
9790 vector extraction if possible and optab is available.
9791 * expr.c (store_constructor): Use convert_optab_handler instead
9792 of optab_handler. Use vector initialization from smaller
9793 vectors if possible and optab is available.
9794 * tree-vect-stmts.c (vectorizable_load): Likewise.
9795 * doc/md.texi (vec_extract, vec_init): Document that the optabs
9797 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
9798 of vec_init from half-sized vectors with the same element mode.
9799 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
9800 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
9801 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
9802 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
9803 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
9804 after mode in gen_vec_extract* calls.
9805 (vec_extract<mode>): Renamed to ...
9806 (vec_extract<mode><ssescalarmodelower>): ... this.
9807 (vec_extract<mode><ssehalfvecmodelower>): New expander.
9808 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
9809 element mode after mode in gen_vec_init* calls.
9810 (VEC_INIT_HALF_MODE): New mode iterator.
9811 (vec_init<mode>): Renamed to ...
9812 (vec_init<mode><ssescalarmodelower>): ... this.
9813 (vec_init<mode><ssehalfvecmodelower>): New expander.
9814 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
9815 (vec_extractv2sfsf): ... this.
9816 (vec_initv2sf): Renamed to ...
9817 (vec_initv2sfsf): ... this.
9818 (vec_extractv2si): Renamed to ...
9819 (vec_extractv2sisi): ... this.
9820 (vec_initv2si): Renamed to ...
9821 (vec_initv2sisi): ... this.
9822 (vec_extractv4hi): Renamed to ...
9823 (vec_extractv4hihi): ... this.
9824 (vec_initv4hi): Renamed to ...
9825 (vec_initv4hihi): ... this.
9826 (vec_extractv8qi): Renamed to ...
9827 (vec_extractv8qiqi): ... this.
9828 (vec_initv8qi): Renamed to ...
9829 (vec_initv8qiqi): ... this.
9830 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
9831 (vec_init<mode>): Renamed to ...
9832 (vec_init<mode><VEC_base_l>): ... this.
9833 (vec_extract<mode>): Renamed to ...
9834 (vec_extract<mode><VEC_base_l>): ... this.
9835 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
9836 (vec_initv2sfsf): ... this.
9837 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
9838 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9839 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
9840 element mode after mode in gen_vec_init* calls.
9841 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
9842 (vec_init<mode><Vel>): ... this.
9843 (vec_extract<mode>): Renamed to ...
9844 (vec_extract<mode><Vel>): ... this.
9845 * config/aarch64/iterators.md (Vel): New mode attribute.
9846 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
9847 Add element mode after mode in gen_vec_extract* calls.
9848 * config/s390/vector.md (non_vec_l): New mode attribute.
9849 (vec_extract<mode>): Renamed to ...
9850 (vec_extract<mode><non_vec_l>): ... this.
9851 (vec_init<mode>): Renamed to ...
9852 (vec_init<mode><non_vec_l>): ... this.
9853 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
9854 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
9856 * config/arm/iterators.md (V_elem_l): New mode attribute.
9857 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
9858 (vec_extract<mode><V_elem_l>): ... this.
9859 (vec_extractv2di): Renamed to ...
9860 (vec_extractv2didi): ... this.
9861 (vec_init<mode>): Renamed to ...
9862 (vec_init<mode><V_elem_l>): ... this.
9863 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
9864 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
9865 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
9866 Add element mode after gen_vec_extract* calls.
9867 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
9868 (vec_init<mode><unitmode>): ... this.
9869 (vec_extract<mode>): Renamed to ...
9870 (vec_extract<mode><unitmode>): ... this.
9871 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
9872 (vec_init<mode><unitmode>): ... this.
9873 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
9874 (vec_initv2sfsf): ... this.
9875 (vec_extractv2sf): Renamed to ...
9876 (vec_extractv2sfsf): ... this.
9877 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
9878 Add element mode after gen_vec_extract* calls.
9879 * config/mips/mips.md (unitmode): New mode iterator.
9880 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
9881 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
9882 * config/spu/spu.md (inner_l): New mode attribute.
9883 (vec_init<mode>): Renamed to ...
9884 (vec_init<mode><inner_l>): ... this.
9885 (vec_extract<mode>): Renamed to ...
9886 (vec_extract<mode><inner_l>): ... this.
9887 * config/sparc/sparc.md (veltmode): New mode iterator.
9888 (vec_init<VMALL:mode>): Renamed to ...
9889 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
9890 * config/ia64/vect.md (vec_initv2si): Renamed to ...
9891 (vec_initv2sisi): ... this.
9892 (vec_initv2sf): Renamed to ...
9893 (vec_initv2sfsf): ... this.
9894 (vec_extractv2sf): Renamed to ...
9895 (vec_extractv2sfsf): ... this.
9896 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
9897 (vec_init<mode>): Renamed to ...
9898 (vec_init<mode><VEC_base_l>): ... this.
9899 (vec_extract<mode>): Renamed to ...
9900 (vec_extract<mode><VEC_base_l>): ... this.
9901 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
9902 (vec_initv2sfsf): ... this.
9903 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
9904 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
9905 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
9906 gen_vec_init* calls.
9908 2017-08-01 Richard Biener <rguenther@suse.de>
9910 PR tree-optimization/81297
9911 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
9912 TREE_OVERFLOW from INTEGER_CSTs.
9914 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
9916 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
9918 2017-07-31 Carl Love <cel@us.ibm.com>
9920 * config/rs6000/rs6000-c: Add support for built-in functions
9921 vector signed char vec_xl_be (signed long long, signed char *);
9922 vector unsigned char vec_xl_be (signed long long, unsigned char *);
9923 vector signed int vec_xl_be (signed long long, signed int *);
9924 vector unsigned int vec_xl_be (signed long long, unsigned int *);
9925 vector signed long long vec_xl_be (signed long long, signed long long *);
9926 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
9927 vector signed short vec_xl_be (signed long long, signed short *);
9928 vector unsigned short vec_xl_be (signed long long, unsigned short *);
9929 vector double vec_xl_be (signed long long, double *);
9930 vector float vec_xl_be (signed long long, float *);
9931 * config/rs6000/altivec.h (vec_xl_be): Add #define.
9932 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
9933 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
9935 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
9936 (altivec_expand_builtin): Add switch statement to call
9937 altivec_expand_xl_be for each builtin.
9938 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
9939 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
9940 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
9941 __builtin_vsx_le_be_v16qi.
9942 * doc/extend.texi: Update the built-in documentation file for the
9943 new built-in functions.
9945 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9948 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
9950 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
9952 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9954 * config.gcc: Add z14.
9955 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
9956 CPU model numbers for z13s and z14.
9957 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
9959 * config/s390/s390-opts.h (enum processor_type): Rename
9960 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9961 * config/s390/s390.c (processor_table): Add field for CPU name to
9962 be passed to Binutils.
9963 (s390_asm_output_machine_for_arch): Use the new field in
9964 processor_table for Binutils.
9965 (s390_expand_builtin): Replace arch12 with z14.
9966 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
9967 (s390_get_sched_attrmask): Likewise.
9968 (s390_get_unit_mask): Likewise.
9969 * config/s390/s390.opt: Add z14 to processor_type enum.
9971 2017-07-31 Martin Jambor <mjambor@suse.cz>
9974 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
9975 regardless of optimization level.
9977 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
9978 Martin Liska <mliska@suse.cz>
9980 * predict.def: Remove old comment and adjust probability.
9981 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
9984 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
9987 * config/i386/i386.c (ix86_function_naked): New function.
9988 (ix86_can_use_return_insn_p): Return false for naked functions.
9989 (ix86_expand_prologue): Skip prologue for naked functions.
9990 (ix86_expand_epilogue): Skip epilogue for naked functions
9991 and emit trap instruction.
9992 (ix86_warn_func_return): New function.
9993 (ix86_attribute_table): Add "naked" attribute specification.
9994 (TARGET_WARN_FUNC_RETURN): Define.
9995 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
9997 2017-07-31 Martin Liska <mliska@suse.cz>
9999 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
10000 (dump_gimple_bb_header): Always dump BB info.
10001 (pp_cfg_jump): Do not append info about BB when dumping a jump.
10003 2017-07-31 Martin Liska <mliska@suse.cz>
10006 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
10007 also with current_function_decl non-null equality.
10009 2017-07-31 Jakub Jelinek <jakub@redhat.com>
10012 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
10013 change type to the element type, instead add eltype variable and
10014 use it where we are interested in the element type.
10016 PR tree-optimization/81603
10017 * ipa-polymorphic-call.c
10018 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
10019 offset arithmetic in offset_int, bail out if the resulting bit offset
10020 doesn't fit into shwi.
10022 2017-07-31 Martin Liska <mliska@suse.cz>
10024 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
10025 (gimplify_save_expr): Fix comment.
10027 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
10030 * config/i386/i386.c (ix86_function_arg): Update arguments for
10032 (ix86_compute_frame_layout): Set the initial stack offset to
10033 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
10034 INCOMING_FRAME_SP_OFFSET.
10035 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
10036 stack before exception handler returns.
10037 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
10038 the 'ERROR_CODE' for exception handler.
10040 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
10042 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
10043 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
10044 (ASM_OUTPUT_REG_POP): Ditto.
10045 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
10046 instead of asm_fprintf to output pure string.
10048 2017-07-29 Jakub Jelinek <jakub@redhat.com>
10050 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
10051 to imported_module_or_decl hook.
10052 (debug_nothing_tree_tree_tree_bool): Remove.
10053 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
10054 * debug.c (do_nothing_debug_hooks): Use
10055 debug_nothing_tree_tree_tree_bool_bool instead of
10056 debug_nothing_tree_tree_tree_bool.
10057 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
10058 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
10059 * sdbout.c (sdb_debug_hooks): Likewise.
10060 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
10061 (gen_namespace_die): Add DW_AT_export_symbols attribute if
10063 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
10064 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
10065 attribute, don't add anything.
10067 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10069 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
10070 (fold_build2_stat_loc): Likewise.
10071 (fold_build3_stat_loc): Likewise.
10072 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
10073 (fold_build1_loc): Remove macro.
10074 (fold_build2_loc): Likewise.
10075 (fold_build3_loc): Likewise.
10077 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10079 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
10080 (gimple_build_debug_bind_source_stat): Likewise.
10081 * gimple.h (gimple_build_debug_bind): Remove macro.
10082 (gimple_build_debug_bind_source): Likewise.
10084 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10086 * bitmap.c (bitmap_alloc): Adjust.
10087 (bitmap_gc_alloc): Likewise.
10088 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
10090 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10092 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
10093 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
10094 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
10095 (bitmap_gc_alloc_stat): Likewise.
10096 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
10098 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10100 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
10101 * rtl.h (shallow_copy_rtx): Remove macro.
10103 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10105 * emit-rtl.c (gen_raw_REG): Adjust.
10106 * gengenrtl.c (gendef): Likewise.
10107 * rtl.c (rtx_alloc_stat): Remove _stat from name.
10108 * rtl.h (rtx_alloc): Remove macro.
10110 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10112 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
10113 (build_tree_list_stat): Likewise.
10114 * tree.h (build_tree_list): Remove macro.
10115 (build_tree_list_vec): Likewise.
10117 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10119 * tree.c (make_vector_stat): Remove _stat from name.
10120 (build_vector_stat): Likewise.
10121 * tree.h (make_vector_stat): Remove macro.
10122 (build_vector_stat): Likewise.
10124 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10126 * tree.h (build_var_debug_value): Remove prototype.
10128 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10130 * tree.c (tree_cons_stat): Remove _stat from name.
10131 * tree.h (tree_cons): Remove macro.
10133 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10135 * tree.c (build_vl_exp_stat): Remove _stat from name.
10136 * tree.h (build_vl_exp): Remove macro.
10138 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10140 * tree.c (build_decl_stat): Remove _stat from name.
10141 * tree.h (build_decl): Remove macro.
10143 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10145 * gimple.c (gimple_build_with_ops_stat): Adjust.
10146 (gimple_alloc_stat): Remove _stat from name.
10147 * gimple.h (gimple_alloc): Remove macro.
10149 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10151 * tree.c (make_tree_vec_stat): Remove _stat from name.
10152 (grow_tree_vec_stat): Likewise.
10153 * tree.h (make_tree_vec_stat): Adjust prototype.
10154 (grow_tree_vec_stat): Likewise.
10155 (make_tree_vec): Remove macro.
10156 (grow_tree_vec): Likewise.
10158 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10160 * fold-const.c (fold_build1_stat_loc): Adjust.
10161 (fold_build2_stat_loc): Likewise.
10162 (fold_build3_stat_loc): Likewise.
10163 * tree.c (build0_stat): Remove _stat from name.
10164 (build1_stat): Likewise.
10165 (build2_stat): Likewise.
10166 (build3_stat): Likewise.
10167 (build4_stat): Likewise.
10168 (build5_stat): Likewise.
10169 * tree.h (build1_loc): Remove macro, and rename _stat function
10171 (build2_loc): Likewise.
10172 (build3_loc): Likewise.
10173 (build4_loc): Likewise.
10174 (build5_loc): Likewise.
10176 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10178 * tree.c (make_int_cst_stat): Remove _stat from name.
10179 * tree.h (make_int_cst_stat): Adjust prototype.
10180 (make_int_cst): Remove macro.
10182 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10184 * tree.c (make_tre_binfo_stat): Remove _stat from name.
10185 * tree.h (make_tree_binfo_stat): Adjust prototype.
10186 (make_tree_binfo): Remove.
10188 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10190 * tree.c (copy_node_stat): Rename to copy_node.
10191 (build_distinct_type_copy): Adjust.
10192 * tree.h (copy_node_stat): Adjust prototype.
10193 (copy_node): Remove macro.
10195 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10197 * tree.c (make_node_stat): rename to make_node.
10198 (build_tree_list_stat): Adjust.
10199 (build0_stat): Likewise.
10200 (build2_stat): Likewise.
10201 (build3_stat): Likewise.
10202 (build4_stat): Likewise.
10203 (build5_stat): Likewise.
10204 (build_decl_stat): Likewise.
10205 * tree.h (make_node_stat): Adjust prototype.
10206 (make_node): remove macro.
10208 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
10210 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
10211 (PPC_FEATURE2_SCV): Likewise.
10212 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
10214 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10216 * config/aarch64/aarch64.c
10217 (aarch64_internal_mov_immediate): Add new special pattern.
10218 * config/aarch64/aarch64.md (*movdi_aarch64):
10219 Add reg/32bit const mov case.
10221 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10222 Richard Sandiford <richard.sandiford@linaro.org>
10224 * config/aarch64/aarch64.md (mov<mode>): Generalize.
10225 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10226 Add integer and movi cases.
10227 (movi-split-hf-df-sf split, fp16): New.
10228 (enabled): Added TARGET_FP_F16INST.
10229 * config/aarch64/iterators.md (GPF_HF): New.
10230 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
10232 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10234 * config/aarch64/aarch64.c
10235 (aarch64_simd_container_mode): Add prototype.
10236 (aarch64_expand_mov_immediate): Add HI support.
10237 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
10238 (aarch64_can_const_movi_rtx_p): New.
10239 (aarch64_preferred_reload_class):
10240 Remove restrictions of using FP registers for certain SIMD operations.
10241 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
10242 (aarch64_valid_floating_const): Add integer move validation.
10243 (aarch64_simd_imm_scalar_p): Remove.
10244 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
10245 (aarch64_legitimate_constant_p): Expand list of supported cases.
10246 * config/aarch64/aarch64-protos.h
10247 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
10248 (aarch64_reinterpret_float_as_int): New.
10249 (aarch64_simd_imm_scalar_p): Remove.
10250 * config/aarch64/constraints.md (Uvi): New.
10251 (Dd): Split into Ds and new Dd.
10252 * config/aarch64/aarch64.md (*movsi_aarch64):
10254 (*movdi_aarch64): Add SIMD mov case.
10256 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10258 * tree-predcom.c: (struct chain): Handle store-store chain in which
10259 stores for elimination only store loop invariant values.
10260 (execute_pred_commoning_chain): Ditto.
10261 (prepare_initializers_chain_store_elim): Ditto.
10262 (prepare_finalizers): Ditto.
10263 (is_inv_store_elimination_chain): New function.
10264 (initialize_root_vars_store_elim_1): New function.
10266 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10268 * tree-predcom.c: Revise general description of the pass.
10269 (enum chain_type): New enum type for store elimination.
10270 (struct chain): New field supporting store elimination.
10271 (struct component): Ditto.
10272 (dump_chain): Dump store-stores chain.
10273 (release_chain): Release resources.
10274 (split_data_refs_to_components): Compute and create component
10275 contains only stores for elimination.
10276 (get_chain_last_ref_at): New function.
10277 (make_invariant_chain): Initialization.
10278 (make_rooted_chain): Specify chain type in parameter and record it.
10279 (add_looparound_copies): Skip for store-stores chain.
10280 (determine_roots_comp): Compute type of chain and pass it to
10282 (initialize_root_vars_store_elim_2): New function.
10283 (finalize_eliminated_stores): New function.
10284 (remove_stmt): Handle store for elimination.
10285 (execute_pred_commoning_chain): Execute predictive commoning on
10286 store-store chains.
10287 (determine_unroll_factor): Skip unroll for store-stores chain.
10288 (prepare_initializers_chain_store_elim): New function.
10289 (prepare_initializers_chain): Hanlde store-store chain.
10290 (prepare_finalizers_chain, prepare_finalizers): New function.
10291 (tree_predictive_commoning_loop): Return integer value indicating
10292 if loop is unrolled or lcssa form is corrupted.
10293 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
10295 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10297 * tree-predcom.c (initialize_root): Delete.
10298 (execute_pred_commoning_chain): Initialize root vars and replace
10299 reference of non-combined chain directly, rather than call above
10302 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10304 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
10305 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
10307 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10309 * tree-predcom.c (struct chain): New field init_seq.
10310 (release_chain): Release init_seq.
10311 (prepare_initializers_chain): Record intialization stmts in above
10313 (insert_init_seqs): New function.
10314 (tree_predictive_commoning_loop): Call insert_init_seqs.
10316 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10318 * tree-predcom.c (determine_roots_comp): Skip trivial components.
10320 2017-07-28 Richard Biener <rguenther@suse.de>
10322 * match.pd: Remove superfluous :c.
10323 * genmatch.c (simplify::id): Add member.
10324 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
10326 (current_id): New global.
10327 (dt_node::parent): Move from ...
10328 (dt_operand::parent): ... here. Add for_id member.
10329 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
10330 (decision_tree::find_node): Relax order requirement when
10331 merging DT_TRUE nodes to ones inbetween the current simplify
10332 and the one we try to merge with. Add diagnostic whenever
10333 we need to enforce pattern order by not merging.
10334 (decision_tree::insert): Set current_id.
10335 (decision_tree::print_node): Dump parent node and for_id.
10336 (parser::last_id): Add member.
10337 (parser::push_simplify): Assign unique id.
10338 (parser::parser): Initialize last_id.
10340 2017-07-28 Martin Liska <mliska@suse.cz>
10343 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
10344 gimple_build_debug_bind.
10346 2017-07-28 Richard Biener <rguenther@suse.de>
10348 PR tree-optimization/81502
10349 * match.pd: Add pattern combining BIT_INSERT_EXPR with
10351 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
10353 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
10354 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
10355 for BIT_FIELD_REF args.
10356 * fold-const.c (make_bit_field_ref): Likewise.
10357 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
10359 2017-07-28 Jakub Jelinek <jakub@redhat.com>
10362 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
10363 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
10364 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
10365 Or it into SANITIZER_UNDEFINED.
10366 * ubsan.c: Include gimple-fold.h and varasm.h.
10367 (ubsan_expand_ptr_ifn): New function.
10368 (instrument_pointer_overflow): New function.
10369 (maybe_instrument_pointer_overflow): New function.
10370 (instrument_object_size): Formatting fix.
10371 (pass_ubsan::execute): Call instrument_pointer_overflow
10372 and maybe_instrument_pointer_overflow.
10373 * internal-fn.c (expand_UBSAN_PTR): New function.
10374 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
10375 * sanitizer.def (__ubsan_handle_pointer_overflow,
10376 __ubsan_handle_pointer_overflow_abort): New builtins.
10377 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
10378 * internal-fn.def (UBSAN_PTR): New internal function.
10379 * opts.c (sanitizer_opts): Add pointer-overflow.
10380 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
10381 * fold-const.c (build_range_check): Compute pointer range check in
10382 integral type if pointer arithmetics would be needed. Formatting
10385 2017-07-28 Martin Liska <mliska@suse.cz>
10388 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
10389 parameters that are of a variable-length.
10391 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
10393 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
10395 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
10396 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
10397 (CRT_CALL_STATIC_FUNCTION): Likewise.
10398 (ASM_DEFAULT_SPEC): New define.
10399 (ASM_SPEC32): Likewise.
10400 (ASM_SPEC64): Likewise.
10401 (ASM_SPEC_COMMON): Likewise.
10402 (ASM_SPEC): Likewise.
10403 (INVALID_64BIT): Likewise.
10404 (LINK_OS_DEFAULT_SPEC): Likewise.
10405 (LINK_OS_SPEC32): Likewise.
10406 (LINK_OS_SPEC64): Likewise.
10407 (POWERPC_LINUX): Likewise.
10408 (PTRDIFF_TYPE): Likewise.
10409 (RESTORE_FP_PREFIX): Likewise.
10410 (RESTORE_FP_SUFFIX): Likewise.
10411 (SAVE_FP_PREFIX): Likewise.
10412 (SAVE_FP_SUFFIX): Likewise.
10413 (SIZE_TYPE): Likewise.
10414 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
10415 (TARGET_64BIT): Likewise.
10416 (TARGET_64BIT): Likewise.
10417 (TARGET_AIX): Likewise.
10418 (WCHAR_TYPE_SIZE): Likewise.
10419 (WCHAR_TYPE): Undefine.
10420 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
10421 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
10422 (CPP_OS_RTEMS_SPEC): Delete.
10423 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
10424 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
10426 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
10428 2017-07-28 Jakub Jelinek <jakub@redhat.com>
10430 PR tree-optimization/81578
10431 * tree-parloops.c (build_new_reduction): Bail out if
10432 reduction_code isn't one of the standard OpenMP reductions.
10433 Move the details printing after that decision.
10435 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10437 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
10438 related to reload_in_progress.
10439 (splat_input_operand): Likewise.
10440 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
10442 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
10444 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
10445 (TARGET_INSTANTIATE_DECLS): Likewise.
10446 (legitimate_indexed_address_p): Delete reload_in_progress code.
10447 (rs6000_debug_legitimate_address_p): Likewise.
10448 (rs6000_eliminate_indexed_memrefs): Likewise.
10449 (rs6000_emit_le_vsx_store): Likewise.
10450 (rs6000_emit_move_si_sf_subreg): Likewise.
10451 (rs6000_emit_move): Likewise.
10452 (register_to_reg_type): Likewise.
10453 (rs6000_pre_atomic_barrier): Likewise.
10454 (rs6000_machopic_legitimize_pic_address): Likewise.
10455 (rs6000_allocate_stack_temp): Likewise.
10456 (rs6000_address_for_fpconvert): Likewise.
10457 (rs6000_address_for_altivec): Likewise.
10458 (rs6000_secondary_memory_needed_rtx): Delete function.
10459 (rs6000_check_sdmode): Likewise.
10460 (rs6000_alloc_sdmode_stack_slot): Likewise.
10461 (rs6000_instantiate_decls): Likewise.
10462 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
10463 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
10464 Delete reload_in_progress.
10465 (*vec_reload_and_plus_<mptrsize>): Likewise.
10466 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
10467 (vsx_div_v2di): Likewise.
10468 (vsx_udiv_v2di): Likewise.
10470 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10472 * config/rs6000/rs6000.opt (mlra): Replace with stub.
10473 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
10474 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
10475 (rs6000_debug_reg_global): Delete print of LRA status.
10476 (rs6000_option_override_internal): Delete dead LRA related code.
10477 (rs6000_lra_p): Delete function.
10478 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
10480 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
10482 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
10483 * config/riscv/rtems.h: New file.
10485 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10486 Sudakshina Das <sudi.das@arm.com>
10488 * config/aarch64/aarch64.md
10489 (define_split for and<mode>3nr_compare): Move
10490 non aarch64_logical_operand to a register.
10491 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
10492 register immediate operand to a register.
10493 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
10495 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10497 PR middle-end/81564
10498 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
10500 2017-07-27 Richard Biener <rguenther@suse.de>
10502 PR tree-optimization/81573
10503 PR tree-optimization/81494
10504 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
10505 multi defuse cycle case.
10507 2017-07-27 Richard Biener <rguenther@suse.de>
10509 PR tree-optimization/81571
10510 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
10513 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
10515 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
10516 earlier and only if MASK_FPU is set. Adjust formatting.
10518 2017-07-27 Martin Liska <mliska@suse.cz>
10520 * opt-functions.awk: Add validation of value of Init.
10521 * optc-gen.awk: Pass new argument.
10523 2017-07-27 Martin Liska <mliska@suse.cz>
10525 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
10526 Fix wrong condition.
10528 2017-07-27 Martin Liska <mliska@suse.cz>
10530 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
10531 BBs and edges seen by autoFDO.
10533 2017-07-27 Richard Biener <rguenther@suse.de>
10535 PR tree-optimization/81502
10536 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
10537 with incompatible but same sized type.
10538 (execute_update_addresses_taken): Likewise.
10540 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
10542 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
10543 flag_tree_loop_vectorize rather than flag_tree_vectorize.
10545 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10548 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
10549 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
10550 Change s_operand to memory_operand.
10552 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
10554 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
10555 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
10556 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
10557 Emit instructions rather than returning an expression. Handle TFmode
10558 and KFmode by casting to TImode.
10559 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
10560 (rs6000_emit_le_vsx_store): Likewise.
10561 * config/rs6000/vsx.md (VSX_TI): New iterator.
10562 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
10563 (*vsx_le_undo_permute_<mode>): Likewise.
10564 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
10565 emit the split sequence.
10566 (*vsx_le_perm_store_<mode>): Likewise.
10568 2017-07-27 Jakub Jelinek <jakub@redhat.com>
10570 PR tree-optimization/81555
10571 PR tree-optimization/81556
10572 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
10573 if true, force CHANGED for the recursive invocation.
10574 (reassociate_bb): Remember original length of ops array, pass
10575 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
10577 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
10578 attributes for noipa attribute. For naked attribute use
10579 lookup_attribute first before lookup_attribute_spec.
10580 * final.c (rest_of_handle_final): Disable IPA RA for functions with
10582 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
10583 for functions with noipa attribute.
10584 (cgraph_externally_visible_p): Return true for functions with noipa
10586 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
10587 for functions with noipa attribute.
10588 * doc/extend.texi: Document noipa function attribute.
10589 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
10590 also for functions with noipa attribute.
10591 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
10593 2017-07-26 Andrew Pinski <apinski@cavium.com>
10595 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
10596 vec_unalign_load_cost and vec_unalign_store_cost.
10598 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10600 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
10601 -mvsx-small-integer option.
10602 (ISA_3_0_MASKS_IEEE): Likewise.
10603 (OTHER_VSX_VECTOR_MASKS): Likewise.
10604 (POWERPC_MASKS): Likewise.
10605 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
10606 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
10607 code, only testing for DImode being allowed in non-VSX floating
10609 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
10610 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
10612 (rs6000_option_override_internal): Delete -mvsx-small-integer.
10613 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
10614 TARGET_P8_VECTOR test.
10615 (rs6000_secondary_reload_simple_move): Likewise.
10616 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
10617 since TARGET_P9_VECTOR was already tested.
10618 (rs6000_opt_masks): Remove -mvsx-small-integer.
10619 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
10620 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10622 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
10623 test for TARGET_VEXTRACTUB was used, and that uses
10625 (p9 extract splitter): Likewise.
10626 (vsx_extract_<mode>_di_p9): Likewise.
10627 (vsx_extract_<mode>_store_p9): Likewise.
10628 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
10629 for TARGET_P9_VECTOR was used. Delete code that is now dead with
10630 the elimination of TARGET_VSX_SMALL_INTEGER.
10631 (vsx_extract_<mode>_p8): Likewise.
10632 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
10633 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
10634 (vsx_set_<mode>_p9): Likewise.
10635 (vsx_set_v4sf_p9): Likewise.
10636 (vsx_set_v4sf_p9_zero): Likewise.
10637 (vsx_insert_extract_v4sf_p9): Likewise.
10638 (vsx_insert_extract_v4sf_p9_2): Likewise.
10639 * config/rs6000/rs6000.md (sign extend splitter): Change
10640 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
10641 (floatsi<mode>2_lfiwax_mem): Likewise.
10642 (floatunssi<mode>2_lfiwzx_mem): Likewise.
10643 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
10644 since a test for TARGET_P9_VECTOR was used.
10645 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10646 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
10647 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10648 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10649 TARGET_P8_VECTOR test.
10650 (fix_trunc<mode>si2_stfiwx): Likewise.
10651 (fix_trunc<mode>si2_internal): Likewise.
10652 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
10653 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10655 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10656 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
10657 TARGET_P8_VECTOR test.
10658 (fixuns_trunc<mode>si2_stfiwx): Likewise.
10659 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
10660 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
10662 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
10663 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
10664 since a test for TARGET_P9_VECTOR was used.
10665 (splitter for loading small constants): Likewise.
10667 2017-07-26 Andrew Pinski <apinski@cavium.com>
10669 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
10672 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
10675 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
10676 (fp_valid_at): Likewise.
10678 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10680 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
10681 (qdf24xx_addrcost_table): Likewise.
10682 (cortexa57_tunings): Update to use generic_branch_cost.
10683 (cortexa72_tunings): Likewise.
10684 (cortexa73_tunings): Likewise.
10685 (qdf24xx_tunings): Likewise.
10687 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
10689 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
10690 (thunderx2t99_branch_cost): Likewise.
10691 (cortexa35_tunings): Update to use generic_branch_cost.
10692 (cortexa53_tunings): Likewise.
10693 (cortexa57_tunings): Likewise.
10694 (cortexa72_tunings): Likewise.
10695 (cortexa73_tunings): Likewise.
10696 (thunderx2t99_tunings): Likewise.
10698 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10700 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
10701 (sparc_option_override): Honour MASK_FSMULD.
10702 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
10703 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
10704 * config/sparc/sparc.opt (mfsmuld): New option.
10705 * doc/invoke.texi (mfsmuld): Document option.
10707 2017-07-26 Marek Polacek <polacek@redhat.com>
10709 PR middle-end/70992
10710 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
10712 2017-07-26 Richard Biener <rguenther@suse.de>
10714 * gimple-match-head.c (do_valueize): Return OP if valueize
10716 (get_def): New helper to get at the def stmt of a SSA name
10717 if valueize allows.
10718 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
10719 do_valueize to get at the def stmt.
10720 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
10722 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
10724 PR middle-end/46932
10725 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
10727 2017-07-26 Martin Liska <mliska@suse.cz>
10730 * function.c (expand_function_start): Make expansion of
10731 nonlocal_goto_save_area after parm_birth_insn.
10733 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
10735 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
10736 from all CPU target flags enable members.
10738 2017-07-26 Richard Biener <rguenther@suse.de>
10740 * genmatch.c (dt_simplify::gen): Make iterator vars const.
10741 (decision_tree::gen): Make 'type' const.
10742 (write_predicate): Likewise.
10744 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10746 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
10747 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
10748 (rs6000_option_override_internal): Likewise.
10749 (rs6000_expand_vector_set): Likewise.
10750 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
10751 (TARGET_UPPER_REGS_SF): Likewise.
10752 (TARGET_UPPER_REGS_DI): Likewise.
10753 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
10754 (TARGET_DIRECT_MOVE_64BIT): Likewise.
10755 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
10756 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
10757 (Splitters for DI constants in Altivec registers): Likewise.
10758 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
10759 (vsx_set_v4sf_p9): Likewise.
10760 (vsx_set_v4sf_p9_zero): Likewise.
10761 (vsx_insert_extract_v4sf_p9): Likewise.
10762 (vsx_insert_extract_v4sf_p9_2): Likewise.
10764 2017-07-25 Carl Love <cel@us.ibm.com>
10766 * doc/extend.texi: Update the built-in documentation file for the
10767 existing built-in functions
10768 vector signed char vec_cnttz (vector signed char);
10769 vector unsigned char vec_cnttz (vector unsigned char);
10770 vector signed short vec_cnttz (vector signed short);
10771 vector unsigned short vec_cnttz (vector unsigned short);
10772 vector signed int vec_cnttz (vector signed int);
10773 vector unsigned int vec_cnttz (vector unsigned int);
10774 vector signed long long vec_cnttz (vector signed long long);
10775 vector unsigned long long vec_cnttz (vector unsigned long long);
10777 2017-07-25 Andrew Pinski <apinski@cavium.com>
10779 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
10780 accesses where the use is for the first operand of a BIT_INSERT.
10782 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10785 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
10786 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
10788 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
10790 * config/i386/gstabs.h: Delete.
10791 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
10793 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
10795 * config/i386/i386.c (ix86_decompose_address): Do not check for
10796 register RTX when looking at index_reg or base_reg.
10797 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
10799 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
10801 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
10802 to update EH info here.
10804 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10806 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
10808 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10810 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
10812 2017-07-25 Torsten Duwe <duwe@suse.de>
10814 * common.opt: Introduce -fpatchable-function-entry
10815 command line option, and its variables function_entry_patch_area_size
10816 and function_entry_patch_area_start.
10817 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
10818 including a two-value parser.
10819 * target.def (print_patchable_function_entry): New target hook.
10820 * targhooks.h (default_print_patchable_function_entry): New function.
10821 * targhooks.c (default_print_patchable_function_entry): Likewise.
10822 * toplev.c (process_options): Switch off IPA-RA if
10823 patchable function entries are being generated.
10824 * varasm.c (assemble_start_function): Look at the
10825 patchable-function-entry command line switch and current
10826 function attributes and maybe generate NOP instructions by
10827 calling the print_patchable_function_entry hook.
10828 * doc/extend.texi: Document patchable_function_entry attribute.
10829 * doc/invoke.texi: Document -fpatchable_function_entry
10830 command line option.
10831 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
10833 * doc/tm.texi: Re-generate.
10835 2017-07-25 Jakub Jelinek <jakub@redhat.com>
10838 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
10839 TARGET_AVX512DQ rather than TARGET_AVX512BW.
10841 2017-07-25 Tamar Christina <tamar.christina@arm.com>
10843 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
10845 2017-07-25 Richard Biener <rguenther@suse.de>
10847 PR tree-optimization/81455
10848 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
10849 not walk in cycles when looking for guards.
10851 2017-07-25 Richard Biener <rguenther@suse.de>
10853 PR tree-optimization/81529
10854 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
10855 when optimizing backedge uses.
10857 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
10859 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
10861 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
10862 to dl_section_ref. On AIX, append an expression to subtract
10863 the size of the section length to dl_section_ref.
10865 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
10867 * configure.ac: If any of the config.* scripts fail, exit 1.
10868 * configure: Regenerate.
10870 2017-07-25 Richard Biener <rguenther@suse.de>
10872 PR middle-end/81546
10873 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
10874 of immediate uses, be more verbose on errors.
10876 2017-07-25 Richard Biener <rguenther@suse.de>
10878 PR tree-optimization/81510
10879 * tree-vect-loop.c (vect_is_simple_reduction): When the
10880 reduction stmt is not inside the loop bail out.
10882 2017-07-25 Richard Biener <rguenther@suse.de>
10884 PR tree-optimization/81303
10885 * tree-vect-loop-manip.c (vect_loop_versioning): Build
10886 profitability check against LOOP_VINFO_NITERSM1.
10888 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
10890 * domwalk.c (cmp_bb_postorder): Simplify.
10891 (sort_bbs_postorder): New function. Use it...
10892 (dom_walker::walk): ...here to optimize common cases.
10894 2017-07-25 Martin Liska <mliska@suse.cz>
10897 * ipa-visibility.c (function_and_variable_visibility): Make the
10898 redirection just on target that supports aliasing.
10899 Fix GNU coding style.
10901 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
10904 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
10906 * config/arm/rtems.h: Likewise.
10907 * config/bfin/rtems.h: Likewise.
10908 * config/i386/rtemself.h: Likewise.
10909 * config/lm32/rtems.h: Likewise.
10910 * config/m32c/rtems.h: Likewise.
10911 * config/m68k/rtemself.h: Likewise.
10912 * config/microblaze/rtems.h: Likewise.
10913 * config/mips/rtems.h: Likewise.
10914 * config/moxie/rtems.h: Likewise.
10915 * config/nios2/rtems.h: Likewise.
10916 * config/powerpcspe/rtems.h: Likewise.
10917 * config/rs6000/rtems.h: Likewise.
10918 * config/rtems.h: Likewise.
10919 * config/sh/rtems.h: Likewise.
10920 * config/sh/rtemself.h: Likewise.
10921 * config/sparc/rtemself.h: Likewise.
10923 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10926 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
10927 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
10928 * tree-ssa-structalias.c (alias_get_name): Same.
10930 2017-07-25 Bin Cheng <bin.cheng@arm.com>
10933 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
10934 instructions if no du chain is found.
10936 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
10938 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
10940 2017-07-25 Richard Biener <rguenther@suse.de>
10942 PR middle-end/81505
10943 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
10946 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
10948 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
10949 upper-regs options.
10950 (ISA_2_7_MASKS_SERVER): Likewise.
10951 (ISA_3_0_MASKS_IEEE): Likewise.
10952 (OTHER_P8_VECTOR_MASKS): Likewise.
10953 (OTHER_VSX_VECTOR_MASKS): Likewise.
10954 (POWERPC_MASKS): Likewise.
10955 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
10956 duplicate list of options.
10957 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
10958 explicit -mupper-regs options.
10959 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
10960 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
10961 alias for -mupper-regs-df.
10962 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
10963 (rs6000_init_hard_regno_mode_ok): Likewise.
10964 (rs6000_option_override_internal): Likewise.
10965 (rs6000_opt_masks): Likewise.
10966 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
10967 options in terms of whether -mvsx or -mpower8-vector was used.
10968 (TARGET_UPPER_REGS_DI): Likewise.
10969 (TARGET_UPPER_REGS_SF): Likewise.
10970 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
10971 -mupper-regs-* options.
10973 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10975 * passes.c (emergency_dump_function): Print some empty lines and a
10976 header before the RTL dump.
10978 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
10980 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
10982 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
10985 * config/aarch64/aarch64.c (aarch64_classify_symbol):
10986 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
10988 2017-07-24 Carl Love <cel@us.ibm.com>
10990 * config/rs6000/rs6000-c.c: Add support for built-in functions
10991 vector float vec_extract_fp32_from_shorth (vector unsigned short);
10992 vector float vec_extract_fp32_from_shortl (vector unsigned short);
10993 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
10994 vec_extract_fp_from_shortl): Add defines for the two builtins.
10995 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
10996 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
10998 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
10999 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
11000 * doc/extend.texi: Update the built-in documentation file for the
11001 new built-in function.
11003 2017-07-24 Jakub Jelinek <jakub@redhat.com>
11006 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
11008 * doc/generic.texi: Likewise.
11009 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
11010 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
11012 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
11014 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
11015 (aarch64_mls_elt_merge<mode>): Likewise.
11017 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
11019 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11020 having __cxa_atexit.
11022 2017-07-23 Michael Collison <michael.collison@arm.com>
11024 * config/arm/arm.c (arm_option_override): Deprecate
11025 use of -mstructure-size-boundary.
11026 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
11027 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
11029 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11032 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
11033 Reduce cost estimate for direct moves.
11035 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
11038 * config/i386/i386.c (ix86_option_override_internal): Disable
11039 BMI, BMI2 and TBM instructions for -m16.
11041 2017-07-21 Carl Love <cel@us.ibm.com>
11043 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11044 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11045 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11046 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11047 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11048 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11049 VMULOSW): New enum "unspec" values.
11050 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11051 altivec_vmulosw): New patterns.
11052 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11053 VMULOSW): Add definitions.
11055 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
11057 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
11058 (qdf24xx): Likewise.
11059 * config/aarch64/aarch64-options-extensions.def (rdma); New.
11060 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
11061 (AARCH64_FL_V8_1): Renumber.
11062 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
11063 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
11064 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
11065 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
11066 rdma to feature modifiers list.
11068 2017-07-21 Yury Gribov <tetra2005@gmail.com>
11070 PR middle-end/56727
11071 * ipa-visibility (function_and_variable_visibility): Convert
11072 recursive PLT call to direct call if appropriate.
11074 2017-07-21 Andrew Pinski <apinski@cavium.com>
11076 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
11077 operand 1 to see if the types precision matches.
11078 * fold-const.c (operand_equal_p): Likewise.
11080 2017-07-21 Richard Biener <rguenther@suse.de>
11082 PR tree-optimization/81303
11083 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
11084 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
11085 (vect_peeling_hash_get_lowest_cost): Adjust.
11086 (vect_enhance_data_refs_alignment): Likewise. Use
11087 vect_get_peeling_costs_all_drs to compute the penalty for no
11088 peeling to match up costs.
11090 2017-07-21 Richard Biener <rguenther@suse.de>
11092 PR tree-optimization/81500
11093 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
11094 we didn't identify a reduction path.
11096 2017-07-21 Tom de Vries <tom@codesourcery.com>
11097 Cesar Philippidis <cesar@codesourcery.com>
11099 PR gcov-profile/81442
11100 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
11103 2017-07-21 Tom de Vries <tom@codesourcery.com>
11106 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
11108 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
11109 nvptx_override_options_after_change.
11111 2017-07-21 Ulrich Drepper <drepper@redhat.com>
11113 * dwarf2out.c (output_file_names): Avoid double testing for
11114 dwarf_version >= 5.
11116 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
11118 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
11120 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11122 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
11124 (try_crossjump_to_edge): Do not punt on partitioned functions.
11126 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11128 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
11129 Put all BBs reachable only via paths crossing cold region to cold
11131 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
11133 2016-07-21 Richard Biener <rguenther@suse.de>
11135 PR tree-optimization/81303
11136 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
11137 into account prologue and epilogue iterations when raising
11138 min_profitable_iters to sth at least covering one vector iteration.
11140 2017-07-21 Tamar Christina <tamar.christina@arm.com>
11142 * config/arm/arm.c (arm_test_cpu_arch_dat):
11145 2017-07-20 Nathan Sidwell <nathan@acm.org>
11147 Remove TYPE_METHODS.
11148 * tree.h (TYPE_METHODS): Delete.
11149 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
11150 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
11151 (dbxout_type_methods): Scan TYPE_FIELDS.
11152 (dbxout_type): Don't check TYPE_METHODS here.
11153 * function.c (use_register_for_decl): Always ignore register for
11154 class types when not optimizing.
11155 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
11156 * tree.c (free_lang_data_in_type): Stitch out member functions and
11157 templates from TYPE_FIELDS.
11158 (build_distinct_type_copy, verify_type_variant,
11159 verify_type): Member fns are on TYPE_FIELDS.
11160 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
11161 * tree-pretty-print.c (dump_generic_node): Likewise.
11163 2017-07-20 Jakub Jelinek <jakub@redhat.com>
11166 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
11167 V2TImode and V4TImode.
11168 (ix86_expand_vector_extract): Likewise.
11169 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
11170 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
11171 (ssescalarmode): Handle V4TImode and V2TImode.
11172 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
11173 (*vec_extractv2ti, *vec_extractv4ti): New insns.
11174 (VEXTRACTI128_MODE): New mode iterator.
11175 (splitter for *vec_extractv?ti first element): New.
11176 (VEC_INIT_MODE): New mode iterator.
11177 (vec_init<mode>): Consolidate 3 expanders into one using
11178 VEC_INIT_MODE mode iterator.
11180 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
11182 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
11183 non_spilled_static_chain_regno_p.
11185 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
11187 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
11189 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
11191 * bb-reorder.c (connect_traces): Allow copying of blocks within
11194 2017-07-20 Richard Biener <rguenther@suse.de>
11196 * gimple.h (gimple_phi_result): Add gphi * overload.
11197 (gimple_phi_result_ptr): Likewise.
11198 (gimple_phi_arg): Likewise. Adjust index assert to only
11199 allow actual argument accesses rather than all slots available
11201 (gimple_phi_arg_def): Add gphi * overload.
11202 * tree-phinodes.c (make_phi_node): Initialize only actual
11204 (resize_phi_node): Clear memory not covered by old node,
11205 do not initialize excess argument slots.
11206 (reserve_phi_args_for_new_edge): Initialize new argument slot
11209 2017-07-20 Bin Cheng <bin.cheng@arm.com>
11211 PR tree-optimization/81388
11213 2016-07-21 Bin Cheng <bin.cheng@arm.com>
11215 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
11216 by removing computation of may_be_zero.
11218 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11219 Tom de Vries <tom@codesourcery.com>
11221 PR middle-end/81030
11222 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
11223 when gimple level profile disagrees with what RTL expander did.
11225 2017-07-20 Richard Biener <rguenther@suse.de>
11227 PR tree-optimization/61171
11228 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
11229 (vect_analyze_stmt): Add slp instance parameter.
11230 (vectorizable_reduction): Likewise.
11231 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
11232 (vect_is_simple_reduction): Deal with chains not detected
11233 as SLP reduction chain, specifically not properly associated
11234 chains containing a mix of plus/minus.
11235 (get_reduction_op): Remove.
11236 (get_initial_defs_for_reduction): Simplify, pass in whether
11237 this is a reduction chain, pass in the SLP node for the PHIs.
11238 (vect_create_epilog_for_reduction): Get the SLP instance as
11240 (vectorizable_reduction): Get the SLP instance as arg.
11241 During analysis remember the SLP node with the PHIs in the
11242 instance. Simplify getting at the vectorized reduction PHIs.
11243 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
11244 through SLP instance.
11245 (vect_slp_analyze_operations): Likewise.
11246 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
11247 (vect_transform_stmt): Likewise.
11249 2017-07-20 Tom de Vries <tom@codesourcery.com>
11251 PR tree-optimization/81489
11252 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
11253 read of phi arg location to before loop that modifies phi.
11255 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11257 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
11260 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11262 PR middle-end/81331
11263 * except.c (execute): Fix ordering issue.
11265 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
11267 PR rtl-optimization/81423
11268 * combine.c (make_compound_operation_int): Don't try to optimize
11269 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
11271 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
11273 PR rtl-optimization/81423
11274 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
11275 with a constant that is -1 in the truncated to mode.
11277 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11279 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
11280 (determine_unlikely_bbs): ... here.
11281 * predict.h (propagate_unlikely_bbs_forward): Declare.
11282 * cfgexpand.c (pass_expand::execute): Use it.
11283 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
11285 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
11286 propagate_unlikely_bbs_forward.
11288 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11290 PR middle-end/81331
11291 * except.c (maybe_add_nop_after_section_switch): New function.
11294 2017-07-19 Tom de Vries <tom@codesourcery.com>
11296 * gimple.h (gimple_phi_set_arg): Make assert more strict.
11298 2017-07-19 Tom de Vries <tom@codesourcery.com>
11300 * gimple.h (gimple_phi_arg): Make assert more strict.
11302 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
11304 * config.gcc (powerpc*-*-*): Add mmintrin.h.
11305 * config/rs6000/mmintrin.h: New file.
11306 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
11308 2017-07-19 Jakub Jelinek <jakub@redhat.com>
11310 PR tree-optimization/81346
11311 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
11313 2017-07-19 Tom de Vries <tom@codesourcery.com>
11315 * config/nvptx/nvptx.md (VECIM): Add V2DI.
11317 2017-07-19 Tom de Vries <tom@codesourcery.com>
11319 * config/nvptx/nvptx-modes.def: Add V2DImode.
11320 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
11321 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
11322 (nvptx_output_mov_insn): Handle lack of mov.b128.
11323 (nvptx_print_operand): Handle 'H' and 'L' codes.
11324 (nvptx_vector_mode_supported): Allow V2DImode.
11325 (nvptx_preferred_simd_mode): New function.
11326 (nvptx_data_alignment): New function.
11327 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
11328 nvptx_preferred_simd_mode.
11329 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
11331 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
11333 2017-07-19 Tom de Vries <tom@codesourcery.com>
11335 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
11336 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
11337 (nvptx_vector_mode_supported): New function. Allow V2SImode.
11338 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
11339 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
11340 (mov<VECIM>_insn): New define_insn.
11341 (define_expand "mov<VECIM>): New define_expand.
11343 2017-07-19 Tom de Vries <tom@codesourcery.com>
11345 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
11347 2017-07-19 Jakub Jelinek <jakub@redhat.com>
11349 PR tree-optimization/81346
11350 * fold-const.h (fold_div_compare, range_check_type): Declare.
11351 * fold-const.c (range_check_type): New function.
11352 (build_range_check): Use range_check_type.
11353 (fold_div_compare): No longer static, rewritten into
11354 a match.pd helper function.
11355 (fold_comparison): Don't call fold_div_compare here.
11356 * match.pd (X / C1 op C2): New optimization using fold_div_compare
11357 as helper function.
11359 2017-07-19 Nathan Sidwell <nathan@acm.org>
11361 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
11362 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
11363 * tree.c (find_decls_types_r, verify_type): Use
11364 TYPE_{MIN,MAX}_VALUE_RAW.
11365 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
11366 (hash_tree): Likewise.
11367 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
11369 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
11372 2017-07-18 Tom de Vries <tom@codesourcery.com>
11374 PR middle-end/81464
11375 * omp-expand.c (expand_omp_for_static_chunk): Handle
11376 equal-argument loop exit phi.
11378 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
11381 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
11382 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
11383 operand 2 predicate.
11384 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
11385 operand 2 predicate.
11386 (ror,rol -> rorx splitters): Use const_int_operand as
11387 operand 2 predicate.
11389 2017-06-18 Richard Biener <rguenther@suse.de>
11391 PR tree-optimization/81410
11392 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
11393 the gap in the ! slp_perm SLP case after each group.
11395 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11397 PR middle-end/81463
11398 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
11401 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11403 PR middle-end/81462
11404 * predict.c (set_even_probabilities): Cleanup; do not affect
11405 probabilities that are already known.
11406 (combine_predictions_for_bb): Call even when count is set.
11408 2017-07-18 Nathan Sidwell <nathan@acm.org>
11410 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
11413 2017-07-18 Bin Cheng <bin.cheng@arm.com>
11416 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
11417 optimization for loop niter analysis.
11419 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
11422 * config/avr/avr.c (avr_optimize_casesi): Don't use
11423 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
11425 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11427 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
11428 body_cost_vec from _vect_peel_extended_info.
11429 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
11430 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
11433 2017-07-18 Bin Cheng <bin.cheng@arm.com>
11435 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
11437 2017-07-18 Richard Biener <rguenther@suse.de>
11439 PR tree-optimization/80620
11440 PR tree-optimization/81403
11441 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
11442 info when re-using a VN table entry.
11444 2017-07-18 Richard Biener <rguenther@suse.de>
11446 PR tree-optimization/81418
11447 * tree-vect-loop.c (vectorizable_reduction): Properly compute
11448 vectype_in. Verify that with lane-reducing reduction operations
11449 we have a single def-use cycle.
11451 2017-07-17 Carl Love <cel@us.ibm.com>
11453 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
11455 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11456 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11457 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11458 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11459 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11460 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11461 VMULOSW): New enum "unspec" values.
11462 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11463 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11464 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11465 altivec_vmulosw): New patterns.
11466 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11467 VMULOSW): Add definitions.
11468 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
11470 * config/alpha/alpha.c: Include predict.h.
11472 2017-07-17 Yury Gribov <tetra2005@gmail.com>
11474 * tree-vrp.c (compare_assert_loc): Fix comparison function
11475 to return predictable results.
11477 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11479 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
11481 (subdi3): Likewise.
11482 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
11483 * doc/invoke.texi (mexpand-adddi): Update text.
11485 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11487 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
11488 that also clobbers the CC register. The old expand code is moved
11490 (*arc_clzsi2): ... here.
11491 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
11492 the CC register. The old expand code is moved to ...
11493 (arc_ctzsi2): ... here.
11495 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11497 * config/arc/arc.opt (mindexed-loads): Use initial value
11498 TARGET_INDEXED_LOADS_DEFAULT.
11499 (mauto-modify-reg): Use initial value
11500 TARGET_AUTO_MODIFY_REG_DEFAULT.
11501 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11502 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11503 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11504 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11506 2017-07-17 Martin Liska <mliska@suse.cz>
11509 * opts.c (finish_options): Do not allow -fgnu-tm
11510 w/ -fsanitize={kernel-,}address. Say sorry.
11512 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11515 * tree-loop-distribution.c (classify_partition): Only assert on
11516 numer of iterations.
11517 (merge_dep_scc_partitions): Delete prameter. Update function call.
11518 (distribute_loop): Remove code handling loop with unknown niters.
11519 (pass_loop_distribution::execute): Skip loop with unknown niters.
11521 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11524 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
11525 function sort_partitions_by_post_order.
11527 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11529 PR tree-optimization/81374
11530 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
11531 the max index of basic blocks, rather than number of basic blocks.
11533 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11535 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
11537 (arc_legitimate_pic_operand_p): Likewise.
11538 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
11540 (arc_needs_pcl_p): Likewise.
11541 (arc_legitimate_pc_offset_p): Likewise.
11542 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
11543 function is also used in constrains.md.
11544 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
11545 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
11546 PLUS. Only return true/false in known cases, otherwise assert.
11547 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
11548 is already called in arc_legitimate_constant_p.
11549 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
11551 (LEGITIMATE_PIC_OPERAND_P): Use
11552 arc_raw_symbolic_reference_mentioned_p function.
11553 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
11558 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11559 Andrew Burgess <andrew.burgess@embecosm.com>
11561 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
11562 (arc_return_address_register): New function.
11563 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
11564 (arc_handle_fndecl_attribute): Add naked attribute.
11565 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
11566 (TARGET_WARN_FUNC_RETURN): Likewise.
11567 (arc_allocate_stack_slots_for_args): New function.
11568 (arc_warn_func_return): Likewise.
11569 (machine_function): Change type fn_type.
11570 (arc_compute_function_type): Consider new naked function type,
11571 change function return type.
11572 (arc_must_save_register): Adapt to handle new
11573 arc_compute_function_type's return type.
11574 (arc_expand_prologue): Likewise.
11575 (arc_expand_epilogue): Likewise.
11576 (arc_return_address_regs): Delete.
11577 (arc_return_address_register): New function.
11578 (arc_epilogue_uses): Use above function.
11579 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
11580 (arc_function_type): Change encoding, add naked type.
11581 (ARC_INTERRUPT_P): Change to handle the new encoding.
11582 (ARC_FAST_INTERRUPT_P): Likewise.
11583 (ARC_NORMAL_P): Define.
11584 (ARC_NAKED_P): Likewise.
11585 (arc_compute_function_type): Delete prototype.
11586 * config/arc/arc.md (in_ret_delay_slot): Use
11587 arc_return_address_register function.
11588 (simple_return): Likewise.
11589 (p_return_i): Likewise.
11591 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11593 PR tree-optimization/81428
11594 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
11595 can't be built for those types.
11597 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11599 Remove stuff dead since r239246.
11601 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
11602 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
11603 (avr_inform_devices): Remove dead stuff.
11605 2017-07-17 Tamar Christina <tamar.christina@arm.com>
11607 * config/arm/arm_neon.h: Fix softp typo.
11609 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11611 PR tree-optimization/81365
11612 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
11613 aggregate moves onto bb predecessor edges, make sure there are no
11614 loads that could alias the lhs in between the start of bb and the
11617 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
11620 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
11621 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
11622 [LSHIFTRT, outer_code = TRUNCATE]: Same.
11624 2017-07-17 Jakub Jelinek <jakub@redhat.com>
11626 PR tree-optimization/81396
11627 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
11628 (init_symbolic_number): Initialize it to 1.
11629 (perform_symbolic_merge): Add n_ops from both operands into the new
11631 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
11632 without base_addr as useless if they need more than one operation.
11633 (bswap_replace): Handle !bswap case for NULL base_addr.
11635 2017-07-17 Tom de Vries <tom@codesourcery.com>
11638 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
11641 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
11643 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
11644 conditional builtin define __FIX_LEON3FT_B2BST.
11646 2017-07-17 Daniel Cederman <cederman@gaisler.com>
11648 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
11649 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
11652 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11654 PR rtl-optimization/81424
11655 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
11656 to remove potential trapping from operands if -fnon-call-exceptions.
11658 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11660 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
11661 profile_proability for scalling.
11662 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
11664 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11666 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
11668 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11670 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
11671 fixpoint arithmetics.
11673 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11675 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
11676 fixpoint arithmetics.
11678 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11680 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
11681 fixpoint arithmetics.
11683 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
11685 * profile-count.h (profile_probability::from_reg_br_prob_note,
11686 profile_probability::to_reg_br_prob_note): New functions.
11687 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
11688 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
11689 * predict.c (probability_reliable_p): Update.
11690 (edge_probability_reliable_p): Update.
11691 (br_prob_note_reliable_p): Update.
11692 (invert_br_probabilities): Update.
11693 (add_reg_br_prob_note): New function.
11694 (combine_predictions_for_insn): Update.
11695 * asan.c (asan_clear_shadow): Update.
11696 * cfgbuild.c (compute_outgoing_frequencies): Update.
11697 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
11698 (update_br_prob_note): Update.
11699 (rtl_verify_edges): Update.
11700 (purge_dead_edges): Update.
11701 (fixup_reorder_chain): Update.
11702 * emit-rtl.c (try_split): Update.
11703 * ifcvt.c (cond_exec_process_insns): Update.
11704 (cond_exec_process_if_block): Update.
11705 (dead_or_predicable): Update.
11706 * internal-fn.c (expand_addsub_overflow): Update.
11707 (expand_neg_overflow): Update.
11708 (expand_mul_overflow): Update.
11709 * loop-doloop.c (doloop_modify): Update.
11710 * loop-unroll.c (compare_and_jump_seq): Update.
11711 * optabs.c (emit_cmp_and_jump_insn_1): Update.
11712 * predict.h: Update.
11713 * reorg.c (mostly_true_jump): Update.
11715 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
11716 * config/alpha/alpha.c (emit_unlikely_jump): Update.
11717 * config/arc/arc.c: (emit_unlikely_jump): Update.
11718 * config/arm/arm.c: (emit_unlikely_jump): Update.
11719 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
11720 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
11721 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
11722 (ix86_print_operand): Update.
11723 (ix86_split_fp_branch): Update.
11724 (predict_jump): Update.
11725 * config/ia64/ia64.c (ia64_print_operand): Update.
11726 * config/mmix/mmix.c (mmix_print_operand): Update.
11727 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
11728 (rs6000_expand_split_stack_prologue): Update.
11729 * config/rs6000/rs6000.c: Update.
11730 * config/s390/s390.c (s390_expand_vec_strlen): Update.
11731 (s390_expand_vec_movstr): Update.
11732 (s390_expand_cs_tdsi): Update.
11733 (s390_expand_split_stack_prologue): Update.
11734 * config/sh/sh.c (sh_print_operand): Update.
11735 (expand_cbranchsi4): Update.
11736 (expand_cbranchdi4): Update.
11737 * config/sparc/sparc.c (output_v9branch): Update.
11738 * config/spu/spu.c (get_branch_target): Update.
11739 (ea_load_store_inline): Update.
11740 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
11741 * config/tilepro/tilepro.c: Update.
11743 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
11745 * gimplify.c (mostly_copy_tree_r): Revert latest change.
11746 (gimplify_save_expr): Likewise.
11748 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11750 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
11752 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11754 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
11756 * timevar.def (TV_IPA_FNSUMMARY): Define.
11758 2017-07-16 Daniel Cederman <cederman@gaisler.com>
11760 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
11761 to back store errata sensitive sequence from being generated.
11762 (sqrtdf2_fix): Likewise.
11764 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
11766 * tree-ssa-threadupdate.c (compute_path_counts,
11767 update_joiner_offpath_counts): Use profile_probability.
11769 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
11772 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11774 * config/arm/arm-c.c (arm_cpu_builtins): Define
11775 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11777 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
11779 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11780 array entries to represent __ieee128 versions of the
11781 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
11782 scalar_extract_sig, and scalar_insert_exp built-in functions.
11783 (altivec_resolve_overloaded_builtin): Add special case handling
11784 for the __builtin_scalar_insert_exp function, as represented by
11785 the P9V_BUILTIN_VEC_VSIEDP constant.
11786 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
11787 exponent support for __ieee128 argument.
11788 (VSESQP): Add scalar extract signature support for __ieee128
11790 (VSTDCNQP): Add scalar test negative support for __ieee128
11792 (VSIEQP): Add scalar insert exponent support for __int128 argument
11793 with __ieee128 result.
11794 (VSIEQPF): Add scalar insert exponent support for __ieee128
11795 argument with __ieee128 result.
11796 (VSTDCQP): Add scalar test data class support for __ieee128
11798 (VSTDCNQP): Add overload support for scalar test negative with
11799 __ieee128 argument.
11800 (VSTDCQP): Add overload support for scalar test data class
11801 __ieee128 argument.
11802 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
11803 UNSPEC_VSX_SXSIGDP.
11804 (UNSPEC_VSX_SIEXPQP): New constant.
11805 (xsxexpqp): New insn for VSX scalar extract exponent quad
11807 (xsxsigqp): New insn for VSX scalar extract significand quad
11809 (xsiexpqpf): New insn for VSX scalar insert exponent quad
11810 precision with floating point argument.
11811 (xststdcqp): New expand for VSX scalar test data class quad
11813 (xststdcnegqp): New expand for VSX scalar test negative quad
11815 (xststdcqp): New insn to match expansions for VSX scalar test data
11816 class quad precision and VSX scalar test negative quad precision.
11817 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
11818 special case operand checking to enforce that second operand of
11819 VSX scalar test data class with quad precision argument is a 7-bit
11821 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
11822 prototypes and descriptions of __ieee128 versions of
11823 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
11824 scalar_test_data_class, and scalar_test_neg built-in functions.
11826 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11828 PR tree-optimization/81162
11829 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
11830 replace a negate with an add.
11832 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
11834 * doc/invoke.texi (arm/-mcpu): Document +crypto.
11836 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11838 * config/arm/arm-c.c (arm_cpu_builtins): Define
11839 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
11841 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11843 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
11844 (armv8-r): Set ARM Cortex-R52 as default CPU.
11845 * config/arm/arm-tables.opt: Regenerate.
11846 * config/arm/arm-tune.md: Regenerate.
11847 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
11849 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
11850 extension for -mcpu=cortex-r52.
11852 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
11854 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
11855 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
11856 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
11857 (fp-armv8): Define it as FP_ARMv8 only.
11858 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
11859 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
11861 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
11862 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
11863 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
11864 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
11865 than TARGET_FPU_ARMV8.
11866 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
11867 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
11868 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
11870 * config/arm/neon.md (neon_vrint): Likewise.
11871 (neon_vcvt): Likewise.
11872 (neon_<fmaxmin_op><mode>): Likewise.
11873 (<fmaxmin><mode>3): Likewise.
11874 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
11875 * config/arm/predicates.md (arm_cond_move_operator): Check against
11876 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
11878 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
11880 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
11881 to top of function.
11883 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11885 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
11886 loop in comment with memset.
11888 2017-07-14 Martin Liska <mliska@suse.cz>
11890 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
11891 * dwarf2out.c (is_java): Remove the function.
11892 (output_pubname): Remove usage of the function.
11893 (lower_bound_default): Remove usage of DW_LANG_Java.
11894 (gen_compile_unit_die): Likewise.
11895 * gcc.c: Remove compiler defaults for .java and .zip files.
11896 * gimple-expr.c (remove_suffix): Change as there's no longer
11897 extension than 4-letter one.
11898 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
11899 (gimplify_save_expr): Likewise.
11900 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
11901 as it's possible even for other languages than Java.
11902 * langhooks.h (struct lang_hooks): Remove Java from a comment.
11903 * lto-opts.c (lto_write_options): Remove reference to Java.
11904 * opts.c (strip_off_ending): Update file extension handling.
11905 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
11906 * tree-eh.c (lower_resx): Likewise.
11907 * tree.c (free_lang_data_in_type): Remove dead code.
11908 (find_decls_types_r): Likewise.
11909 (build_common_builtin_nodes): Remove Java from a comment.
11910 (verify_type): Remove dead code.
11911 * varasm.c (assemble_external): Remove Java from a comment.
11913 2017-07-14 Martin Liska <mliska@suse.cz>
11915 * opts.c (finish_options): Add quotes.
11916 (common_handle_option): Likewise.
11918 2017-07-14 Martin Liska <mliska@suse.cz>
11920 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
11921 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
11922 Remove N_SO_PASCAL.
11923 * dwarf2out.c (lower_bound_default): Do not handle
11925 (gen_compile_unit_die): Likewise.
11926 * gcc.c: Remove default extension binding for GNU Pascal.
11927 * stmt.c: Remove Pascal language from a comment.
11928 * xcoffout.c: Likewise.
11930 2017-07-13 David Malcolm <dmalcolm@redhat.com>
11933 * diagnostic-show-locus.c (fixit_cmp): New function.
11934 (layout::layout): Sort m_fixit_hints.
11935 (column_range::column_range): Assert that the values are valid.
11936 (struct char_span): New struct.
11937 (correction::overwrite): New method.
11938 (struct source_line): New struct.
11939 (line_corrections::add_hint): Add assertions. Reimplement memcpy
11940 calls in terms of classes source_line and char_span, and
11941 correction::overwrite.
11942 (selftest::test_overlapped_fixit_printing_2): New function.
11943 (selftest::diagnostic_show_locus_c_tests): Call it.
11945 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
11947 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
11948 early if there is no lhs.
11950 2017-07-13 Martin Liska <mliska@suse.cz>
11952 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
11953 (gen_reference_type_die): Likewise.
11954 * stor-layout.c: Remove Pascal-related comment.
11956 2017-07-13 Martin Liska <mliska@suse.cz>
11958 * opts.c (finish_options): Add quotes to error messages.
11959 (parse_sanitizer_options): Likewise.
11961 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
11963 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
11965 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
11967 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
11969 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
11971 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
11973 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
11975 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
11978 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
11979 provides the hardware capability bits, define the macro
11980 __BUILTIN_CPU_SUPPORTS__.
11981 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
11982 if GLIBC does not provide the hardware capability bits. Add a
11983 gcc_unreachable call if the built-in cpu function is neither
11984 __builtin_cpu_is nor __builtin_cpu_supports.
11985 (rs6000_get_function_versions_dispatcher): Change the warning
11986 that an old GLIBC is used which does not export the capability
11987 bits to be an error.
11988 * doc/extend.texi (target_clones attribute): Document the
11989 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
11990 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
11991 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
11992 the macros defined by GCC if the newer GLIBC is available.
11994 2017-07-12 Jeff Law <law@redhat.com>
11996 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
11997 remaining includes slightly.
11998 * config/riscv/riscv-builtins.c: Include profile-count.h.
12000 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
12003 * config/avr/avr.c (avr_set_current_function): In diagnostic
12004 messages: Quote keywords and (parts of) identifiers.
12005 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
12008 2017-07-12 Carl Love <cel@us.ibm.com>
12010 * config/rs6000/rs6000-c.c: Add support for built-in functions
12011 vector bool char vec_revb (vector bool char);
12012 vector bool short vec_revb (vector short char);
12013 vector bool int vec_revb (vector bool int);
12014 vector bool long long vec_revb (vector bool long long);
12015 * doc/extend.texi: Update the built-in documentation file for the
12016 new built-in functions.
12018 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12020 * config/s390/s390.md: Remove movcc splitter.
12022 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12024 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
12025 load/store on condition.
12027 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
12030 * config/avr/avr.c (avr_encode_section_info)
12031 [progmem && !TREE_READONLY]: Error if progmem object needs
12034 2017-07-11 Michael Collison <michael.collison@arm.com>
12036 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
12039 2017-07-11 Carl Love <cel@us.ibm.com>
12041 * config/rs6000/rs6000-c.c: Add support for builtins
12042 vector unsigned int vec_parity_lsbb (vector signed int);
12043 vector unsigned int vec_parity_lsbb (vector unsigned int);
12044 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
12045 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
12046 vector unsigned long long vec_parity_lsbb (vector signed long long);
12047 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
12048 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
12049 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
12050 * doc/extend.texi: Update the built-in documentation file for the
12051 new built-in functions.
12053 2017-07-11 David Malcolm <dmalcolm@redhat.com>
12055 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
12056 (layout::m_primary_loc): New field.
12057 (layout::layout): Initialize new field. Move location filtering
12058 logic from here to...
12059 (layout::maybe_add_location_range): ...this new method. Add
12060 support for filtering to just the lines already specified by other
12062 (layout::will_show_line_p): New method.
12063 (gcc_rich_location::add_location_if_nearby): New method.
12064 (selftest::test_add_location_if_nearby): New test function.
12065 (selftest::diagnostic_show_locus_c_tests): Call it.
12066 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
12069 2017-07-11 Tom de Vries <tom@codesourcery.com>
12071 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
12072 (bb_first_real_insn): New function.
12073 (nvptx_single): Add extra initialization of broadcasted condition
12076 2017-07-11 Nathan Sidwell <nathan@acm.org>
12078 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
12080 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
12082 * doc/extend.texi (AVR Function Attributes): Remove weblink to
12083 Binutils doc as TEXI will mess them up.
12084 * doc/invoke.texi (AVR Options): Same here.
12086 2017-07-11 Daniel Cederman <cederman@gaisler.com>
12088 * config/sparc/sparc.opt (mfix-ut700): New option.
12089 (mfix-gr712rc): Likewise.
12090 (sparc_fix_b2bst): New variable.
12091 * doc/invoke.texi (SPARC options): Document them.
12092 (ARM options): Fix warnings.
12093 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
12094 instructions to prevent sequences that can trigger the store-store
12095 errata for certain LEON3FT processors.
12096 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
12097 (sparc_option_override): Set sparc_fix_b2bst appropriately.
12098 * config/sparc/sparc.md (fix_b2bst): New attribute.
12099 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
12101 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
12104 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
12106 (*rsqrtsf2_sse): Ditto.
12108 (div<mode>3): Macroize insn from divdf3 and divsf3
12109 using MODEF mode iterator.
12111 2017-07-10 Martin Sebor <msebor@redhat.com>
12113 PR tree-optimization/80397
12114 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
12115 instead of testing for equality to INTEGER_TYPE.
12117 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
12119 * config.gcc: Remove uclibc from arc target spec.
12121 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
12123 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
12125 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12128 * lto-wrapper.c (remove_option): New function.
12129 (merge_and_complain): Merge PIC/PIE options more realistically.
12131 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12133 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
12137 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
12138 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
12139 * config.in: Regenerate.
12140 * configure: Regenerate.
12141 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
12142 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
12143 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
12144 (TARGET_GASISR_PROLOGUES): ...target mask.
12145 * common/config/avr/avr-common.c
12146 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
12147 Set -mgas-isr-prologues.
12148 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
12149 INSERT_PASS_BEFORE for it.
12150 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
12151 * config/avr/avr.c (avr_option_override)
12152 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
12153 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
12154 (avr_attribute_table) <no_gccisr>: Add new function attribute.
12155 (avr_set_current_function) <is_no_gccisr>: Init machine field.
12156 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
12158 (make_avr_pass_pre_proep): New function.
12159 (emit_push_sfr) <treg>: Add argument to function and use it
12160 instead of TMP_REG.
12161 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
12162 and set machine->gasisr.yes.
12163 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
12164 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
12165 __gcc_isr.n_pushed to .L__stack_usage.
12166 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
12167 (avr_asm_final_postscan_insn): ...this new static function.
12168 * config/avr/avr.h (machine_function)
12169 <is_no_gccisr, use_L__stack_usage>: New fields.
12170 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
12171 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
12172 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
12173 (gasisr, *gasisr): New expander and insn.
12174 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12175 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
12176 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
12178 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
12180 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
12183 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12185 Move jump-tables out of .text again.
12188 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
12189 (ASM_OUTPUT_ADDR_VEC): New function.
12190 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
12191 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
12192 INSN_ADDRESSes as asm comment.
12193 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
12194 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
12195 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
12196 * config/avr/avr.md (*tablejump): Adjust comment.
12197 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
12198 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
12200 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
12201 (avr_output_addr_vec): New proto.
12202 (avr_log_t) <insn_addresses>: New field.
12204 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
12207 * config/i386/i386.c (ix86_function_arg_advance): Set
12208 outgoing_args_on_stack to true if there are outgoing arguments
12210 (ix86_function_arg): Likewise.
12211 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
12212 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
12213 * config/i386/i386.h (machine_function): Add
12214 outgoing_args_on_stack.
12216 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
12218 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
12219 supporting pthreds.
12220 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
12222 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
12224 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
12225 (REAL_H): Remove $(MACHMODE_H).
12226 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
12228 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
12229 $(MACHMODE_H) and double-int.h.
12230 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
12232 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
12235 2017-07-07 Andrew Pinski <apinski@cavium.com>
12237 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
12238 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
12240 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12242 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
12243 Add warning if GCC was not configured to link against a GLIBC that
12244 exports the hardware capability bits.
12245 (make_resolver_func): Make resolver function private and not a
12246 COMDAT function. Create the name with clone_function_name instead
12247 of make_unique_name.
12250 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
12251 correct operand in doing the split.
12253 2017-07-07 Carl Love <cel@us.ibm.com>
12255 * config/rs6000/rs6000-c: Add support for built-in function
12256 vector unsigned short vec_pack_to_short_fp32 (vector float,
12258 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
12259 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
12260 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
12261 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
12262 (convert_4f32_8i16): Add define_expand.
12263 * doc/extend.texi: Update the built-in documentation file for the
12264 new built-in function.
12266 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12268 * config/sparc/m8.md: New file.
12269 * config/sparc/sparc.md: Include m8.md.
12271 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12273 * config/sparc/sparc.opt: New option -mvis4b.
12274 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
12275 (sparc_option_override): Handle VIS4B.
12276 (enum sparc_builtins): Define
12277 SPARC_BUILTIN_DICTUNPACK{8,16,32},
12278 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
12279 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
12280 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
12281 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
12282 (check_constant_argument): New function.
12283 (sparc_vis_init_builtins): Define builtins
12284 __builtin_vis_dictunpack{8,16,32},
12285 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
12286 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
12287 __builtin_vis_fpcmpde{8,16,32}shl and
12288 __builtin_vis_fpcmpur{8,16,32}shl.
12289 (sparc_expand_builtin): Check that the constant operands to
12290 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
12291 constant and in range.
12292 * config/sparc/sparc-c.c (sparc_target_macros): Handle
12294 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
12295 (SPARC_IMM5_P): Likewise.
12296 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
12297 (enabled): Handle vis4b.
12298 (UNSPEC_DICTUNPACK): New unspec.
12299 (UNSPEC_FPCMPSHL): Likewise.
12300 (UNSPEC_FPUCMPSHL): Likewise.
12301 (UNSPEC_FPCMPDESHL): Likewise.
12302 (UNSPEC_FPCMPURSHL): Likewise.
12303 (cpu_feature): New CPU feature `vis4b'.
12304 (dictunpack{8,16,32}): New insns.
12305 (FPCSMODE): New mode iterator.
12306 (fpcscond): New code iterator.
12307 (fpcsucond): Likewise.
12308 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
12309 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
12310 (fpcmpde{8,16,32}{si,di}shl): Likewise.
12311 (fpcmpur{8,16,32}{si,di}shl): Likewise.
12312 * config/sparc/constraints.md: Define constraints `q' for unsigned
12313 2-bit integer constants and `t' for unsigned 5-bit integer
12315 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
12317 (imm5_operand_dictunpack16): Likewise.
12318 (imm5_operand_dictunpack32): Likewise.
12319 (imm2_operand): Likewise.
12320 * doc/invoke.texi (SPARC Options): Document -mvis4b.
12321 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
12322 ditunpack* and fpcmp*shl builtins.
12324 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12326 * config.gcc: Handle m8 in --with-{cpu,tune} options.
12327 * config.in: Add HAVE_AS_SPARC6 define.
12328 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
12330 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
12332 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
12333 (CPP_CPU_SPEC): Handle m8.
12334 (ASM_CPU_SPEC): Likewise.
12335 * config/sparc/sparc-opts.h (enum processor_type): Add
12337 * config/sparc/sparc.c (m8_costs): New struct.
12338 (sparc_option_override): Handle TARGET_CPU_m8.
12339 (sparc32_initialize_trampoline): Likewise.
12340 (sparc64_initialize_trampoline): Likewise.
12341 (sparc_issue_rate): Likewise.
12342 (sparc_register_move_cost): Likewise.
12343 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
12344 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
12345 (ASM_CPU64_DEFAULT_SPEC): Likewise.
12346 (CPP_CPU_SPEC): Handle M8.
12347 (ASM_CPU_SPEC): Likewise.
12348 (AS_M8_FLAG): Define.
12349 * config/sparc/sparc.md: Add m8 to the cpu attribute.
12350 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
12351 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
12353 * configure: Regenerate.
12354 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
12357 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12359 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
12361 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
12362 ("*movdi_insn_sp32"): Do not set v3pipe.
12363 ("*movsi_insn"): Likewise.
12364 ("*movdi_insn_sp64"): Likewise.
12365 ("*movsf_insn"): Likewise.
12366 ("*movdf_insn_sp32"): Likewise.
12367 ("*movdf_insn_sp64"): Likewise.
12368 ("*zero_extendsidi2_insn_sp64"): Likewise.
12369 ("*sign_extendsidi2_insn"): Likewise.
12370 ("*mov<VM32:mode>_insn"): Likewise.
12371 ("*mov<VM64:mode>_insn_sp64"): Likewise.
12372 ("*mov<VM64:mode>_insn_sp32"): Likewise.
12373 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
12374 ("<vlop:code><VL:mode>3"): Likewise.
12375 ("*not_<vlop:code><VL:mode>3"): Likewise.
12376 ("*nand<VL:mode>_vis"): Likewise.
12377 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
12378 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
12379 ("one_cmpl<VL:mode>2"): Likewise.
12380 ("faligndata<VM64:mode>_vis"): Likewise.
12381 ("alignaddrsi_vis"): Likewise.
12382 ("alignaddrdi_vis"): Likweise.
12383 ("alignaddrlsi_vis"): Likewise.
12384 ("alignaddrldi_vis"): Likewise.
12385 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
12386 ("bmaskdi_vis"): Likewise.
12387 ("bmasksi_vis"): Likewise.
12388 ("bshuffle<VM64:mode>_vis"): Likewise.
12389 ("cmask8<P:mode>_vis"): Likewise.
12390 ("cmask16<P:mode>_vis"): Likewise.
12391 ("cmask32<P:mode>_vis"): Likewise.
12392 ("pdistn<P:mode>_vis"): Likewise.
12393 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
12395 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12397 * config/sparc/sparc.md ("subtype"): New insn attribute.
12398 ("*wrgsr_sp64"): Set insn subtype.
12399 ("*rdgsr_sp64"): Likewise.
12400 ("alignaddrsi_vis"): Likewise.
12401 ("alignaddrdi_vis"): Likewise.
12402 ("alignaddrlsi_vis"): Likewise.
12403 ("alignaddrldi_vis"): Likewise.
12404 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
12405 ("fexpand_vis"): Likewise.
12406 ("fpmerge_vis"): Likewise.
12407 ("faligndata<VM64:mode>_vis"): Likewise.
12408 ("bshuffle<VM64:mode>_vis"): Likewise.
12409 ("cmask8<P:mode>_vis"): Likewise.
12410 ("cmask16<P:mode>_vis"): Likewise.
12411 ("cmask32<P:mode>_vis"): Likewise.
12412 ("fchksm16_vis"): Likewise.
12413 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
12414 ("fmean16_vis"): Likewise.
12415 ("fp<plusminus_insn>64_vis"): Likewise.
12416 ("<plusminus_insn>v8qi3"): Likewise.
12417 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
12418 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
12419 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
12420 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
12421 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
12422 ("*movqi_insn"): Likewise.
12423 ("*movhi_insn"): Likewise.
12424 ("*movsi_insn"): Likewise.
12425 ("movsi_pic_gotdata_op"): Likewise.
12426 ("*movdi_insn_sp32"): Likewise.
12427 ("*movdi_insn_sp64"): Likewise.
12428 ("movdi_pic_gotdata_op"): Likewise.
12429 ("*movsf_insn"): Likewise.
12430 ("*movdf_insn_sp32"): Likewise.
12431 ("*movdf_insn_sp64"): Likewise.
12432 ("*zero_extendhisi2_insn"): Likewise.
12433 ("*zero_extendqihi2_insn"): Likewise.
12434 ("*zero_extendqisi2_insn"): Likewise.
12435 ("*zero_extendqidi2_insn"): Likewise.
12436 ("*zero_extendhidi2_insn"): Likewise.
12437 ("*zero_extendsidi2_insn_sp64"): Likewise.
12438 ("ldfsr"): Likewise.
12439 ("prefetch_64"): Likewise.
12440 ("prefetch_32"): Likewise.
12441 ("tie_ld32"): Likewise.
12442 ("tie_ld64"): Likewise.
12443 ("*tldo_ldub_sp32"): Likewise.
12444 ("*tldo_ldub1_sp32"): Likewise.
12445 ("*tldo_ldub2_sp32"): Likewise.
12446 ("*tldo_ldub_sp64"): Likewise.
12447 ("*tldo_ldub1_sp64"): Likewise.
12448 ("*tldo_ldub2_sp64"): Likewise.
12449 ("*tldo_ldub3_sp64"): Likewise.
12450 ("*tldo_lduh_sp32"): Likewise.
12451 ("*tldo_lduh1_sp32"): Likewise.
12452 ("*tldo_lduh_sp64"): Likewise.
12453 ("*tldo_lduh1_sp64"): Likewise.
12454 ("*tldo_lduh2_sp64"): Likewise.
12455 ("*tldo_lduw_sp32"): Likewise.
12456 ("*tldo_lduw_sp64"): Likewise.
12457 ("*tldo_lduw1_sp64"): Likewise.
12458 ("*tldo_ldx_sp64"): Likewise.
12459 ("*mov<VM32:mode>_insn"): Likewise.
12460 ("*mov<VM64:mode>_insn_sp64"): Likewise.
12461 ("*mov<VM64:mode>_insn_sp32"): Likewise.
12463 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12465 * config/sparc/sparc.md ("type"): New insn type viscmp.
12466 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
12468 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
12469 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
12470 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
12471 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
12473 ("n7_vis_logical_11cycle"): Likewise.
12474 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
12475 * config/sparc/niagara2.md ("niag3_vis": Likewise.
12476 * config/sparc/niagara.md ("niag_vis"): Likewise.
12477 * config/sparc/ultra3.md ("us3_fga"): Likewise.
12478 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
12480 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12482 * config/sparc/sparc.md: New instruction type `bmask'.
12483 (bmaskdi_vis): Use the `bmask' type.
12484 (bmasksi_vis): Likewise.
12485 * config/sparc/ultra3.md (us3_array): Likewise.
12486 * config/sparc/niagara7.md (n7_array): Likewise.
12487 * config/sparc/niagara4.md (n4_array): Likewise.
12488 * config/sparc/niagara2.md (niag2_vis): Likewise.
12489 (niag3_vis): Likewise.
12490 * config/sparc/niagara.md (niag_vis): Likewise.
12492 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12494 * ipa-comdats.c: Remove optimize check from gate.
12495 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
12496 for functions not optimized.
12497 (ipa_fn_summary_read): Skip optimize check.
12498 (ipa_fn_summary_write): Likewise.
12499 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
12501 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
12503 (can_inline_edge_p): Check flag_pcc_struct_return for match.
12504 (check_callers): Give up on caller which is not optimized.
12505 (inline_small_functions): Likewise.
12506 (ipa_inline): Do not give up when not optimizing.
12507 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
12508 away unoptimizes cdtors.
12509 (whole_program_function_and_variable_visibility): Do
12510 ipa_discover_readonly_nonaddressable_vars in LTO mode.
12511 * ipa.c (process_references): Do not check optimize.
12512 (symbol_table::remove_unreachable_nodes): Update optimize check.
12513 (set_writeonly_bit): Update optimize check.
12514 (pass_ipa_cdtor_merge::gate): Do not check optimize.
12515 (pass_ipa_single_use::gate): Remove.
12517 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12519 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
12520 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
12521 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
12522 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
12523 permute_load, permute_store, adjust_extract, adjust_splat,
12524 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
12525 replace_swap_with_copy, dump_swap_insn_table,
12526 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
12527 recombine_lvx_pattern, recombine_stvx_pattern,
12528 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
12529 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
12530 to file rs6000-p8swap.c.
12531 * config/rs6000/rs6000-p8swap.c: New file.
12532 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
12533 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
12534 and rs6000*-*-* targets.
12536 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12538 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
12540 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12542 * lto-wrapper.c (merge_and_complain): Do not merge
12543 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
12544 fsigned_zeros, ftrapping_math, fwrapv.
12545 (append_compiler_options): Do not track these options.
12546 (append_linker_options): Likewie
12548 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12550 * cgraphunit.c (cgraph_node::finalize_function): When
12551 !flag_toplevel_reorde set no_reorder flag.
12552 (varpool_node::finalize_decl): Likewise.
12553 (symbol_table::compile): Drop no toplevel reorder path.
12555 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12557 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
12558 edges; zero probability is not better than uninitialized.
12560 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
12562 * asan.h (asan_sanitize_allocas_p): Declare.
12563 * asan.c (asan_sanitize_allocas_p): New function.
12564 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
12565 (handle_builtin_alloca): Likewise.
12566 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
12567 if !asan_sanitize_allocas_p.
12568 * params.def (asan-instrument-allocas): Add new option.
12569 * params.h (ASAN_PROTECT_ALLOCAS): Define.
12570 * opts.c (common_handle_option): Disable allocas sanitization for
12573 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
12575 * asan.c: Include gimple-fold.h.
12576 (get_last_alloca_addr): New function.
12577 (handle_builtin_stackrestore): Likewise.
12578 (handle_builtin_alloca): Likewise.
12579 (asan_emit_allocas_unpoison): Likewise.
12580 (get_mem_refs_of_builtin_call): Add new parameter, remove const
12581 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
12582 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
12583 (instrument_builtin_call): Pass gimple iterator to
12584 get_mem_refs_of_builtin_call.
12585 (last_alloca_addr): New global.
12586 * asan.h (asan_emit_allocas_unpoison): Declare.
12587 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
12588 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
12589 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
12590 if function calls alloca.
12591 * gimple-fold.c (replace_call_with_value): Remove static keyword.
12592 * gimple-fold.h (replace_call_with_value): Declare.
12593 * internal-fn.c: Include asan.h.
12594 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
12595 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
12597 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12599 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
12600 (C_SELFTEST_FLAGS): New.
12601 (CPP_SELFTEST_FLAGS): New.
12602 (SELFTEST_DEPS): New, from deps of s-selftest.
12603 (C_SELFTEST_DEPS): New, from deps of s-selftest.
12604 (CPP_SELFTEST_DEPS): New.
12605 (selftest): Add dependency on s-selftest-c++.
12606 (s-selftest): Rename to...
12607 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
12608 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
12609 than SELFTEST_FLAGS.
12610 (selftest-gdb): Rename to...
12611 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
12613 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
12614 (selftest-valgrind): Rename to...
12615 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
12617 (selftest-valgrind): Reintroduce as an alias for
12618 selftest-c-valgrind.
12619 (s-selftest-c++): New.
12620 (selftest-c++-gdb): New.
12621 (selftest-c++-valgrind): New.
12623 2017-07-06 Olivier Hainque <hainque@adacore.com>
12625 * gcc.c (process_command): When deciding if undefined variables
12626 should be ignored when processing specs, accept "gcc -v" as well.
12628 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12630 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
12631 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
12633 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12635 * config/arm/arm-cpus.in (armv8-r): Add new entry.
12636 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
12637 * config/arm/arm-tables.opt: Regenerate.
12638 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
12640 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
12642 2017-07-06 Carl Love <cel@us.ibm.com>
12644 * ChangeLog: Clean up from mid air collision
12646 2017-07-06 Carl Love <cel@us.ibm.com>
12648 * config/rs6000/rs6000-c.c: Add support for built-in functions
12649 vector signed int vec_subc (vector signed int, vector signed int);
12650 vector signed __int128 vec_subc (vector signed __int128,
12651 vector signed __int128);
12652 vector unsigned __int128 vec_subc (vector unsigned __int128,
12653 vector unsigned __int128);
12654 vector signed int vec_sube (vector signed int, vector signed int,
12655 vector signed int);
12656 vector unsigned int vec_sube (vector unsigned int,
12657 vector unsigned int,
12658 vector unsigned int);
12659 vector signed __int128 vec_sube (vector signed __int128,
12660 vector signed __int128,
12661 vector signed__int128);
12662 vector unsigned __int128 vec_sube (vector unsigned __int128,
12663 vector unsigned __int128,
12664 vector unsigned __int128);
12665 vector signed int vec_subec (vector signed int, vector signed int,
12666 vector signed int);
12667 vector unsigned int vec_subec (vector unsigned int,
12668 vector unsigned int,
12669 vector unsigned int);
12670 vector signed __int128 vec_subec (vector signed __int128,
12671 vector signed __int128,
12672 vector signed__int128);
12673 vector unsigned __int128 vec_subec (vector unsigned __int128,
12674 vector unsigned __int128,
12675 vector unsigned __int128);
12676 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
12677 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
12678 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
12679 BU_ALTIVEC_OVERLOAD_X definitions.
12680 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
12681 * doc/extend.texi: Update the built-in documentation file for the new
12682 built-in functions.
12684 2017-07-06 David Malcolm <dmalcolm@redhat.com>
12687 * diagnostic-show-locus.c (layout::layout): Use start and finish
12688 spelling location for the start and finish of each range.
12689 * genmatch.c (linemap_client_expand_location_to_spelling_point):
12690 Add unused aspect param.
12691 * input.c (expand_location_1): Add "aspect" param, and use it
12692 to access the correct part of the location.
12693 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
12695 (expand_location_to_spelling_point): Likewise.
12696 (linemap_client_expand_location_to_spelling_point): Add "aspect"
12697 param, and pass it to expand_location_1.
12699 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
12701 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
12702 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
12703 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
12704 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
12705 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
12706 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
12707 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
12708 _mm_maskz_getmant_ss): New intrinsics.
12709 (__builtin_ia32_getexpss128_mask): Changed to ...
12710 __builtin_ia32_getexpss128_round ... this.
12711 (__builtin_ia32_getexpsd128_mask): Changed to ...
12712 __builtin_ia32_getexpsd128_round ... this.
12713 * config/i386/i386-builtin-types.def
12714 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
12715 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
12716 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
12717 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
12718 __builtin_ia32_getmantss_mask_round): New builtins.
12719 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
12720 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
12721 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
12722 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
12723 * config/i386/sse.md
12724 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
12725 avx512f_sgetexp<mode><mask_scalar_name>
12726 <round_saeonly_scalar_name> ... this.
12727 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
12728 %0, %1, %2<round_saeonly_op3>}): Changed to ...
12729 vgetexp<ssescalarmodesuffix>
12730 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12731 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
12732 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
12733 avx512f_vgetmant<mode><mask_scalar_name>
12734 <round_saeonly_scalar_name> ... this.
12735 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
12736 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
12737 vgetmant<ssescalarmodesuffix>
12738 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
12739 %0<mask_scalar_operand4>, %1, %2
12740 <round_saeonly_scalar_mask_op4>, %3} ... this.
12741 * config/i386/subst.md (mask_scalar_operand4,
12742 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
12743 round_saeonly_scalar_nimm_predicate): New subst attributes.
12745 2017-07-06 Julia Koval <julia.koval@intel.com>
12747 * config/i386/i386.c (ix86_erase_embedded_rounding):
12748 Remove code for old rounding pattern.
12750 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
12752 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
12754 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
12756 * doc/sourcebuild.texi (Test Directives, Variants of
12757 dg-require-support): Add documentation for dg-require-stack-check.
12759 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
12761 * config/i386/subst.md (mask_scalar, round_scalar,
12762 round_saeonly_scalar): New meta-templates.
12763 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
12764 round_scalar_mask_operand3, round_scalar_mask_op3,
12765 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
12766 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
12767 round_saeonly_scalar_constraint,
12768 round_saeonly_scalar_prefix): New subst attribute.
12769 * config/i386/sse.md
12770 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
12771 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
12772 <round_scalar_name> ... this.
12773 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
12774 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
12775 <round_scalar_name> ... this.
12776 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
12777 <sse>_vm<code><mode>3<mask_scalar_name>
12778 <round_saeonly_scalar_name> ... this.
12779 (v<plusminus_mnemonic><ssescalarmodesuffix>
12780 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12781 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12782 v<plusminus_mnemonic><ssescalarmodesuffix>
12783 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12784 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12785 (v<multdiv_mnemonic><ssescalarmodesuffix>
12786 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
12787 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
12788 v<multdiv_mnemonic><ssescalarmodesuffix>
12789 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12790 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
12791 (v<maxmin_float><ssescalarmodesuffix>
12792 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
12793 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
12794 v<maxmin_float><ssescalarmodesuffix>
12795 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
12796 %0<mask_scalar_operand3>, %1, %<iptr>2
12797 <round_saeonly_scalar_mask_op3>} ... this.
12799 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
12801 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
12802 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
12804 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
12805 Alan Hayward <alan.hayward@arm.com>
12806 David Sherwood <david.sherwood@arm.com>
12808 * combine.c (simplify_if_then_else): Remove "enum" before
12810 * compare-elim.c (can_eliminate_compare): Likewise.
12811 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12813 (aarch64_lookup_simd_builtin_type): Likewise.
12814 (aarch64_simd_builtin_type): Likewise.
12815 (aarch64_init_simd_builtin_types): Likewise.
12816 (aarch64_simd_expand_args): Likewise.
12817 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
12819 (aarch64_reverse_mask): Likewise.
12820 (aarch64_simd_emit_reg_reg_move): Likewise.
12821 (aarch64_gen_adjusted_ldpstp): Likewise.
12822 (aarch64_ccmp_mode_to_code): Likewise.
12823 (aarch64_operands_ok_for_ldpstp): Likewise.
12824 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12825 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
12827 (aarch64_min_divisions_for_recip_mul): Likewise.
12828 (aarch64_reassociation_width): Likewise.
12829 (aarch64_get_condition_code_1): Likewise.
12830 (aarch64_simd_emit_reg_reg_move): Likewise.
12831 (aarch64_simd_attr_length_rglist): Likewise.
12832 (aarch64_reverse_mask): Likewise.
12833 (aarch64_operands_ok_for_ldpstp): Likewise.
12834 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
12835 (aarch64_gen_adjusted_ldpstp): Likewise.
12836 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
12838 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
12839 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12840 (arm_lookup_simd_builtin_type): Likewise.
12841 (arm_simd_builtin_type): Likewise.
12842 (arm_init_simd_builtin_types): Likewise.
12843 (arm_expand_builtin_args): Likewise.
12844 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
12845 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
12846 (ft32_setup_incoming_varargs): Likewise.
12847 (ft32_function_arg): Likewise.
12848 (ft32_function_arg_advance): Likewise.
12849 (ft32_pass_by_reference): Likewise.
12850 (ft32_arg_partial_bytes): Likewise.
12851 (ft32_valid_pointer_mode): Likewise.
12852 (ft32_addr_space_pointer_mode): Likewise.
12853 (ft32_addr_space_legitimate_address_p): Likewise.
12854 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
12856 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
12857 (ix86_emit_outlined_ms2sysv_restore): Likewise.
12858 (iamcu_alignment): Likewise.
12859 (canonicalize_vector_int_perm): Likewise.
12860 (ix86_noce_conversion_profitable_p): Likewise.
12861 (ix86_mpx_bound_mode): Likewise.
12862 (ix86_operands_ok_for_move_multiple): Likewise.
12863 * config/microblaze/microblaze-protos.h
12864 (microblaze_expand_conditional_branch_reg): Likewise.
12865 * config/microblaze/microblaze.c
12866 (microblaze_expand_conditional_branch_reg): Likewise.
12867 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12869 (rs6000_reassociation_width): Likewise.
12870 (rs6000_invalid_binary_op): Likewise.
12871 (fusion_p9_p): Likewise.
12872 (emit_fusion_p9_load): Likewise.
12873 (emit_fusion_p9_store): Likewise.
12874 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
12876 (riscv_hard_regno_mode_ok_p): Likewise.
12877 (riscv_address_insns): Likewise.
12878 (riscv_split_symbol): Likewise.
12879 (riscv_legitimize_move): Likewise.
12880 (riscv_function_value): Likewise.
12881 (riscv_hard_regno_nregs): Likewise.
12882 (riscv_expand_builtin): Likewise.
12883 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
12884 (riscv_build_integer): Likewise.
12885 (riscv_split_integer): Likewise.
12886 (riscv_legitimate_constant_p): Likewise.
12887 (riscv_cannot_force_const_mem): Likewise.
12888 (riscv_regno_mode_ok_for_base_p): Likewise.
12889 (riscv_valid_base_register_p): Likewise.
12890 (riscv_valid_offset_p): Likewise.
12891 (riscv_valid_lo_sum_p): Likewise.
12892 (riscv_classify_address): Likewise.
12893 (riscv_legitimate_address_p): Likewise.
12894 (riscv_address_insns): Likewise.
12895 (riscv_load_store_insns): Likewise.
12896 (riscv_force_binary): Likewise.
12897 (riscv_split_symbol): Likewise.
12898 (riscv_force_address): Likewise.
12899 (riscv_legitimize_address): Likewise.
12900 (riscv_move_integer): Likewise.
12901 (riscv_legitimize_const_move): Likewise.
12902 (riscv_legitimize_move): Likewise.
12903 (riscv_address_cost): Likewise.
12904 (riscv_subword): Likewise.
12905 (riscv_output_move): Likewise.
12906 (riscv_canonicalize_int_order_test): Likewise.
12907 (riscv_emit_int_order_test): Likewise.
12908 (riscv_function_arg_boundary): Likewise.
12909 (riscv_pass_mode_in_fpr_p): Likewise.
12910 (riscv_pass_fpr_single): Likewise.
12911 (riscv_pass_fpr_pair): Likewise.
12912 (riscv_get_arg_info): Likewise.
12913 (riscv_function_arg): Likewise.
12914 (riscv_function_arg_advance): Likewise.
12915 (riscv_arg_partial_bytes): Likewise.
12916 (riscv_function_value): Likewise.
12917 (riscv_pass_by_reference): Likewise.
12918 (riscv_setup_incoming_varargs): Likewise.
12919 (riscv_print_operand): Likewise.
12920 (riscv_elf_select_rtx_section): Likewise.
12921 (riscv_save_restore_reg): Likewise.
12922 (riscv_for_each_saved_reg): Likewise.
12923 (riscv_register_move_cost): Likewise.
12924 (riscv_hard_regno_mode_ok_p): Likewise.
12925 (riscv_hard_regno_nregs): Likewise.
12926 (riscv_class_max_nregs): Likewise.
12927 (riscv_memory_move_cost): Likewise.
12928 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
12929 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
12930 (rl78_addr_space_address_mode): Likewise.
12931 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12933 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12934 (rs6000_reassociation_width): Likewise.
12935 (rs6000_invalid_binary_op): Likewise.
12936 (fusion_p9_p): Likewise.
12937 (emit_fusion_p9_load): Likewise.
12938 (emit_fusion_p9_store): Likewise.
12939 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
12940 (ok_for_simple_move_operands): Likewise.
12941 (ok_for_simple_move_strict_operands): Likewise.
12942 (ok_for_simple_arith_logic_operands): Likewise.
12943 (visium_legitimize_reload_address): Likewise.
12944 (visium_select_cc_mode): Likewise.
12945 (output_cbranch): Likewise.
12946 (visium_split_double_move): Likewise.
12947 (visium_expand_copysign): Likewise.
12948 (visium_expand_int_cstore): Likewise.
12949 (visium_expand_fp_cstore): Likewise.
12950 * config/visium/visium.c (visium_pass_by_reference): Likewise.
12951 (visium_function_arg): Likewise.
12952 (visium_function_arg_advance): Likewise.
12953 (visium_libcall_value): Likewise.
12954 (visium_setup_incoming_varargs): Likewise.
12955 (visium_legitimate_constant_p): Likewise.
12956 (visium_legitimate_address_p): Likewise.
12957 (visium_legitimize_address): Likewise.
12958 (visium_secondary_reload): Likewise.
12959 (visium_register_move_cost): Likewise.
12960 (visium_memory_move_cost): Likewise.
12961 (prepare_move_operands): Likewise.
12962 (ok_for_simple_move_operands): Likewise.
12963 (ok_for_simple_move_strict_operands): Likewise.
12964 (ok_for_simple_arith_logic_operands): Likewise.
12965 (visium_function_value_1): Likewise.
12966 (rtx_ok_for_offset_p): Likewise.
12967 (visium_legitimize_reload_address): Likewise.
12968 (visium_split_double_move): Likewise.
12969 (visium_expand_copysign): Likewise.
12970 (visium_expand_int_cstore): Likewise.
12971 (visium_expand_fp_cstore): Likewise.
12972 (visium_split_cstore): Likewise.
12973 (visium_select_cc_mode): Likewise.
12974 (visium_split_cbranch): Likewise.
12975 (output_cbranch): Likewise.
12976 (visium_print_operand_address): Likewise.
12977 * expmed.c (flip_storage_order): Likewise.
12978 * expmed.h (emit_cstore): Likewise.
12979 (flip_storage_order): Likewise.
12980 * genrecog.c (validate_pattern): Likewise.
12981 * hsa-gen.c (gen_hsa_addr): Likewise.
12982 * internal-fn.c (expand_arith_overflow): Likewise.
12983 * ira-color.c (allocno_copy_cost_saving): Likewise.
12984 * lra-assigns.c (find_hard_regno_for_1): Likewise.
12985 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
12986 (process_invariant_for_inheritance): Likewise.
12987 * lra-eliminations.c (move_plus_up): Likewise.
12988 * omp-low.c (lower_oacc_reductions): Likewise.
12989 * simplify-rtx.c (simplify_subreg): Likewise.
12990 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
12991 (TARGET_CHKP_BOUND_MODE): Likewise..
12992 * targhooks.c (default_chkp_bound_mode): Likewise.
12993 (default_setup_incoming_vararg_bounds): Likewise.
12994 * targhooks.h (default_chkp_bound_mode): Likewise.
12995 (default_setup_incoming_vararg_bounds): Likewise.
12996 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
12997 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
12998 (have_whole_vector_shift): Likewise.
12999 * tree-vect-stmts.c (vectorizable_load): Likewise.
13000 * doc/tm.texi: Regenerate.
13002 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
13004 Graceful degrade if Binutils PR21472 is not available.
13007 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
13008 .rodata in flash test fails.
13009 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
13010 * confgure: Regenerate.
13011 * config.in: Regenerate.
13012 * config/avr/avr.c (avr_asm_named_section)
13013 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
13014 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
13015 (avr_asm_init_sections): Same.
13017 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13019 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
13020 (fma<VH:mode>4_intrinsic): Likewise.
13021 (*fmsub<VCVTF:mode>4): Likewise.
13022 (*fmsub<VH:mode>4_intrinsic): Likewise.
13024 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
13027 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
13028 Don't depend on "optimize > 0".
13029 (out_movhi_r_mr, out_movqi_mr_r): Same.
13030 (out_movhi_mr_r, out_movqi_r_mr): Same.
13031 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
13032 io_address_operand on "optimize > 0".
13034 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13036 * tree-loop-distribution.c: Add general explanantion on the pass.
13037 (generate_loops_for_partition): Mark distributed loop.
13038 (pg_add_dependence_edges): New parameter. Handle alias data
13039 dependence specially and record it in the parameter if asked.
13040 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
13041 (init_partition_graph_vertices, add_partition_graph_edge): New.
13042 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
13043 (free_partition_graph_vdata, build_partition_graph): New.
13044 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
13045 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
13046 (data_ref_segment_size, latch_dominated_by_data_ref): New.
13047 (compute_alias_check_pairs, version_loop_by_alias_check): New.
13048 (version_for_distribution_p, finalize_partitions): New.
13049 (distribute_loop): Handle alias data dependence specially. Factor
13050 out loop fusion code as functions and call these functions.
13052 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13054 * tree-loop-distribution.c (classify_partition): New parameter and
13055 better handle reduction statement.
13056 (rdg_build_partitions): Revise comment.
13057 (distribute_loop): Compute statements in all partitions and pass it
13058 to classify_partition.
13060 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13062 * tree-loop-distribution.c (enum partition_type): New.
13063 (struct partition): New field type.
13064 (partition_merge_into): Add parameter. Update partition type.
13065 (data_dep_in_cycle_p, update_type_for_merge): New functions.
13066 (build_rdg_partition_for_vertex): Compute partition type.
13067 (rdg_build_partitions): Dump partition type.
13068 (distribute_loop): Update calls to partition_merge_into.
13070 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13072 * tree-loop-distribution.c (struct ddr_hasher): New.
13073 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
13075 (classify_partition): Call get_data_dependence.
13076 (pg_add_dependence_edges): Ditto.
13077 (distribute_loop): Release data dependence hash table.
13079 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13081 * tree-loop-distribution.c (ref_base_address): Delete.
13082 (similar_memory_accesses): Rename ...
13083 (share_memory_accesses): ... to this. Check if partitions access
13084 the same memory reference.
13085 (distribute_loop): Call share_memory_accesses.
13087 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13089 * tree-loop-distribution.c (struct partition): New field recording
13090 its data reference.
13091 (partition_alloc, partition_free): Init and release data refs.
13092 (partition_merge_into): Merge data refs.
13093 (build_rdg_partition_for_vertex): Collect data refs for partition.
13094 (pg_add_dependence_edges): Change parameters from vector to bitmap.
13096 (distribute_loop): Remve data refs from vertice data of partition
13099 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13101 * tree-loop-distribution.c (params.h): Include header file.
13102 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
13103 (datarefs_vec): New global var.
13104 (create_rdg_vertices): Use datarefs_vec directly.
13105 (free_rdg): Don't free data references.
13106 (build_rdg): Update use. Don't free data references.
13107 (distribute_loop): Compute global variable for data references.
13108 Bail out if there are too many data references.
13110 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13112 * tree-loop-distribution.c (loop_nest): New global var.
13113 (build_rdg): Use loop directly, rather than loop nest.
13114 (pg_add_dependence_edges): Remove loop nest parameter. Use global
13116 (distribute_loop): Compute global variable loop nest. Update use.
13118 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13120 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
13121 (partition_merge_into): New parameter. Dump reason for fusion.
13122 (distribute_loop): Update use of partition_merge_into.
13124 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13126 * tree-loop-distribution.c (bb_top_order_index): New.
13127 (bb_top_order_index_size, bb_top_order_cmp): New.
13128 (stmts_from_loop): Use topological order.
13129 (pass_loop_distribution::execute): Compute and release topological
13130 order for basic blocks.
13132 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13134 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
13137 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13139 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
13140 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
13141 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
13142 * internal-fn.def (LOOP_DIST_ALIAS): New.
13143 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
13144 (fold_loop_internal_call): ... this.
13145 (vect_loop_dist_alias_call): New function.
13146 (set_uid_loop_bbs): Call fold_loop_internal_call.
13147 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
13150 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13153 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
13154 Require dead FLAGS_REG at the beginning of a peephole.
13156 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13159 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
13160 arguments in the call to __builtin_ia32_sbb_u32.
13161 (_subborrow_u64): Swap _X and _Y arguments in the call to
13162 __builtin_ia32_sbb_u64.
13164 2017-07-04 Jakub Jelinek <jakub@redhat.com>
13167 * tree-vrp.c (compare_assert_loc): Turn into a function template
13168 with stable template parameter. Only test if a->e is NULL,
13169 !a->e == !b->e has been verified already. Use e == NULL or
13170 e != NULL instead of e or ! e tests. If stable is true, don't use
13171 iterative_hash_expr, on the other side allow a or b or both NULL
13172 and sort the NULLs last.
13173 (process_assert_insertions): Sort using compare_assert_loc<false>
13174 instead of compare_assert_loc, later sort using
13175 compare_assert_loc<true> before calling process_assert_insertions_for
13176 in a loop. Use break instead of continue once seen NULL pointer.
13178 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
13180 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13181 Cortex-R7 and Cortex-R8 processors.
13183 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13185 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
13186 uninitialized while src is not.
13188 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
13190 * common/config/arm/arm-common.c: Adjust include path for
13192 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
13193 (arm-cpu.h): Create in build directory. Adjust dependency rules.
13194 (arm-cpu-data.h): Likewise.
13195 (arm-cpu-cdata.h): Likewise.
13196 * config/arm/arm-cpu.h: Delete.
13197 * config/arm/arm-cpu-cdata.h: Delete.
13198 * config/arm/arm-cpu-data.h: Delete.
13200 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
13202 * config/arm/arm-cpus.in (cortex-a55): New.
13203 (cortex-a75): Likewise.
13204 (cortex-a75.cortex-a55): Likewise.
13205 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
13207 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
13208 * config/arm/arm-cpu-cdata.h: Regenerate.
13209 * config/arm/arm-cpu-data.h: Regenerate.
13210 * config/arm/arm-cpu.h: Regenerate.
13211 * config/arm/arm-tables.opt: Regenerate.
13212 * config/arm/arm-tune.md: Regenerate.
13214 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13216 * haifa-sched.c (sched_create_recovery_edges): Update profile.
13218 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13220 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
13223 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
13225 PR tree-optimization/81292
13226 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
13227 full_string_p, also call adjust_related_strinfos if the adjustment
13228 is simple, otherwise invalidate related strinfos.
13230 2017-07-04 Martin Liska <mliska@suse.cz>
13233 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
13234 newly created variable as DECL_IGNORED_P.
13236 2017-07-04 Martin Liska <mliska@suse.cz>
13239 * ipa-inline.c (inline_small_functions):
13240 Use xstrdup_for_dump.
13242 2017-07-04 Tom de Vries <tom@codesourcery.com>
13244 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
13246 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
13249 * config/darwin.c (darwin_function_switched_text_sections):
13252 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
13254 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
13256 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
13258 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
13260 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13262 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
13263 min_profitable_iters, and th as inclusive lower bounds.
13264 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
13265 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
13266 for min_profitable_iters and min_profitable_estimate.
13267 (vect_transform_loop): Treat th as an inclusive lower bound.
13268 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
13270 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
13273 * config/darwin.c (darwin_function_switched_text_sections):
13274 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
13275 in two pieces, and suppress the use of buf.
13277 2017-07-03 Nathan Sidwell <nathan@acm.org>
13279 * hash-table.h (hash_table_mod1): Fix indentation.
13281 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13283 PR middle-end/81290
13284 * predict.c (force_edge_cold): Be more careful about propagation
13286 * profile-count.h (profile_probability::guessed,
13287 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
13289 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
13291 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
13293 * doc/invoke.texi (rcpc architecture extension): Document it.
13295 2017-07-03 Richard Biener <rguenther@suse.de>
13297 PR tree-optimization/60510
13298 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
13299 the scalar reduction PHI and use it.
13300 (vectorizable_reduction): Properly guard the single_defuse_cycle
13301 path for non-SLP reduction chains where we cannot use it.
13302 Rework reduc_def/index and vector type deduction. Rework
13303 vector operand gathering during reduction op code-gen.
13304 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
13305 chains dissolve the chain and leave it to non-SLP reduction
13308 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13310 * tree-data-ref.h (dr_alignment): Declare.
13311 * tree-data-ref.c (dr_alignment): New function.
13312 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
13313 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
13315 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
13317 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13319 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
13320 and base_misalignment fields.
13321 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
13322 * tree-data-ref.c: Include builtins.h.
13323 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
13324 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
13325 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
13326 * tree-vect-data-refs.c: Include tree-cfg.h.
13327 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
13328 fields instead of calculating an alignment here.
13329 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
13330 innermost_loop_behavior fields.
13332 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13334 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
13336 (DR_STEP_ALIGNMENT): New macro.
13337 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
13338 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
13339 (create_data_ref): Print it.
13340 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
13341 to tell whether the step preserves vector (mis)alignment.
13342 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
13343 Move the check for an integer step and generalise to all INTEGER_CST.
13344 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
13345 Print the outer step alignment.
13347 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13349 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
13350 with offset_alignment.
13351 (DR_ALIGNED_TO): Delete.
13352 (DR_OFFSET_ALIGNMENT): New macro.
13353 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
13354 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
13355 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
13356 (create_data_ref): Likewise.
13357 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
13358 (vect_analyze_data_refs): Likewise.
13359 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
13360 creating dummy innermost behavior.
13362 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13364 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
13365 with a "innermost_loop_behavior *" and refeence tree.
13366 * tree-data-ref.c (dr_analyze_innermost): Likewise.
13367 (create_data_ref): Update call accordingly.
13368 * tree-predcom.c (find_looparound_phi): Likewise.
13370 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13372 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
13373 fields with dr_wrt_vec_loop.
13374 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
13375 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
13376 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
13377 (vect_dr_behavior): New function.
13378 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
13379 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
13380 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
13381 track whether the step preserves the misalignment.
13382 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
13383 Use vect_dr_behavior.
13384 (vect_setup_realignment): Update call accordingly.
13385 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
13386 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
13387 call to vect_create_addr_base_for_vector_ref.
13388 (vect_create_cond_for_align_checks): Likewise.
13389 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
13390 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
13391 (vect_recog_mask_conversion_pattern): Likewise.
13392 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
13393 (new_stmt_vec_info): Remove redundant zeroing.
13395 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
13397 * common/config/arm/arm-common.c (arm_be8_option): New function.
13398 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
13399 (ISA_ARMv6): Add isa_bit_be8.
13400 * config/arm/arm.h (arm_be8_option): Add prototype.
13401 (BE8_SPEC_FUNCTION): New define.
13402 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
13403 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
13404 (mlittle-endian): Similarly.
13405 (mbe8, mbe32): New options.
13406 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
13407 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
13409 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13411 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
13413 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13415 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
13416 (cleanup_tree_cfg_bb): Use it.
13417 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
13419 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
13421 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13424 * loop-doloop.c (add_test): Update profile.
13426 2017-07-03 Martin Liska <mliska@suse.cz>
13429 * sanopt.c (rewrite_usage_of_param): New function.
13430 (sanitize_rewrite_addressable_params): Likewise.
13431 (pass_sanopt::execute): Call rewrite_usage_of_param.
13433 2017-07-03 Richard Biener <rguenther@suse.de>
13435 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
13436 back to using VIEW_CONVERT_EXPR.
13438 2017-07-03 Martin Liska <mliska@suse.cz>
13441 * doc/extend.texi: Document when a resolver function is
13442 generated for target_clones.
13444 2017-07-03 Martin Liska <mliska@suse.cz>
13446 * asan.c (asan_emit_stack_protection): Unpoison just red zones
13447 and shadow memory of auto variables which are subject of
13448 use-after-scope sanitization.
13449 (asan_expand_mark_ifn): Add do set only when is_poison.
13451 2016-07-03 Richard Biener <rguenther@suse.de>
13453 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
13455 (vect_force_simple_reduction): Record reduction def -> phi mapping.
13456 (vectorizable_reduction): Perform reduction PHI creation when
13457 visiting a reduction PHI and adjust and simplify code generation
13458 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
13459 (vect_transform_loop): Visit reduction PHIs.
13460 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
13461 defs into the SLP tree.
13462 (vect_build_slp_tree): Reduction defs terminate the recursion.
13463 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
13465 (vect_get_vec_defs_for_stmt_copy): Export.
13466 (vect_get_vec_defs): Likewise.
13467 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
13469 (vect_get_vec_defs_for_stmt_copy): Declare.
13470 (vect_get_vec_defs): Likewise.
13472 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13474 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
13475 parameter with a "loop" parameter and use it instead of the
13476 loop containing DR_STMT. Don't check simple_iv when doing
13477 BB analysis. Describe the two analysis modes in the comment.
13479 2017-07-03 Tom de Vries <tom@codesourcery.com>
13481 PR tree-optimization/69468
13482 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
13483 (find_same_succ_bb): Handle ignore_edge_flags.
13485 2017-07-03 Tom de Vries <tom@codesourcery.com>
13487 PR tree-optimization/81192
13488 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
13490 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
13492 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
13494 2017-07-03 Tom de Vries <tom@codesourcery.com>
13496 PR tree-optimization/81192
13497 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
13498 BB_SAME_SUCC (bb) == NULL.
13500 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13502 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
13505 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13507 * dumpfile.c: Include profile-count.h
13508 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
13510 (insert_cond_bb): Update profile.
13511 * tree-cfg.h (insert_cond_bb): Update prototype.
13512 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
13513 * tree-dump.c: Do not include tree-cfg.
13515 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13517 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
13519 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13521 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
13524 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13526 * tree-complex.c (expand_complex_div_wide): update profile.
13528 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13529 Alan Hayward <alan.hayward@arm.com>
13530 David Sherwood <david.sherwood@arm.com>
13532 * Makefile.in (MACHMODE_H): Remove insn-modes.h
13533 (CORETYPES_H): New define.
13534 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
13535 (insn-modes-inline.h, s-modes-inline-h): New rules.
13536 (generated_files): Add insn-modes-inline.h.
13537 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
13538 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
13539 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
13540 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
13541 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
13542 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
13543 (build/gencodes.o, build/genconditions.o): Likewise.
13544 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
13545 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
13546 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
13547 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
13548 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
13549 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
13550 * coretypes.h: Include everything up to real.h for generators.
13551 Include insn-modes.h first. Include wide-int-print.h after
13552 wide-int.h. Include insn-modes-inline.h and then machmode.h.
13553 * machmode.h: Don't include insn-modes.h here.
13554 * function-tests.c: Remove includes of signop.h, machmode.h,
13555 double-int.h and wide-int.h.
13557 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
13559 * optc-save-gen.awk: Likewise.
13560 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
13561 * godump.c: Remove include of wide-int-print.h.
13562 * pretty-print.h: Likewise.
13563 * wide-int-print.cc: Likewise.
13564 * wide-int.cc: Likewise.
13565 * hash-map-tests.c: Remove include of signop.h.
13566 * hash-set-tests.c: Likewise.
13567 * rtl-tests.c: Likewise.
13568 * mkconfig.sh: Remove include of machmode.h.
13569 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
13571 (emit_insn_modes_inline_h): ...this new function. Emit the code
13572 into an insn-modes-inline.h header file, adding appropriate
13573 include guards and end comments.
13574 (emit_insn_modes_c_header): Remove include of machmode.h.
13575 (emit_min_insn_modes_c_header): Include coretypes.h rather than
13577 (main): Handle -i flag and call emit_insn_modes_inline_h when
13580 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13582 * tree-ssa-strlen.c (strinfo): Rename the length field to
13583 nonzero_chars. Add a full_string_p field.
13584 (compare_nonzero_chars, zero_length_string_p): New functions.
13585 (get_addr_stridx): Add an offset_out parameter.
13586 Use compare_nonzero_chars.
13587 (get_stridx): Update accordingly. Use compare_nonzero_chars.
13588 (new_strinfo): Update after above changes to strinfo.
13589 (set_endptr_and_length): Set full_string_p.
13590 (get_string_length): Update after above changes to strinfo.
13591 (unshare_strinfo): Update call to new_strinfo.
13592 (maybe_invalidate): Likewise.
13593 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
13594 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
13595 as a uhwi instead of an shwi. Update after above changes to
13596 strinfo and new_strinfo.
13597 (zero_length_string): Assert that chainsi contains full strings.
13598 Use zero_length_string_p. Update call to new_strinfo.
13599 (adjust_related_strinfos): Update after above changes to strinfo.
13600 Copy full_string_p from origsi.
13601 (adjust_last_stmt): Use zero_length_string_p.
13602 (handle_builtin_strlen): Update after above changes to strinfo and
13603 new_strinfo. Install the lhs as the string length if the previous
13604 entry didn't describe a full string.
13605 (handle_builtin_strchr): Update after above changes to strinfo
13607 (handle_builtin_strcpy): Likewise.
13608 (handle_builtin_strcat): Likewise.
13609 (handle_builtin_malloc): Likewise.
13610 (handle_pointer_plus): Likewise.
13611 (handle_builtin_memcpy): Likewise. Track nonzero characters
13612 that aren't necessarily followed by a nul terminator.
13613 (handle_char_store): Likewise.
13615 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13617 PR tree-optimization/80769
13618 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
13619 for malloc and calloc. Document the new invariant that all related
13620 strinfos have delayed lengths or none do.
13621 (verify_related_strinfos): Move earlier in file.
13622 (set_endptr_and_length): New function, split out from...
13623 (get_string_length): ...here. Also set the lengths of related
13625 (zero_length_string): Assert that chainsi has known (rather than
13627 (adjust_related_strinfos): Likewise.
13629 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
13631 PR tree-optimization/81136
13632 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
13633 assert that two references with the same misalignment have the same
13634 compile-time misalignment if those compile-time misalignments
13637 2017-07-01 Andi Kleen <ak@linux.intel.com>
13639 * print-tree.c (print_node): Print all attributes.
13641 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13643 * cfg.c (scale_bbs_frequencies): New function.
13644 * cfg.h (scale_bbs_frequencies): Declare it.
13645 * cfgloopanal.c (single_likely_exit): Cleanup.
13646 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
13648 (scale_loop_profile): Likewise.
13649 (loop_version): Likewise.
13650 (create_empty_loop_on_edge): Update.
13651 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
13652 scale_loop_frequencies, scale_loop_profile, loopify,
13653 loop_version): Update prototypes.
13654 * modulo-sched.c (sms_schedule): Update.
13655 * predict.c (unlikely_executed_edge_p): Also check probability.
13656 (probably_never_executed_edge_p): Fix typo.
13657 * tree-if-conv.c (version_loop_for_if_conversion): Update.
13658 * tree-parloops.c (gen_parallel_loop): Update.
13659 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
13660 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
13661 * tree-ssa-loop-split.c (split_loop): Update.
13662 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
13663 * tree-vect-loop-manip.c (vect_do_peeling): Update.
13664 (vect_loop_versioning): Update.
13665 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
13667 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13669 * trans-mem.c (split_bb_make_tm_edge): Update profile.
13671 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13673 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
13674 to keep profile consistent.
13676 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13678 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
13679 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
13680 * profile-count.h (max_safe_multiplier): Make unsigned.
13681 (profile_count::guessed_zero): New.
13683 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13685 * bb-reorder.c (fix_up_crossing_landing_pad,
13686 fix_crossing_conditional_branches): Use make_single_succ_edge
13687 to keep profile consistent.
13689 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
13691 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
13694 2017-07-01 Jakub Jelinek <jakub@redhat.com>
13697 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
13698 the right scopes, make sure cond_jump isn't preserved between multiple
13699 iterations. Search for fallthru edge whenever there are 3+ edges and
13700 use find_fallthru_edge for it.
13702 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13704 Patch by Alexander Monakov <amonakov@ispras.ru>
13705 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
13706 probabilities consistently.
13708 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13710 * pa.c (pa_expand_compare_and_swap_loop): Update call of
13711 emit_cmp_and_jump_insns.
13713 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13716 * tree-inline.c (expand_call_inline): Combine profile statuses.
13718 2017-06-30 Andrew Pinski <apinski@cavium.com>
13720 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
13721 fold_stmt returned true.
13723 2017-06-30 Nathan Sidwell <nathan@acm.org>
13725 * ggc.h (empty_string): Delete.
13726 * cfgexpand.c (expand_asm_stmt): Use plain "".
13727 * optabs.c (expand_asm_memory_barrier): Likewise.
13728 * stringpool.c (empty_string): Delete.
13729 (digit_vector, digit_string): Delete.
13730 (ggc_alloc_string): Use plain "", don't optimize single digit
13731 strings. Use ggc_alloc_atomic.
13733 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
13735 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
13736 comparison set and one other set, use the cost of the non-comparison
13739 2017-06-30 Nathan Sidwell <nathan@acm.org>
13741 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
13744 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
13746 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
13747 loops. Remove now unneeded calls to gimple_switch_set_label() that
13748 just set removed labels to NULL_TREE.
13750 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
13752 * tree-ssanames.c (set_range_info_raw): Abstract from ...
13753 (set_range_info): ...here. Only call set_range_info_raw if domain
13755 (set_nonzero_bits): Call set_range_info_raw.
13756 * tree-ssanames.h (set_range_info_raw): New.
13758 2017-06-30 Jakub Jelinek <jakub@redhat.com>
13761 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
13762 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
13763 of nonimmediate_operand and <store_mask_constraint> instead of m
13764 for the input operand. For V8FI iterator, always split if input
13765 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
13766 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
13767 <store_mask_predicate> instead of register_operand and
13768 <store_mask_constraint> instead of v for the input operand. Make
13769 sure both operands aren't MEMs for if not <mask_applied>.
13771 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
13773 * lto-wrapper.c (copy_file) Close both file descriptors before
13776 2017-06-30 Martin Liska <mliska@suse.cz>
13779 * multiple_target.c (create_dispatcher_calls): Make ifunc
13780 also for function that don't have calls or are not referenced.
13782 2017-06-30 Richard Biener <rguenther@suse.de>
13784 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
13785 analyze the first scalar stmt. Move vector type computation
13786 for the BB case here from ...
13787 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
13788 live operation processing in the SLP case properly.
13790 2017-06-30 Richard Biener <rguenther@suse.de>
13792 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
13794 2017-06-30 Martin Liska <mliska@suse.cz>
13797 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
13798 before BUILT_IN_UNWIND_RESUME when ASAN is used.
13800 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
13802 * doc/invoke.texi (AArch64): Add missing options and remove redundant
13805 2017-06-30 Richard Biener <rguenther@suse.de>
13807 PR tree-optimization/81249
13808 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
13809 condition reduction result to original scalar type.
13811 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13813 * profile-count.h (enum profile_quality): Fix typos and whitespace
13816 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13818 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
13819 type for branch probabilities.
13821 2017-06-29 Julian Brown <julian@codesourcery.com>
13822 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13824 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
13825 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
13826 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
13827 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
13829 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
13831 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
13832 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
13833 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
13834 CC usage from generic code to here.
13835 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
13836 CC usage into the target macros.
13838 2017-06-29 Maya Rashish <coypu@sdf.org>
13840 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
13843 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13845 * arm/arm-builtins.c: Include profile-count.h
13846 * except.c (sjlj_emit_function_enter): Use
13847 profile_probability::unlikely.
13849 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13851 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
13852 and tocrel_offset be pointer args rather than implicitly using
13854 (legitimate_constant_pool_address_p, rs6000_emit_move,
13855 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
13856 tocrel_offset and use in toc_relative_expr_p call.
13857 (print_operand, print_operand_address): Use static tocrel_base_oac
13858 and tocrel_offset_oac.
13859 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
13862 2017-06-29 Maya Rashish <coypu@sdf.org>
13864 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
13866 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
13868 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
13869 objects, take into account only the alignment of 'op0' and 'mode1' if
13872 2017-06-29 Steve Ellcey <sellcey@cavium.com>
13874 * ccmp.c (ccmp_tree_comparison_p): New function.
13875 (ccmp_candidate_p): Update to use above function.
13876 (get_compare_parts): New function.
13877 (expand_ccmp_next): Update to use new functions.
13878 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
13880 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
13881 take mode as argument.
13882 * ccmp.h (expand_ccmp_expr): Add mode as argument.
13883 * expr.c (expand_expr_real_1): Pass mode as argument.
13885 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
13887 * combine.c (combine_instructions): Print insns to dump_file, together
13890 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
13892 * asan.c (asan_emit_stack_protection): Update.
13893 (create_cond_insert_point): Update.
13894 * auto-profile.c (afdo_propagate_circuit): Update.
13895 * basic-block.h (struct edge_def): Turn probability to
13896 profile_probability.
13897 (EDGE_FREQUENCY): Update.
13898 * bb-reorder.c (find_traces_1_round): Update.
13899 (better_edge_p): Update.
13900 (sanitize_hot_paths): Update.
13901 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
13902 (make_single_succ_edge): Update.
13903 (check_bb_profile): Update.
13904 (dump_edge_info): Update.
13905 (update_bb_profile_for_threading): Update.
13906 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
13907 probabilitycount to 0.
13908 * cfgbuild.c (compute_outgoing_frequencies): Update.
13909 * cfgcleanup.c (try_forward_edges): Update.
13910 (outgoing_edges_match): Update.
13911 (try_crossjump_to_edge): Update.
13912 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
13913 (expand_gimple_tailcall): Update.
13914 (construct_init_block): Use make_single_succ_edge.
13915 (construct_exit_block): Use make_single_succ_edge.
13916 * cfghooks.c (verify_flow_info): Update.
13917 (redirect_edge_succ_nodup): Update.
13918 (split_edge): Update.
13919 (account_profile_record): Update.
13920 * cfgloopanal.c (single_likely_exit): Update.
13921 * cfgloopmanip.c (scale_loop_profile): Update.
13922 (set_zero_probability): Remove.
13923 (duplicate_loop_to_header_edge): Update.
13924 * cfgloopmanip.h (loop_version): Update prototype.
13925 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
13926 (force_nonfallthru_and_redirect): Update.
13927 (update_br_prob_note): Update.
13928 (rtl_verify_edges): Update.
13929 (purge_dead_edges): Update.
13930 (rtl_lv_add_condition_to_bb): Update.
13931 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
13932 * cgraphunit.c (init_lowered_empty_function): Update.
13933 (cgraph_node::expand_thunk): Update.
13934 * cilk-common.c: Include profile-count.h
13935 * dojump.c (inv): Remove.
13936 (jumpifnot): Update.
13937 (jumpifnot_1): Update.
13938 (do_jump_1): Update.
13940 (do_jump_by_parts_greater_rtx): Update.
13941 (do_compare_rtx_and_jump): Update.
13942 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
13943 do_jump_1. do_compare_rtx_and_jump): Update prototype.
13944 * dwarf2cfi.c: Include profile-count.h
13945 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
13946 (sjlj_emit_dispatch_table): Likewise.
13947 * explow.c: Include profile-count.h
13948 * expmed.c (emit_store_flag_force): Update.
13949 (do_cmp_and_jump): Update.
13950 * expr.c (compare_by_pieces_d::generate): Update.
13951 (compare_by_pieces_d::finish_mode): Update.
13952 (emit_block_move_via_loop): Update.
13953 (store_expr_with_bounds): Update.
13954 (store_constructor): Update.
13955 (expand_expr_real_2): Update.
13956 (expand_expr_real_1): Update.
13957 * expr.h (try_casesi, try_tablejump): Update prototypes.
13958 * gimple-pretty-print.c (dump_probability): Update.
13959 (dump_profile): New.
13960 (dump_gimple_label): Update.
13961 (dump_gimple_bb_header): Update.
13962 * graph.c (draw_cfg_node_succ_edges): Update.
13963 * hsa-gen.c (convert_switch_statements): Update.
13964 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
13965 (find_if_case_1): Update.
13966 (find_if_case_2): Update.
13967 * internal-fn.c (expand_arith_overflow_result_store): Update.
13968 (expand_addsub_overflow): Update.
13969 (expand_neg_overflow): Update.
13970 (expand_mul_overflow): Update.
13971 (expand_vector_ubsan_overflow): Update.
13972 * ipa-cp.c (good_cloning_opportunity_p): Update.
13973 * ipa-split.c (split_function): Use make_single_succ_edge.
13974 * ipa-utils.c (ipa_merge_profiles): Update.
13975 * loop-doloop.c (add_test): Update.
13976 (doloop_modify): Update.
13977 * loop-unroll.c (compare_and_jump_seq): Update.
13978 (unroll_loop_runtime_iterations): Update.
13979 * lra-constraints.c (lra_inheritance): Update.
13980 * lto-streamer-in.c (input_cfg): Update.
13981 * lto-streamer-out.c (output_cfg): Update.
13982 * mcf.c (adjust_cfg_counts): Update.
13983 * modulo-sched.c (sms_schedule): Update.
13984 * omp-expand.c (expand_omp_for_init_counts): Update.
13985 (extract_omp_for_update_vars): Update.
13986 (expand_omp_ordered_sink): Update.
13987 (expand_omp_for_ordered_loops): Update.
13988 (expand_omp_for_generic): Update.
13989 (expand_omp_for_static_nochunk): Update.
13990 (expand_omp_for_static_chunk): Update.
13991 (expand_cilk_for): Update.
13992 (expand_omp_simd): Update.
13993 (expand_omp_taskloop_for_outer): Update.
13994 (expand_omp_taskloop_for_inner): Update.
13995 * omp-simd-clone.c (simd_clone_adjust): Update.
13996 * optabs.c (expand_doubleword_shift): Update.
13997 (expand_abs): Update.
13998 (emit_cmp_and_jump_insn_1): Update.
13999 (expand_compare_and_swap_loop): Update.
14000 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
14001 * predict.c (predictable_edge_p): Update.
14002 (edge_probability_reliable_p): Update.
14003 (set_even_probabilities): Update.
14004 (combine_predictions_for_insn): Update.
14005 (combine_predictions_for_bb): Update.
14006 (propagate_freq): Update.
14007 (estimate_bb_frequencies): Update.
14008 (force_edge_cold): Update.
14009 * profile-count.c (profile_count::dump): Add missing space into dump.
14010 (profile_count::debug): Add newline.
14011 (profile_count::differs_from_p): Explicitly convert to unsigned.
14012 (profile_count::stream_in): Update.
14013 (profile_probability::dump): New member function.
14014 (profile_probability::debug): New member function.
14015 (profile_probability::differs_from_p): New member function.
14016 (profile_probability::differs_lot_from_p): New member function.
14017 (profile_probability::stream_in): New member function.
14018 (profile_probability::stream_out): New member function.
14019 * profile-count.h (profile_count_quality): Rename to ...
14020 (profile_quality): ... this one.
14021 (profile_probability): New.
14022 (profile_count): Update.
14023 * profile.c (compute_branch_probabilities): Update.
14024 * recog.c (peep2_attempt): Update.
14025 * sched-ebb.c (schedule_ebbs): Update.
14026 * sched-rgn.c (find_single_block_region): Update.
14027 (compute_dom_prob_ps): Update.
14028 (schedule_region): Update.
14029 * sel-sched-ir.c (compute_succs_info): Update.
14030 * stmt.c (struct case_node): Update.
14031 (do_jump_if_equal): Update.
14032 (get_outgoing_edge_probs): Update.
14033 (conditional_probability): Update.
14034 (emit_case_dispatch_table): Update.
14035 (expand_case): Update.
14036 (expand_sjlj_dispatch_table): Update.
14037 (emit_case_nodes): Update.
14038 * targhooks.c: Update.
14039 * tracer.c (better_p): Update.
14040 (find_best_successor): Update.
14041 * trans-mem.c (expand_transaction): Update.
14042 * tree-call-cdce.c: Update.
14043 * tree-cfg.c (gimple_split_edge): Upate.
14044 (move_sese_region_to_fn): Upate.
14045 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
14046 * tree-eh.c (lower_resx): Upate.
14047 (cleanup_empty_eh_move_lp): Upate.
14048 * tree-if-conv.c (version_loop_for_if_conversion): Update.
14049 * tree-inline.c (copy_edges_for_bb): Update.
14050 (copy_cfg_body): Update.
14051 * tree-parloops.c (gen_parallel_loop): Update.
14052 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
14053 (gimple_gen_time_profiler): Update.
14054 * tree-ssa-dce.c (remove_dead_stmt): Update.
14055 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14056 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
14057 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
14058 (unloop_loops): Update.
14059 (try_peel_loop): Update.
14060 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14061 * tree-ssa-loop-split.c (connect_loops): Update.
14062 (split_loop): Update.
14063 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
14064 (hoist_guard): Update.
14065 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
14066 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
14067 (value_replacement): Update.
14068 * tree-ssa-reassoc.c (branch_fixup): Update.
14069 * tree-ssa-tail-merge.c (replace_block_by): Update.
14070 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
14071 (create_edge_and_update_destination_phis): Update.
14072 (compute_path_counts): Update.
14073 (recompute_probabilities): Update.
14074 (update_joiner_offpath_counts): Update.
14075 (freqs_to_counts_path): Update.
14076 (duplicate_thread_path): Update.
14077 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
14078 (struct switch_conv_info): Update.
14079 (gen_inbound_check): Update.
14080 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14081 (vect_do_peeling): Update.
14082 (vect_loop_versioning): Update.
14083 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14084 (optimize_mask_stores): Update.
14085 * ubsan.c (ubsan_expand_null_ifn): Update.
14086 * value-prof.c (gimple_divmod_fixed_value): Update.
14087 (gimple_divmod_fixed_value_transform): Update.
14088 (gimple_mod_pow2): Update.
14089 (gimple_mod_pow2_value_transform): Update.
14090 (gimple_mod_subtract): Update.
14091 (gimple_mod_subtract_transform): Update.
14092 (gimple_ic): Update.
14093 (gimple_stringop_fixed_value): Update.
14094 (gimple_stringops_transform): Update.
14095 * value-prof.h: Update.
14097 2017-06-29 Carl Love <cel@us.ibm.com>
14099 * config/rs6000/rs6000-c.c: Add support for built-in functions
14100 vector signed int vec_signed (vector float);
14101 vector signed long long vec_signed (vector double);
14102 vector signed int vec_signed2 (vector double, vector double);
14103 vector signed int vec_signede (vector double);
14104 vector signed int vec_signedo (vector double);
14105 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
14106 instruction generator.
14107 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14108 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
14109 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
14111 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
14112 vunsignede_v2df): Add define_expands.
14113 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
14114 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
14115 VEC_UNSIGNEDO): Add definitions.
14116 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14117 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
14118 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
14119 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
14120 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
14121 * config/rs6000/altivec.h (vec_signed, vec_signed2,
14122 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
14123 vec_unsignede, vec_unsignedo): Add builtin defines.
14124 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
14126 * doc/extend.texi: Update the built-in documentation file for the
14127 new built-in functions.
14129 2017-06-29 Richard Biener <rguenther@suse.de>
14131 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
14132 reduction chains to LOOP_VINFO_REDUCTIONS.
14133 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
14134 SLP reductions after processing reduction chains.
14136 2017-06-29 Nathan Sidwell <nathan@acm.org>
14138 * builtins.c (fold_builtin_FUNCTION): Use
14139 lang_hooks.decl_printable_name.
14141 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
14143 PR middle-end/81194
14144 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
14145 with only one label.
14146 * stmt.c (expand_case): Assert NCASES is greater than one.
14148 2017-06-29 Richard Biener <rguenther@suse.de>
14150 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
14152 (group_case_labels): Likewise.
14153 (find_taken_edge): Push sanity checking on val to workers...
14154 (find_taken_edge_cond_expr): ... here
14155 (find_taken_edge_switch_expr): ... and here, handle cases
14156 with just a default label.
14157 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
14158 (group_case_labels): Likewise.
14159 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
14160 group_case_labels does anything cleanup the CFG again.
14162 2017-06-29 Bin Cheng <bin.cheng@arm.com>
14164 PR tree-optimization/81196
14165 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
14166 exit condition comparing two IVs.
14168 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
14170 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
14171 profile to the dummy entry at the end of the list of architectures.
14172 * config/arm/arm-cpu-cdata.h: Regenerated.
14174 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14175 Michael Collison <michael.collison@arm.com>
14178 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
14180 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
14181 (*aarch64_reg_<mode>3_minus_mask): New pattern.
14182 (*aarch64_<optab>_reg_di3_mask2): New pattern.
14183 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
14184 of shift when the shift amount is masked with constant equal to
14185 the size of the mode.
14186 * config/aarch64/predicates.md (subreg_lowpart_operator): New
14189 2017-06-29 Martin Liska <mliska@suse.cz>
14191 * config/i386/i386.opt: Change range from [1,5] to [0,5].
14193 2017-06-29 Yury Gribov <tetra2005@gmail.com>
14196 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
14199 (edge_growth_cache_entry::edge_growth_cache_entry): New
14201 (reset_edge_growth_cache): Update to use constructor.
14203 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14205 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
14206 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
14207 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
14209 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
14211 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
14212 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
14214 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
14216 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
14217 (*-linux-uclibc*): Add t-uclibc tmake_file.
14218 * config/t-musl: New.
14219 * config/t-uclibc: New.
14221 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
14223 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
14225 (gen_comm_data): Emit architectural setting of arch_prof.
14226 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
14228 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
14229 (armv8-m.base, armv8-m.main): Likewise.
14230 * arm-protos.h (arm_build_target): Add profile field.
14231 (arch_option): Likewise.
14232 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
14234 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
14235 arm_active_target.profile.
14237 2017-06-28 Richard Biener <rguenther@suse.de>
14239 PR middle-end/81227
14240 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
14241 TYPE_OVERFLOW_WRAPS.
14242 * match.pd (negate_expr_p): Likewise.
14243 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
14244 fold_build2, not fold_binary.
14246 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
14248 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
14249 Convert memory address to Pmode.
14250 (aarch64_print_operand): Assert MEM operands are always Pmode.
14252 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
14255 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
14256 Remove redundant if.
14257 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
14258 * config/arm/aarch-common-protos.h
14259 (aarch_forward_to_shift_is_not_shifted_re): Remove.
14260 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
14262 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
14265 * multiple_target.c (create_dispatcher_calls): Set the default
14266 clone to be static, not public.
14268 2017-06-28 Richard Biener <rguenther@suse.de>
14270 * tree-vect-loop.c (vectorizable_reduction): Move special
14271 cond reduction IV var creation ...
14272 (vect_create_epilog_for_reduction): ... here. Remove induction_index
14273 parameter. Use STMT_VINFO_VECTYPE.
14274 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
14277 2017-06-28 Martin Liska <mliska@suse.cz>
14280 * ipa-visibility.c (non_local_p): Handle visibility.
14282 2017-06-28 Martin Liska <mliska@suse.cz>
14285 * common.opt: Add IntegerRange to various options.
14286 * opt-functions.awk (integer_range_info): New function.
14287 * optc-gen.awk: Add integer_range_info to cl_options struct.
14288 * opts-common.c (decode_cmdline_option): Handle
14289 CL_ERR_INT_RANGE_ARG.
14290 (cmdline_handle_error): Likewise.
14291 * opts.c (print_filtered_help): Show valid interval in
14292 when --help is provided.
14293 * opts.h (struct cl_option): Add range_min and range_max fields.
14294 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
14296 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
14298 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
14299 (x * C EQ/NE y * C): New transformation.
14301 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
14303 * genmultilib (combination_space): Accept '+' in option names.
14305 2017-06-28 Martin Liska <mliska@suse.cz>
14308 * asan.c (instrument_derefs): Bail out inner references
14309 that are hard register variables.
14311 2017-06-28 Jakub Jelinek <jakub@redhat.com>
14314 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
14315 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
14317 2017-06-28 Richard Biener <rguenther@suse.de>
14319 * tree-vectorizer.h (vect_get_vec_defs): Remove.
14320 (vect_get_slp_defs): Adjust.
14321 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
14323 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
14325 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
14326 get_initial_defs_for_reduction instead of vect_get_vec_defs.
14327 (vectorizable_reduction): Adjust.
14328 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
14330 (vect_get_slp_defs): Likewise.
14331 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
14332 (vectorizable_bswap): Adjust.
14333 (vectorizable_call): Likewise.
14334 (vectorizable_conversion): Likewise.
14335 (vectorizable_assignment): Likewise.
14336 (vectorizable_shift): Likewise.
14337 (vectorizable_operation): Likewise.
14338 (vectorizable_store): Likewise.
14339 (vectorizable_condition): Likewise.
14340 (vectorizable_comparison): Likewise.
14342 2017-06-28 Michael Collison <michael.collison@arm.com>
14345 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
14347 (arm_gen_movmemqi): Removed unused variable 'i'.
14348 Convert 'for' loop into 'while' loop.
14349 (arm_expand_prologue): Remove last unnecessary set of insn.
14350 (thumb_pop): Remove unused variable 'pushed_words'.
14351 (thumb_exit): Remove last unnecessary set of regs_to_pop.
14353 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14355 * config/s390/predicates.md: Use s390_rel_address_ok_p.
14356 * config/s390/s390-protos.h: Add prototype of
14357 s390_rel_address_ok_p.
14358 * config/s390/s390.c (s390_got_symbol): New function.
14359 (s390_rel_address_ok_p): New function.
14360 (legitimize_pic_address): Use s390_rel_address_ok_p.
14361 (s390_load_got): Use s390_got_symbol.
14362 (s390_option_override): Issue error if
14363 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
14364 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
14366 * config/s390/s390.opt: New option mpic-data-is-text-relative.
14368 2017-06-27 Andrew Pinski <apinski@cavium.com>
14370 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
14371 (X * copysign (1.0, X)): New pattern.
14372 (X * copysign (1.0, -X)): New pattern.
14373 (copysign (-1.0, CST)): New pattern.
14375 2017-06-27 Joseph Myers <joseph@codesourcery.com>
14377 * genmultilib (combination_space): Remove variable.
14378 Validate reuse rules against regular expression for any sequence
14379 of multilib options in any order.
14381 2017-06-27 Michael Collison <michael.collison@arm.com>
14383 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
14384 call aarch64_split_simd_combine.
14385 * (aarch64_combine_internal<mode>): Delete pattern.
14386 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
14387 Allow register and subreg operands.
14389 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14391 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
14392 specific need, just fallback on defaults.
14393 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
14395 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14396 Olivier Hainque <hainque@adacore.com>
14398 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
14400 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
14401 targets. Pick a default if no particular attempt applied.
14402 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
14405 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14407 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
14408 (x86_64-wrs-vxworks7): Likewise.
14410 2017-06-27 Marek Polacek <polacek@redhat.com>
14413 * ubsan.c (instrument_null): Check get_base_address's result for null.
14415 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
14417 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
14419 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
14421 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
14422 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
14423 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
14424 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
14425 New function types.
14426 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
14427 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
14428 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
14429 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
14430 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
14431 BUILT_IN_FEUPDATEENV): New builtins.
14432 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
14433 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
14434 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
14435 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
14437 (builtin_structptr_types): Adjust size.
14438 * tree.c (builtin_structptr_types): Add four entries.
14440 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14441 Olivier Hainque <hainque@adacore.com>
14443 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
14444 (TLS_SYM): New local macro, forcing reference to __tls__ on
14445 link command lines for VxWorks 7 RTPs, triggering initialization
14447 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
14448 OS features TLS support, true for RTPs on VxWorks 7.
14449 * config/vxworks.c (vxworks_override_options): Setup emutls
14452 2017-06-27 Jakub Jelinek <jakub@redhat.com>
14454 * predict.c (test_prediction_value_range): Use -1U instead of -1
14455 to avoid narrowing conversion warning.
14456 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
14457 to avoid narrowing conversion warning.
14458 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
14460 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
14462 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14464 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
14465 64bit configurations.
14466 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
14467 (SIZE_TYPE): Likewise.
14468 * config/vxworks.c (vxworks_emutls_var_fields): Use
14469 long_unsigned_type_node instead of unsigned_type_node as the offset
14470 field type, which is "pointer" mode in emutls.c.
14472 2017-06-27 Jakub Jelinek <jakub@redhat.com>
14475 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
14477 PR middle-end/81207
14478 * gimple-fold.c (replace_call_with_call_and_fold): Handle
14479 gimple_vuse copying separately from gimple_vdef copying.
14481 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14483 * value-prof.c (free_hist): Remove call to memset and the enclosing if
14486 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14487 Olivier Hainque <hainque@adacore.com>
14489 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
14490 for all vxworks7 targets.
14491 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
14492 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
14493 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
14494 variations for VX6/VX7 and 32/64bits later on in ...
14495 (VXWORKS_LIB_SPEC): Leverage new macros.
14496 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
14497 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
14499 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14501 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
14502 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
14504 2017-06-26 Carl Love <cel@us.ibm.com>
14506 * config/rs6000/rs6000-c.c: Add support for built-in functions
14507 vector bool char vec_reve (vector bool char);
14508 vector signed char vec_reve (vector signed char);
14509 vector unsigned char vec_reve (vector unsigned char);
14510 vector bool int vec_reve (vector bool int);
14511 vector signed int vec_reve (vector signed int);
14512 vector unsigned int vec_reve (vector unsigned int);
14513 vector bool long long vec_reve (vector bool long long);
14514 vector signed long long vec_reve (vector signed long long);
14515 vector unsigned long long vec_reve (vector unsigned long long);
14516 vector bool short vec_reve (vector bool short);
14517 vector signed short vec_reve (vector signed short);
14518 vector double vec_reve (vector double);
14519 vector float vec_reve (vector float);
14520 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
14521 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
14522 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
14523 (altivec_vreve): New pattern.
14524 * config/rs6000/altivec.h (vec_reve): New define.
14525 * doc/extend.texi (vec_rev): Update the built-in documentation file
14526 for the new built-in functions.
14528 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14530 PR tree-optimization/71815
14531 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
14533 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
14535 (slsr_process_phi): Likewise.
14536 (replace_uncond_cands_and_profitable_phis): Don't replace a
14537 multiply candidate with a stride of 1 (copy or cast).
14538 (phi_incr_cost): Call uses_consumed_by_stmt rather than
14540 (lowest_cost_path): Likewise.
14541 (total_savings): Likewise.
14543 2017-06-26 Richard Biener <rguenther@suse.de>
14546 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
14547 Use def_builtin_pure for all gather builtins.
14549 2017-06-26 Richard Biener <rguenther@suse.de>
14551 PR tree-optimization/81203
14552 * tree-tailcall.c (find_tail_calls): Do not move stmts into
14553 non-dominating BBs.
14555 2017-06-26 Marek Polacek <polacek@redhat.com>
14558 * doc/invoke.texi: Document -Wmultistatement-macros.
14560 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
14562 * doc/sourcebuild.texi (ARM-specific attributes): Document new
14563 arm_neon_ok_no_float_abi effective target.
14565 2017-06-26 Richard Biener <rguenther@suse.de>
14567 PR tree-optimization/80928
14568 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
14569 (copy_bbs): Set BB_DUPLICATED flag early.
14570 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
14572 (execute_on_shrinking_pred): Likewise.
14573 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
14574 BB_DUPLICATED blocks.
14575 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
14576 iterate over all PHIs considering removal of *gsi.
14578 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
14580 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
14583 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14585 * config/rs6000/rs6000-string.c: (expand_block_clear,
14586 do_load_for_compare, select_block_compare_mode,
14587 compute_current_alignment, expand_block_compare,
14588 expand_strncmp_align_check, expand_strn_compare,
14589 expand_block_move, rs6000_output_load_multiple)
14590 Move functions related to string/block move/compare
14591 to a separate file.
14592 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
14593 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
14594 for this function which is now used in two files.
14595 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
14596 * config.gcc: Add rs6000-string.o to extra_objs for
14597 targets powerpc*-*-* and rs6000*-*-*.
14599 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
14602 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
14603 32-bit, since indexed is not valid for DImode.
14604 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
14605 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
14606 (define_peephole2 for Altivec d-form load): Add 32-bit support.
14607 (define_peephole2 for Altivec d-form store): Likewise.
14610 * multiple_target.c (create_dispatcher_calls): Only create the
14611 dispatcher call if the function is the default clone of a
14612 versioned function.
14614 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
14616 PR middle-end/80902
14617 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
14618 a call, force the call to not be a tail call.
14620 2017-06-23 Jeff Law <law@redhat.com>
14622 * doc/contrib.texi: Add entry for Steven Pemberton's work on
14625 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
14627 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
14628 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
14629 handling for early expansion of vector shifts (sl,sr,sra,rl).
14630 (builtin_function_type): Add vector shift right instructions
14631 to the unsigned argument list.
14633 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
14635 rtl-optimizatoin/79286
14636 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
14637 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
14638 trap. PIC register plus a const unspec without offset can never trap.
14640 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
14642 * tree.h (builtin_structptr_type): New type.
14643 (builtin_structptr_types): Declare new array.
14644 * tree.c (builtin_structptr_types): New array.
14645 (free_lang_data, build_common_tree_nodes): Use it.
14647 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
14650 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
14653 2017-06-22 Matt Turner <mattst88@gmail.com>
14655 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
14656 Lake models to skylake case. Assume skylake for unknown
14657 models with clflushopt.
14659 2017-06-22 Jeff Law <law@redhat.com>
14661 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
14662 frame sizes that do not satisfy aarch64_uimm12_shift.
14664 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
14666 * profile-count.h (apply_probability,
14667 apply_scale, probability_in): Fix checks for zero.
14669 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14671 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
14672 * doc/cppdiropts.texi (-I @var{dir}): Document it.
14674 2016-06-22 Richard Biener <rguenther@suse.de>
14676 * tree-vect-loop.c (vect_model_reduction_cost): Handle
14677 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
14678 REDUC_MAX_EXPR support.
14679 (vectorizable_reduction): Likewise.
14680 (vect_create_epilog_for_reduction): Likewise.
14682 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
14684 * match.pd (A / (1 << B) -> A >> B): New.
14685 * generic-match-head.c: Include optabs-tree.h.
14686 * gimple-match-head.c: Likewise.
14687 * optabs-tree.h (target_supports_op_p): New.
14688 * optabs-tree.c (target_supports_op_p): New.
14690 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14692 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
14693 $gcc_cv_ld --help output.
14694 (gcc_cv_ld_demangle): Likewise.
14695 (gcc_cv_ld_eh_frame_hdr): Likewise.
14696 (gcc_cv_ld_pie): Likewise.
14697 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
14698 (gcc_cv_ld_buildid): Likewise.
14699 (gcc_cv_ld_sysroot): Likewise.
14700 (ld_bndplt_support): Likewise.
14701 (ld_pushpopstate_support): Likewise.
14702 * configure: Regenerate.
14703 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
14705 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14708 * config/i386/sse.md (round<mode>2): Renumber match_dup and
14709 operands indexes to avoid gap between operands and match_dups.
14711 2017-06-21 Andrew Pinski <apinski@cavium.com>
14713 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
14714 Increment Arith_shift and Arith_shift_reg by 1.
14715 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
14717 * config/aarch64/aarch64.c (thunderx_tunings): Enable
14718 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
14719 (aarch64_strip_extend): Add new argument and test for it.
14720 (aarch64_cheap_mult_shift_p): New function.
14721 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
14722 add a cost if it is true.
14723 Update calls to aarch64_strip_extend.
14724 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
14726 2017-06-21 Andrew Pinski <apinski@cavium.com>
14728 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
14730 (thunderxt88): Likewise.
14731 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
14732 (thunderx_prefetch_tune): New variable.
14733 (thunderx2t99_prefetch_tune): Update for the correct values.
14734 (thunderxt88_tunings): New variable.
14735 (thunderx_tunings): Use thunderx_prefetch_tune instead of
14736 generic_prefetch_tune.
14737 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
14739 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14741 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
14742 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
14743 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
14744 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
14745 (aarch64_atomic_cas<mode>, GPI): Likewise.
14747 2017-06-21 Martin Liska <mliska@suse.cz>
14749 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
14750 statements on cold and hot labels.
14751 * predict.c (tree_estimate_probability_bb): Remove the
14752 prediction from this place.
14754 2017-06-21 Martin Liska <mliska@suse.cz>
14756 PR tree-optimization/79489
14757 * gimplify.c (maybe_add_early_return_predict_stmt): New
14759 (gimplify_return_expr): Call the function.
14760 * predict.c (tree_estimate_probability_bb): Remove handling
14762 * predict.def: Update comment about early return predictor.
14763 * gimple-predict.h (is_gimple_predict): New function.
14764 * predict.def: Change default value of early return to 66.
14765 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
14767 * passes.def: Put pass_strip_predict_hints to the beginning of
14770 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
14772 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
14773 FUNCTION_DECL declarations.
14774 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
14776 (dwaf2out_decl): Likewise.
14777 * godump.c (go_early_global_decl): Skip call to the real debug hook
14778 for FUNCTION_DECL declarations.
14779 * passes.c (rest_of_decl_compilation): Skip call to the
14780 early_global_decl debug hook for FUNCTION_DECL declarations, unless
14781 -fdump-go-spec is passed.
14783 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14785 * config/i386/i386.c (struct builtin_isa): New field pure_p.
14786 Reorder for compactness.
14787 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
14788 (def_builtin_pure, def_builtin_pure2): New functions.
14789 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
14791 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
14793 * match.pd (nop_convert): New predicate.
14794 ((A +- CST1) +- CST2): Allow some NOP conversions.
14796 2017-06-21 Jakub Jelinek <jakub@redhat.com>
14799 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
14800 with ctors/dtors if GOVD_SHARED is set.
14802 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14804 * config/aarch64/aarch64.md (movti_aarch64):
14805 Emit mov rather than orr.
14806 (movtf_aarch64): Likewise.
14807 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
14808 Emit mov rather than orr.
14810 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14812 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
14813 Swap alternatives, make integer dup more expensive.
14815 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
14817 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
14818 Return true for non-tls symbols.
14820 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
14822 * config/aarch64/aarch64-cores.def (cortex-a55): New.
14823 (cortex-a75): Likewise.
14824 (cortex-a75.cortex-a55): Likewise.
14825 * config/aarch64/aarch64-tune.md: Regenerate.
14826 * doc/invoke.texi (-mtune): Document new values for -mtune.
14828 2017-06-21 Tom de Vries <tom@codesourcery.com>
14830 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
14831 stack_size feature.
14832 (Effective-Target Keywords, Other attributes): Suggest using
14833 dg-add-options stack_size feature to get stack limit in stack_size
14834 effective target documentation.
14836 2017-06-21 Julian Brown <julian@codesourcery.com>
14837 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14839 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
14840 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
14841 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
14843 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
14844 attribute type list for neon_multiply.
14845 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
14846 attribute type list for neon_multiply.
14847 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
14848 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
14849 attribute type list for neon_multiply.
14850 * config/arm/types.md (crypto_pmull): Add.
14851 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
14852 attribute type list.
14854 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
14856 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
14859 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14861 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
14862 to make sure not to dereference a NULL cost_classes_ptr pointer.
14864 2017-06-20 Carl Love <cel@us.ibm.com>
14866 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14867 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14868 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14869 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14870 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14871 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14872 VMULOSW): New enum "unspec" values.
14873 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14874 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
14875 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14876 altivec_vmulosw): New patterns.
14877 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14878 VMULOSW): Add definitions.
14880 2017-06-20 Julia Koval <julia.koval@intel.com>
14882 * config/i386/i386.c: Fix rounding expand for new pattern.
14883 * config/i386/subst.md: Fix pattern (parallel -> unspec).
14885 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14887 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
14888 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
14890 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14892 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
14895 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
14897 * config/aarch64/aarch64-cores.def: Rearrange to sort by
14898 architecture, then by implementer ID.
14899 * config/aarch64/aarch64-tune.md: Regenerate.
14901 2017-06-20 Richard Biener <rguenther@suse.de>
14903 PR middle-end/81097
14904 * fold-const.c (split_tree): Fold to type before negating.
14906 2017-06-20 David Malcolm <dmalcolm@redhat.com>
14908 * diagnostic-show-locus.c
14909 (selftest::test_fixit_deletion_affecting_newline): New function.
14910 (selftest::diagnostic_show_locus_c_tests): Call it.
14912 2017-06-20 Andreas Schwab <schwab@suse.de>
14915 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
14918 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14920 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
14921 __ARM_FEATURE_COPROC according to support.
14923 2017-06-20 Jakub Jelinek <jakub@redhat.com>
14925 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
14926 Rewritten to avoid overflow for > 32-bit pointers.
14929 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
14930 by removing enum keyword.
14931 (ubsan_type_descriptor): Likewise. Formatting fix.
14934 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
14935 splitter): Require TARGET_SSE2 in the condition.
14937 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
14940 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
14941 for doing vector set of SFmode on ISA 3.0.
14942 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
14943 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
14945 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
14946 SFmode value into a V4SF variable that was extracted from another
14947 V4SF variable without converting the element to double precision
14948 and back to single precision vector format.
14949 (vsx_insert_extract_v4sf_p9_2): Likewise.
14951 2017-06-19 Jakub Jelinek <jakub@redhat.com>
14953 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
14954 in UWHI to avoid undefined overflow.
14957 * ubsan.h (enum ubsan_encode_value_phase): New.
14958 (ubsan_encode_value): Change second argument to
14959 enum ubsan_encode_value_phase with default value of
14960 UBSAN_ENCODE_VALUE_GENERIC.
14961 * ubsan.c (ubsan_encode_value): Change second argument to
14962 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
14963 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
14964 create_tmp_var_raw instead of create_tmp_var and use a
14966 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
14967 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
14968 ubsan_encode_value callers.
14971 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
14972 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
14973 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
14975 2017-06-19 Richard Biener <rguenther@suse.de>
14977 PR middle-end/81118
14978 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
14979 estimates if we changed anything.
14981 2017-06-19 Richard Biener <rguenther@suse.de>
14983 PR tree-optimization/80887
14984 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
14985 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
14986 simplified lookups, then reset mprts_hook.
14987 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
14989 (try_to_simplify): Likewise.
14991 2017-06-19 Martin Liska <mliska@suse.cz>
14994 * gimplify.c (gimplify_switch_expr):
14995 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
14997 2017-06-19 Martin Liska <mliska@suse.cz>
14999 * doc/install.texi: Document that PGO runs in 4 stages.
15001 2017-06-19 Martin Liska <mliska@suse.cz>
15004 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
15005 to dispatcher function name.
15006 * multiple_target.c (replace_function_decl): New function.
15007 (create_dispatcher_calls): Redirect both edges and references.
15009 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
15011 * profile-count.c (profile_count::dump): Dump quality.
15012 (profile_count::differs_from_p): Update for unsigned val.
15013 * profile-count.h (profile_count_quality): New enum.
15014 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
15016 2017-06-19 Richard Biener <rguenther@suse.de>
15018 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
15019 struct function as arg.
15020 (estimate_numbers_of_iterations): Export overload with loop arg.
15021 (free_numbers_of_iterations_estimates_loop): Use an overload of
15022 free_numbers_of_iterations_estimates instead.
15023 * tree-cfg.c (remove_bb): Adjust.
15024 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
15025 * tree-parloops.c (gen_parallel_loop): Likewise.
15026 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
15028 (tree_unroll_loops_completely): Likewise.
15029 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
15030 Use an overload instead and export.
15031 (estimated_loop_iterations): Adjust.
15032 (max_loop_iterations): Likewise.
15033 (likely_max_loop_iterations): Likewise.
15034 (estimate_numbers_of_iterations): Take struct function as arg
15036 (loop_exits_before_overflow): Adjust.
15037 (free_numbers_of_iterations_estimates_loop): Use an overload.
15038 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
15039 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
15041 2017-06-19 Richard Biener <rguenther@suse.de>
15044 * ipa-prop.c (find_constructor_constant_at_offset): Handle
15045 RANGE_EXPR conservatively.
15047 2017-06-16 Carl Love <cel@us.ibm.com>
15049 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15050 definitions for vec_float, vec_float2, vec_floato,
15051 vec_floate built-ins.
15052 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
15053 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
15055 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
15056 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
15057 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
15058 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
15059 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
15060 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
15061 vec_floato): Add builtin defines.
15062 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
15063 Update the built-in documentation file for the new built-in
15066 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15068 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
15069 (mthumb): Mark as the negative of -marm.
15071 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15073 * doc/invoke.texi (ARM Options, -mcpu): Document supported
15075 (ARM Options, -mtune): Document that this accepts the same
15076 extension options as -mcpu.
15077 (ARM Options, -mfpu): Document addition of -mfpu=auto.
15079 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15081 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
15082 permitted extensions.
15084 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15086 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
15087 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
15088 (armv8-m.main): Add option +nodsp.
15089 * config/arm/arm-cpu-cdata.h: Regenerated.
15091 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15093 * config/arm/t-fuchsia: New file.
15094 * config.gcc (arm*-*-fuchsia*): Use it.
15096 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15098 * config/arm/t-symbian: Rewrite for new option infrastructure.
15100 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15102 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
15103 (MULTILIB_REQUIRED): Likewise.
15105 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15107 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
15108 (MULTILIB_RESUE): Likewise.
15109 (MULTILIB_MATCHES): Likewise.
15110 (MULTLIB_REQUIRED): Likewise.
15112 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15114 * config/arm/t-rtems: Rewrite for new option framework.
15116 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15118 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
15119 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
15120 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
15121 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
15122 * config/arm/t-multilib: ... here.
15123 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
15124 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
15125 armv7-a and armv8*-a when A-profile libraries have not been built.
15126 * config/arm/t-rmprofile: Rewrite.
15128 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15130 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
15131 with a backslash. Remove the backslash after substituting unescaped
15133 * doc/fragments.texi (MULTILIB_REUSE): Document it.
15135 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15137 * config.gcc: (arm*-*-*): When building a-profile libraries, force
15138 the driver to pass through the default setting of -mfloat-abi.
15139 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
15141 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
15142 (all_feat_combs): New rule.
15143 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
15145 * config/arm/t-aprofile: Rewrite.
15147 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15149 * config/arm/arm.h (FPUTYPE_AUTO): Define.
15150 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
15151 fpu is not specified by the user/command-line.
15152 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
15153 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
15154 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
15155 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
15156 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
15157 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
15159 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15161 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
15162 * config/arm/t-arm-elf: Rewritten.
15164 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15166 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
15167 have some floating-point instructions.
15168 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
15169 (TARGET_MAYBE_HARD_FLOAT): New macro.
15170 * config/arm/arm-builtins.c (arm_init_builtins): Use
15171 TARGET_MAYBE_HARD_FLOAT.
15172 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
15174 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15176 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
15177 (configargs.h): Include it.
15178 (arm_print_hint_for_fpu_option): New function.
15179 (arm_parse_fpu_option): New function.
15180 (candidate_extension): New class.
15181 (arm_canon_for_multilib): New function.
15182 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
15183 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
15184 (ARCH_CANONICAL_SPECS): New macro.
15185 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
15187 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15189 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
15190 are set after handling multilib fragments. Set target_cpu_default2
15193 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15195 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
15197 (arm*-*-*): Set target_cpu_default2 to a quoted string.
15198 * config/arm/parsecpu.awk (check_cpu): Validate any extension
15200 (check_arch): Likewise.
15201 * config/arm/arm.c (arm_configure_build_target): Handle
15202 TARGET_CPU_DEFAULT being a string constant. Scan any feature
15203 options in the default.
15205 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15207 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
15208 when an option is an alias of another.
15209 * config/arm/parsecpu.awk (optalias): New parser token.
15210 (gen_comm_data): Mark non-alias options as such. Emit entries
15211 for extension aliases.
15212 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
15213 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
15214 (armv6kz, armv6zk, armv6t2): Likewise.
15215 (armv7): Make vfpv3-d16 an alias.
15216 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
15218 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
15219 Sort in canonical order.
15220 (armv8-a): Sort in canonical order.
15221 (armv8.1-a, armv8.2-a): Likewise.
15222 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
15224 (cortex-a9): Sort in canonical order.
15225 * config/arm/arm.c (selftests.h): Include it.
15226 (arm_test_cpu_arch_data): New function.
15227 (arm_run_self_tests): New function.
15228 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
15229 (targetm): Move declaration to the end of the file.
15230 * arm-cpu-cdata.h: Regenerated.
15232 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15234 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
15235 call to target_mode_check describing the type of option passed.
15236 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
15237 (arm_target_thumb_only): Use arm_parse_arch_option_name or
15238 arm_parse_cpu_option_name to match parameters against list of
15240 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
15241 arm_arch_core_flags data structure.
15242 * config/arm/arm-cpu_cdata.h: Regenerated.
15244 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15246 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
15248 (arm_print_hint_for_cpu_option): Likewise.
15249 (arm_print_hint_for_arch_option): Likewise.
15250 (arm_parse_cpu_option_name): Likewise.
15251 (arm_parse_arch_option_name): Likewise.
15252 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
15253 of entries in the all_fpus list.
15254 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
15255 (arm_parse_cpu_option_name): Declare.
15256 (arm_parse_arch_option_name): Declare.
15257 (arm_parse_option_features): Declare.
15258 (arm_intialize_isa): Declare.
15259 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
15261 (gen_comm_data): ... here. Make definitions non-static.
15262 * config/arm/arm-cpu-data.h: Regenerated.
15263 * config/arm/arm-cpu-cdata.h: Regenerated.
15265 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15267 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
15268 (cpu_arch_extension): New structure.
15269 (cpu_arch_option, arch_option, cpu_option): New structures.
15270 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
15271 architecture types.
15272 (gen_data): Generate new format data tables.
15273 * config/arm/arm.c (cpu_tune): New structure.
15274 (cpu_option, processors): Delete.
15275 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
15276 (arm_print_hint_for_cpu_option): ... this and ...
15277 (arm_print_hint_for_arch_option): ... this.
15278 (arm_parse_arch_cpu_name): Delete. Replace with ...
15279 (arm_parse_cpu_option_name): ... this and ...
15280 (arm_parse_arch_option_name): ... this.
15281 (arm_unrecognized_feature): Change type of target parameter to
15283 (arm_parse_arch_cpu_features): Delete. Replace with ...
15284 (arm_parse_option_features): ... this.
15285 (arm_configure_build_target): Rework to use new configuration data
15287 (arm_print_tune_info): Rework for new configuration data tables.
15288 * config/arm/arm-cpu-data.h: Regenerated.
15289 * config/arm/arm-cpu.h: Regenerated.
15291 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15293 * Makefile.in (OBJS): Move sbitmap.o from here ...
15294 (OBJS-libcommon): ... to here.
15296 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15298 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
15299 (ISA_ALL_CRYPTO): New macro.
15300 (ISA_ALL_SIMD): New macro
15301 (ISA_ALL_FP): New macro.
15302 * config/arm/arm.c (fpu_bitlist): Update initializer.
15303 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
15305 (arm9e): Add fpu. Add option for nofp
15306 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
15307 (arm926ej-s, arm1026ej-s): Likewise.
15308 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
15309 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
15310 neon-fp16, neon-vfpv4, nofp and nosimd.
15311 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
15312 (cortex-a8): Add fpu. Add option for nofp.
15313 (cortex-a9): Add fpu. Add options for nosimd and nofp.
15314 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
15315 (cortex-r4f): Add fpu.
15316 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
15317 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
15319 (cortex-r8): Likewise.
15320 (cortex-m4): Add fpu. Add option for nofp.
15321 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
15322 (cortex-a17.cortex-a7): Likewise.
15323 (cortex-a32): Add fpu. Add options for crypto and nofp.
15324 (cortex-a35, cortex-a53): Likewise.
15325 (cortex-a57): Add fpu. Add option for crypto.
15326 (cortex-a72, cortex-a73): Likewise.
15327 (exynos-m1): Likewise.
15328 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
15329 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
15330 (cortex-m33): Add fpu. Add option for nofp.
15331 * config/arm/arm-cpu-cdata.h: Regenerated
15332 * config/arm/arm-cpu-data.h: Regenerated.
15334 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15336 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
15337 (armv5te, armv5tej): Likewise.
15338 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
15339 (armv7): Add options fp and vfpv3-d16.
15340 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
15341 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
15343 (armv7ve): Likewise.
15344 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
15345 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
15346 (armv8-a): Add nocrypto option.
15347 (armv8.1-a, armv8.2-a): Likewise.
15348 (armv8-m.main): add options fp, fp.dp and nofp.
15350 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15352 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
15354 (armv8-a+crc): Delete.
15355 (armv8.1-a): Add options simd, crypto and nofp.
15356 (armv8.2-a): Add options fp16, simd, crypto and nofp.
15357 (armv8.2-a+fp16): Delete.
15358 (armv8-m.main): Add option dsp.
15359 (armv8-m.main+dsp): Delete.
15360 (cortex-a8): Add fpu. Add nofp option.
15361 (cortex-a9): Add fpu. Add nofp and nosimd options.
15362 * config/arm/parsecpu.awk (gen_data): Generate option tables and
15363 link to main cpu and architecture data structures.
15364 (gen_comm_data): Only put isa attributes from the main architecture
15366 (option): New statement for architecture and CPU entries.
15367 * arm.c (struct cpu_option): New structure.
15368 (struct processors): Add entry for options.
15369 (arm_unrecognized_feature): New function.
15370 (arm_parse_arch_cpu_name): Ignore any characters after the first
15372 (arm_parse_arch_cpu_feature): New function.
15373 (arm_configure_build_target): Separate out any CPU and architecture
15374 features and parse separately. Don't error out if -mfpu=auto is
15375 used with only an architecture string.
15376 (arm_print_asm_arch_directives): New function.
15377 (arm_file_start): Call it.
15378 * config/arm/arm-cpu-cdata.h: Regenerated.
15379 * config/arm/arm-cpu-data.h: Likewise.
15380 * config/arm/arm-tables.opt: Likewise.
15382 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15384 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
15385 assembler when it is not -mfpu=auto.
15387 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15389 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
15390 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
15391 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
15392 (ASM_CPU_SPEC): Rewrite.
15393 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
15394 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
15395 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
15396 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
15397 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
15398 copied string is NUL-terminated. Also strip any characters prefixed
15400 (arm_rewrite_selected_arch): New function.
15401 (arm_rewrite_march): New function.
15403 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15405 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
15406 (x_arm_cpu_string, x_arm_tune_string): Likewise.
15407 (march, mcpu, mtune): Convert to string-based options.
15408 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
15409 (arm_parse_arch_cpu_name): New function.
15410 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
15411 identify selected architecture or CPU.
15412 (arm_option_save): New function.
15413 (TARGET_OPTION_SAVE): Redefine.
15414 (arm_option_restore): Restore string options.
15415 (arm_option_print): Print string options.
15417 2017-06-16 Martin Sebor <msebor@redhat.com>
15419 PR tree-optimization/80933
15420 PR tree-optimization/80934
15421 * builtins.c (fold_builtin_3): Do not handle bcmp here.
15422 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
15423 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
15424 (gimple_fold_builtin): Call them.
15426 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15428 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
15429 as unlikely; update profile.
15431 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15433 * predict.c (force_edge_cold): Handle declaring edges impossible
15436 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15438 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
15440 (try_unroll_loop_completely): Fix reporting.
15442 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15444 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
15446 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
15449 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
15450 if given a non-constant argument for an intrinsic which requires a
15453 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15455 * profile.c (compare_freqs): New function.
15456 (branch_prob): Sort edge list.
15457 (find_spanning_tree): Assume that the list is priority sorted.
15459 2017-06-16 Richard Biener <rguenther@suse.de>
15461 PR tree-optimization/81090
15462 * passes.def (pass_record_bounds): Remove.
15463 * tree-pass.h (make_pass_record_bounds): Likewise.
15464 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
15465 make_pass_record_bounds): Likewise.
15466 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
15467 not free niter estimates at the beginning but at the end.
15468 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
15470 2017-06-16 Richard Biener <rguenther@suse.de>
15472 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
15473 initializer to workaround ICE in host GCC 4.8.
15475 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15477 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
15479 (clone_inlined_nodes): Update.
15481 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15483 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
15484 prefetch settings, and enable prefetching by default at -O3.
15486 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15488 * config/aarch64/aarch64.c (aarch64_override_options_internal):
15489 Set flag_prefetch_loop_arrays according to tuning data.
15491 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15493 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
15494 New tune structure.
15495 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
15496 [Unrelated to main purpose of the patch] Place the pointer field last
15497 to enable type checking errors when tune structure are wrongly merged.
15498 * config/aarch64/aarch64.c (generic_prefetch_tune,)
15499 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
15500 (thunderx2t99_prefetch_tune): New tune constants.
15501 (tune_params *_tunings): Update all tunings (no functional change).
15502 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
15503 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
15504 from tunings structures.
15506 2017-06-16 Jakub Jelinek <jakub@redhat.com>
15509 * ubsan.c (instrument_null): Add T argument, use it instead
15510 of computing it based on IS_LHS.
15511 (instrument_object_size): Likewise.
15512 (pass_ubsan::execute): Adjust instrument_null and
15513 instrument_object_size callers to pass gimple_get_lhs or
15514 gimple_assign_rhs1 result to it. Use instrument_null instead of
15515 calling get_base_address and instrument_mem_ref. Handle
15516 aggregate call arguments for object-size sanitization.
15518 2017-06-16 Yury Gribov <tetra2005@gmail.com>
15520 PR tree-optimization/81089
15521 * tree-vrp.c (is_masked_range_test): Validate operands of
15524 2017-06-15 Martin Sebor <msebor@redhat.com>
15527 * dumpfile.c (dump_register): Avoid calling memset to initialize
15528 a class with a default ctor.
15529 * gcc.c (struct compiler): Remove const qualification.
15530 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
15531 * hash-table.h: Ditto.
15532 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
15534 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
15535 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
15537 * params.h (struct param_info): Make struct members non-const.
15538 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
15539 with default initialization.
15540 * vec.h (vec_copy_construct, vec_default_construct): New helper
15542 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
15543 with vec_copy_construct.
15544 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
15545 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
15546 * doc/invoke.texi (-Wclass-memaccess): Document.
15548 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15550 * emit-rtl.h (is_leaf): Update comment about local
15551 register allocator.
15553 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
15556 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
15557 for a variable to have a section before checking if the section has a
15559 Set section to.persistent if persistent attribute is set.
15560 Warn if .persistent attribute is used on an automatic variable.
15562 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
15564 PR rtl-optimization/80474
15565 * reorg.c (update_block): Do not ignore instructions in a delay slot.
15567 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
15569 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
15572 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
15574 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
15575 (casesi): Emit bounds checking as RTL.
15576 (casesi_internal_mips16_<mode>): Remove bounds checking.
15578 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
15580 * config/xtensa/xtensa.c (xtensa_option_override): Append
15581 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
15582 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
15583 xtensa_doloop_hooks): Define unconditionally.
15584 (xtensa_reorg_loops): Only call reorg_loops in the presence of
15586 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
15587 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
15588 for it in xtensa_option_override.
15589 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
15590 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
15592 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
15594 * doc/cppopts.texi: Document '-' special value to -MF.
15596 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
15598 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
15599 (cortex_a53_fconst): Likewise.
15600 (cortex_a53_fpmul): Likewise.
15601 (cortex_a53_f_load_64): Likewise.
15602 (cortex_a53_f_load_many): Likewise.
15603 (cortex_a53_advsimd_alu): Likewise.
15604 (cortex_a53_advsimd_alu_q): Likewise.
15605 (cortex_a53_advsimd_mul): Likewise.
15606 (cortex_a53_advsimd_mul_q): Likewise.
15607 (fpmac bypass): Add new bypass for fpmac-fpmac case.
15608 Add missing fmul, r2f_cvt and fconst cases.
15610 2017-06-14 Richard Biener <rguenther@suse.de>
15612 PR middle-end/81088
15613 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
15615 (fold_binary_loc): When associating do not treat pre-existing
15616 TREE_OVERFLOW on literal constants as a reason to allow
15617 TREE_OVERFLOW on associated literal constants.
15619 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15621 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
15622 (MASK_FEATURES): New macro.
15623 * config/sparc/sparc.c (sparc_option_override): Remove the special
15624 handling of -mfpu and generalize it to all MASK_FEATURES switches.
15626 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15628 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
15629 a division of 0 if non-call exceptions are enabled.
15631 2017-06-14 Andrew Pinski <apinski@cavium.com>
15632 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15635 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
15636 Improve vector initialization code gen for only variable case.
15638 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
15640 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
15642 2017-06-14 Richard Biener <rguenther@suse.de>
15644 PR tree-optimization/81083
15645 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
15648 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15650 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
15651 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
15652 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
15653 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
15654 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
15655 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
15657 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15659 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
15660 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
15662 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15664 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
15666 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15668 * config/rs6000/t-rtems: Don't handle SPE.
15670 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15672 * config/rs6000/t-linux: Don't handle SPE.
15674 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15676 * config/rs6000/eabispe.h: Delete file.
15678 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15680 * config/rs6000/t-spe: Delete file.
15682 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15684 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
15685 (rs6000_legitimate_offset_address_p): Return false for anything in
15686 V2SImode or V2SFmode.
15688 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
15690 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
15691 except V2SF and V2SI. Rearrange the vector modes, and add comments.
15692 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
15694 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
15695 (rs6000_legitimate_offset_address_p): Ditto.
15696 (rs6000_emit_move): Ditto.
15697 (rs6000_init_builtins): Remove V4HI_type_node.
15699 2017-06-13 Martin Liska <mliska@suse.cz>
15702 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
15703 (gate_asan): Likewise.
15704 * asan.h (asan_no_sanitize_address_p): Remove the function.
15705 (sanitize_flags_p): New function.
15706 * builtins.def: Fix coding style.
15707 * common.opt: Use renamed enum value.
15708 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
15709 * doc/extend.texi: Document no_sanitize attribute.
15710 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
15711 to SANITIZE_UNDEFINED_NONDEFAULT.
15712 * gcc.c (sanitize_spec_function): Use the renamed enum value.
15713 * gimple-fold.c (optimize_atomic_compare_exchange_p):
15714 Use sanitize_flags_p.
15715 * gimplify.c (gimplify_function_tree): Likewise.
15716 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
15717 * opts.c (parse_no_sanitize_attribute): New function.
15718 (common_handle_option): Use renamed enum value.
15719 * opts.h (parse_no_sanitize_attribute): Declare.
15720 * tree.c (sanitize_flags_p): New function.
15721 * tree.h: Declared here.
15722 * tsan.c: Use sanitize_flags_p.
15723 * ubsan.c (ubsan_expand_null_ifn): Likewise.
15724 (instrument_mem_ref): Likewise.
15725 (instrument_bool_enum_load): Likewise.
15726 (do_ubsan_in_current_function): Remove the function.
15727 (pass_ubsan::execute): Use sanitize_flags_p.
15728 * ubsan.h: Remove do_ubsan_in_current_function
15729 * tree-cfg.c (print_no_sanitize_attr_value): New function.
15730 (dump_function_to_file): Use it here.
15732 2017-06-13 Martin Jambor <mjambor@suse.cz>
15734 PR tree-optimization/80803
15735 PR tree-optimization/81063
15736 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
15737 (propagate_subaccesses_across_link): Enqueue subtree whenever
15738 necessary instead of relying on the caller.
15740 2017-06-13 Martin Jambor <mjambor@suse.cz>
15742 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
15743 that have a first_link.
15744 (sort_and_splice_var_accesses): Do not check first_link before
15746 (subtree_mark_written_and_enqueue): Likewise.
15747 (propagate_all_subaccesses): Likewise and do not stop at first
15748 parent with a first_link.
15750 2017-06-13 Martin Jambor <mjambor@suse.cz>
15752 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
15755 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15757 * match.pd: New pattern.
15759 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15761 * tree-vrp.c (is_masked_range_test): New function.
15762 (register_edge_assert_for): Determine ranges for
15765 2017-06-13 Yury Gribov <tetra2005@gmail.com>
15767 PR tree-optimization/67328
15768 * fold-const.c (maskable_range_p): New function.
15769 (build_range_check): Generate bittests if possible.
15771 2017-06-13 Martin Liska <mliska@suse.cz>
15773 * gimple-pretty-print.c (dump_probability): Add new argument.
15774 (dump_edge_probability): Dump both probability and count.
15775 (dump_gimple_label): Likewise.
15776 (dump_gimple_bb_header): Likewise.
15778 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
15781 * config/avr/avr-devices.c: Fix indentation.
15782 * config/avr/gen-avr-mmcu-specs.c: Dito.
15784 2017-06-13 Richard Biener <rguenther@suse.de>
15786 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
15787 instead get vector type from stmt_info.
15788 (vectorizable_reduction): Adjust. Remove dead code.
15790 2017-06-13 Richard Biener <rguenther@suse.de>
15792 PR middle-end/81065
15793 * fold-const.c (extract_muldiv_1): Remove bogus distribution
15794 case of C * (x * C2 + C3).
15795 (fold_addr_of_array_ref_difference): Properly fold index difference.
15797 2017-06-12 David S. Miller <davem@davemloft.net>
15800 * config/sparc/sparc.md (return expander): Emit frame blockage if
15801 function uses alloca.
15803 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
15805 * combine.c (make_field_assignment): Check len rather than the mode
15806 precision when calling force_to_mode.
15808 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
15810 Support multilibs and devices that see flash in RAM address range.
15813 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
15814 (avr_mcu_t) <flash_pm_offset>: New field.
15815 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
15816 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
15817 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
15818 (AVR_TINY_PM_OFFSET): Remove macro.
15819 * config/avr/avr.opt (-mshort-calls): New option.
15820 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15821 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
15822 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
15823 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
15824 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
15825 instead of avr_arch->have_jmp_call.
15826 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
15827 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
15828 avr_arch->flash_pm_offset to define.
15829 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
15830 new field flash_pm_offset. Add entry for avrxmega3.
15831 (avr_texinfo): Add entry for avrxmega3.
15832 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
15833 attiny212, attiny214,
15834 attiny412, attiny414, attiny416, attiny417,
15835 attiny814, attiny816, attiny817,
15836 attiny1614, attiny1616, attiny1617,
15837 attiny3214, attiny3216, attiny3217.
15838 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
15839 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
15840 (avr_print_operand_address) [AVR_TINY]: Same.
15841 (avr_asm_init_sections) <readonly_data_section>: Only patch
15842 callback if avr_arch->flash_pm_offset = 0.
15843 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
15844 for rodata if avr_arch->flash_pm_offset != 0.
15845 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
15846 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
15847 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
15848 (m_options): Append opt_rcall.
15849 (m_dirnames): Append dir_rcall.
15850 * config/avr/t-multilib: Regenerate.
15852 * configure.ac [target=avr]: Check whether avrxmega3 default
15853 linker description file works as needed.
15854 * configure: Regenerate.
15855 * doc/avr-mmcu.texi: Regenerate.
15856 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
15857 <__AVR_ARCH__>: Document avrxmega3 and 103.
15858 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
15859 <__AVR_SHORT_CALLS__>: Document it.
15860 <__AVR_PM_BASE_ADDRESS__>: Document it.
15861 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
15862 (AVR Variable Attributes) <progmem>: Document this is
15863 not needed for avrxmega3.
15864 (AVR Named Address Spaces) <__flash>: Dito.
15866 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
15868 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
15870 2017-06-12 Doug Rupp <rupp@adacore.com>
15872 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
15873 Append vxworks-stdint.h to the tm_file list.
15874 * config/vxworks-stdint.h: New file.
15876 2017-06-12 Martin Liska <mliska@suse.cz>
15878 PR tree-optimization/81041
15879 * tree-profile.c (gimple_gen_ic_func_profiler):
15880 Create an extra BB in profile-generate
15881 (gimple_gen_time_profiler): Likewise.
15883 2017-06-12 Jakub Jelinek <jakub@redhat.com>
15885 PR tree-optimization/81003
15886 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
15887 (update_range_test): Use it instead of force_gimple_operand_gsi.
15889 2017-06-12 Richard Biener <rguenther@suse.de>
15891 PR tree-optimization/81053
15892 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
15893 with backedge value not defined in loop. Simplify def stmt
15896 2017-06-11 Tom de Vries <tom@codesourcery.com>
15899 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
15901 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
15902 nvptx_cannot_force_const_mem.
15904 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15906 * opts.c (finish_options): Move test for flag_split_stack after
15907 it has been initialized.
15909 2017-06-11 Jason Merrill <jason@redhat.com>
15911 * tree.h (id_equal): New.
15912 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
15913 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
15914 instead of strcmp of IDENTIFIER_POINTER.
15916 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15918 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
15919 (mark_all_inlined_calls_cdtor): Fix formating.
15920 (inline_transform): Rescale profile before inlining.
15922 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15924 * cgraph.h (cgraph_edge::clone): Update prototype.
15925 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
15926 (cgraph_node::create_clone): Update.
15927 (cgraph_node::create_version_clone): Update.
15928 * tree-inline.c (copy_bb): Update.
15929 (expand_call_inline): Update.
15931 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15933 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
15934 factored out from ...
15935 (rs6000_emit_prologue): ... here.
15937 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
15939 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
15940 factored out from ...
15941 (rs6000_emit_prologue): ... here.
15943 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15945 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
15947 (handle_missing_profiles): Fix computation of tp_first_run.
15948 (counts_to_freqs): Do not touch freqs when count is 0.
15950 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
15952 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
15955 2017-06-10 Tom de Vries <tom@codesourcery.com>
15957 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
15958 attributes): Document signal effective target.
15960 2017-06-10 Tom de Vries <tom@codesourcery.com>
15962 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
15963 Document effective target stack_size.
15965 2017-06-09 David Malcolm <dmalcolm@redhat.com>
15967 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
15968 to the edit_context if they can be auto-applied.
15970 2017-06-9 Ian Lance Taylor <iant@golang.org>
15972 * opts.c (finish_options): If -fsplit-stack, disable implicit
15973 -forder-blocks-and-partition.
15974 * doc/invoke.texi (Optimize Options): Document that when using
15975 -fsplit-stack -forder-blocks-and-partition is not implicitly
15978 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15980 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
15981 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
15982 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
15983 * builtins.def (abort, trap, unreachable): Declare cold.
15984 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
15985 * tree-core.h (ECF_COLD): New.
15986 * tree.c (set_call_expr_flags): Handle ECF_COLD.
15987 (build_common_builtin_nodes): Mark unreachable and abort as cold.
15989 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
15991 * predict.c (unlikely_executed_stmt_p): Cleanup.
15993 2017-06-09 Richard Biener <rguenther@suse.de>
15995 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
15996 model if the ref is always written to.
15998 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16000 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
16002 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16004 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
16007 2017-06-09 Tom de Vries <tom@codesourcery.com>
16010 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
16011 "target cannot support label values" when encountering LABEL_REF.
16013 2017-06-09 Martin Liska <mliska@suse.cz>
16015 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
16016 (gimple_gen_ic_func_profiler): Emit direct comparison
16017 of __gcov_indirect_call_callee with NULL.
16018 (gimple_gen_time_profiler): Change probability from
16019 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
16021 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16023 * profile.c (edge_gcov_counts): Turn to pointer.
16024 (compute_branch_probabilities, compute_branch_probabilities): Update.
16025 (branch_prob): Do not clear edge_gcov_count.
16026 * profile.h (edge_gcov_counts): Turn to pointer.
16027 (edge_gcov_count): Update.
16029 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16031 * gimple.h (gimple_check_failed): Mark cold.
16033 2017-06-09 Richard Biener <rguenther@suse.de>
16035 PR tree-optimization/66623
16036 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
16037 refactor check_reduction into two parts, properly computing
16038 whether we have to check reduction validity for outer loop
16041 2017-06-09 Richard Biener <rguenther@suse.de>
16043 PR tree-optimization/79483
16044 * graphite-scop-detection.c (order): New global.
16045 (get_order): Compute bb to order mapping that satisfies code
16046 generation constraints.
16047 (cmp_pbbs): New helper.
16048 (build_scops): Start domwalk at entry block, sort generated
16051 2017-06-09 Richard Biener <rguenther@suse.de>
16053 PR middle-end/81007
16054 * ipa-polymorphic-call.c
16055 (ipa_polymorphic_call_context::restrict_to_inner_class):
16056 Skip FIELD_DECLs with error_mark_node type.
16057 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
16060 2017-06-09 Martin Liska <mliska@suse.cz>
16062 * predict.c (struct branch_predictor): New struct.
16063 (test_prediction_value_range): New test.
16064 (predict_c_tests): New function.
16065 * selftest-run-tests.c (selftest::run_tests): Run the function.
16066 * selftest.h: Declare new tests.
16068 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
16071 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
16072 gen_add3_insn did not fail.
16073 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
16074 r0, construct that number in a temporary reg and add that reg to r0.
16075 If asked to put the result in r0 as well, fail.
16077 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
16079 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16080 for early expansion of vec_eqv.
16082 2017-06-08 Jakub Jelinek <jakub@redhat.com>
16084 PR middle-end/81005
16085 * ubsan.c (instrument_null): Avoid pointless code temporary.
16086 (pass_ubsan::execute): Instrument aggregate arguments of calls.
16088 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
16092 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
16095 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
16098 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16100 * predict.c (unlikely_executed_edge_p): Move ahead.
16101 (probably_never_executed_edge_p): Use it.
16103 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16105 PR middle-end/79988
16106 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
16107 gimple_call_builtin_p call.
16109 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16111 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
16112 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
16113 rtl_check_failed_type2, rtl_check_failed_code1,
16114 rtl_check_failed_code2, rtl_check_failed_code_mode,
16115 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
16116 rtvec_check_failed_bounds, rtl_check_failed_flag,
16117 _fatal_insn_not_found, _fatal_insn): Likewise.
16118 * tree.h (tree_contains_struct_check_failed,
16119 tree_check_failed, tree_not_check_failed,
16120 tree_class_check_failed, tree_range_check_failed,
16121 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
16122 tree_vec_elt_check_failed, phi_node_elt_check_failed,
16123 tree_operand_check_failed, omp_clause_check_failed,
16124 omp_clause_operand_check_failed, omp_clause_range_check_failed):
16127 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16129 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
16130 flag_branch_probabilities.
16131 * ipa-inline.c (edge_badness): Likewise.
16132 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
16133 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
16134 * predict.c (maybe_hot_frequency_p): Likewise.
16135 (probably_never_executed): Likewise.
16136 * sched-ebb.c (schedule_ebbs): Likewise.
16137 * sched-rgn.c (find_single_block_region): Likewise.
16138 * tracer.c (tail_duplicate): Likewise.
16140 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16142 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
16143 longer requires x_flag_profile_use.
16145 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16147 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
16148 instead of flag_reorder_blocks_and_partition.
16149 * dbxout.c (dbxout_function_end): Likewise.
16150 * dwarf2out.c (gen_subprogram_die): Likewise.
16151 * haifa-sched.c (sched_create_recovery_edges): Likewise.
16152 * hw-doloop.c (reorg_loops): Likewise.
16153 * varasm.c (assemble_start_function,
16154 assemble_end_function): Likewise.
16155 (decide_function_section): Do not check for
16156 flag_reorder_blocks_and_partition.
16158 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16160 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16162 (chkp_get_hard_register_fake_addr_expr): Ditto.
16163 (chkp_build_addr_expr): Add check for hard reg case.
16164 (chkp_parse_array_and_component_ref): Ditto.
16165 (chkp_find_bounds_1): Ditto.
16166 (chkp_process_stmt): Don't generate bounds store for
16169 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16171 * predict.c (maybe_hot_bb_p): Do not check profile status.
16172 (maybe_hot_edge_p): Likewise.
16173 (probably_never_executed): Check for zero counts even if profile
16175 (unlikely_executed_edge_p): New function.
16176 (unlikely_executed_stmt_p): New function.
16177 (unlikely_executed_bb_p): New function.
16178 (set_even_probabilities): Use unlikely predicates.
16179 (combine_predictions_for_bb): Likewise.
16180 (predict_paths_for_bb): Likewise.
16181 (predict_paths_leading_to_edge): Likewise.
16182 (determine_unlikely_bbs): New function.
16183 (estimate_bb_frequencies): Use it.
16184 (compute_function_frequency): Use zero counts even if profile is
16186 * profile-count.h: Fix typo.
16188 2017-08-08 Julia Koval <julia.koval@intel.com>
16190 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
16191 _mm512_mask_cvtsepi16_storeu_epi8,
16192 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
16193 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
16194 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
16195 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
16196 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
16197 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
16198 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
16199 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
16200 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
16201 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
16202 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
16203 __builtin_ia32_pmovuswb256mem_mask,
16204 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
16205 __builtin_ia32_pmovwb512mem_mask): New builtins.
16207 2017-08-08 Julia Koval <julia.koval@intel.com>
16209 PR target/73350,80862
16210 * config/i386/subst.md (round): Fix round pattern.
16211 * config/i386/i386.c (ix86_erase_embedded_rounding):
16212 Fix erasing rounding for the fixed pattern.
16214 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16216 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
16218 2017-06-08 Martin Liska <mliska@suse.cz>
16220 PR gcov-profile/80911
16221 * gcov.c (block_info::block_info): New constructor.
16223 2017-06-07 Carl Love <cel@us.ibm.com>
16225 * config/rs6000/rs6000-c: The return type of the following
16226 built-in functions was implemented as int not long long. Fix sign
16227 of return value for the unsigned version of vec_mulo and vec_mule.
16228 vector unsigned long long vec_bperm (vector unsigned long long,
16229 vector unsigned char)
16230 vector signed long long vec_mule (vector signed int,
16232 vector unsigned long long vec_mule (vector unsigned int,
16233 vector unsigned int)
16234 vector signed long long vec_mulo (vector signed int,
16236 vector unsigned long long vec_mulo (vector unsigned int,
16237 vector unsigned int)
16238 * doc/extend.texi: Fix the documentation for the built-in
16241 2017-06-07 Carl Love <cel@us.ibm.com>
16244 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
16247 2017-06-07 Carl Love <cel@us.ibm.com>
16249 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
16250 support, Generate doublehv for signed int/float for BE case only.
16252 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
16254 * doc/invoke.texi (mcx16): Rewrite.
16256 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16258 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
16259 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
16260 *mov<mode>_softfloat, and an anonymous splitter): Use
16261 nonimmediate_operand instead of rs6000_nonimmediate_operand.
16263 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16265 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
16267 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
16268 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
16269 (rs6000_debug_reg_global): Adjust.
16270 (rs6000_init_hard_regno_mode_ok): Adjust.
16271 (rs6000_dbx_register_number): Adjust.
16272 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
16273 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
16274 Remove SPE_ACC and SPEFSCR.
16275 (REG_ALLOC_ORDER): Ditto.
16276 (FRAME_POINTER_REGNUM): Change to 111.
16277 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
16278 (REG_CLASS_NAMES): Ditto.
16279 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
16280 (REGISTER_NAMES): Ditto.
16281 (ADDITIONAL_REG_NAMES): Ditto.
16282 (rs6000_reg_names): Ditto.
16283 * config/rs6000/rs6000.md: Renumber some register number
16286 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16288 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
16290 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
16291 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
16293 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
16294 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
16295 Delete the SPE high registers.
16296 (REG_ALLOC_ORDER): Ditto.
16297 (enum reg_class): Remove SPE_HIGH_REGS.
16298 (REG_CLASS_NAMES): Ditto.
16299 (REG_CLASS_CONTENTS): Delete the SPE high registers.
16300 (REGISTER_NAMES): Ditto.
16301 (rs6000_reg_names): Ditto.
16302 * doc/tm.texi.in: Remove SPE as example.
16303 * doc/tm.texi: Regenerate.
16305 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16307 * config/rs6000/8540.md (ppc8540_brinc): Delete.
16308 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
16309 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
16310 * config/rs6000/rs6000.md (type): Remove "brinc".
16312 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16314 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
16315 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
16316 * config/rs6000/linuxspe.h: Delete file.
16317 * config/rs6000/rs6000.md: Don't include spe.md.
16318 * config/rs6000/spe.h: Delete file.
16319 * config/rs6000/spe.md: Delete file.
16320 * config/rs6000/t-rs6000: Remove spe.md.
16322 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16324 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
16325 (reg_or_none500mem_operand): Delete.
16326 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
16327 instead of reg_or_none500mem_operand.
16329 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16331 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16332 handling of SPE flags.
16333 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
16335 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16337 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
16339 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
16340 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
16341 paired_divv2sf3): Similar.
16342 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
16343 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
16344 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
16346 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
16347 Rename the paired_* instruction patterns.
16348 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
16350 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
16351 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
16352 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
16353 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
16354 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
16355 PAIRED_VECTOR_MODE.
16356 (struct machine_function): Delete field spe_insn_chain_scanned_p.
16357 (spe_func_has_64bit_regs_p): Delete.
16358 (spe_expand_predicate_builtin): Delete.
16359 (spe_expand_evsel_builtin): Delete.
16360 (TARGET_DWARF_REGISTER_SPAN): Do not define.
16361 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
16362 (invalid_e500_subreg): Delete.
16363 (rs6000_legitimize_address): Always force_reg op2 as well, for
16364 paired single memory accesses.
16365 (rs6000_member_type_forces_blk): Delete.
16366 (rs6000_spe_function_arg): Delete.
16367 (rs6000_expand_unop_builtin): Delete SPE handling.
16368 (rs6000_expand_binop_builtin): Ditto.
16369 (spe_expand_stv_builtin): Delete.
16370 (bdesc_2arg_spe): Delete.
16371 (spe_expand_builtin): Delete.
16372 (spe_expand_predicate_builtin): Delete.
16373 (spe_expand_evsel_builtin): Delete.
16374 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
16375 (spe_init_builtins): Delete.
16376 (spe_func_has_64bit_regs_p): Delete.
16377 (savres_routine_name): Delete "info" parameter. Adjust callers.
16378 (rs6000_emit_stack_reset): Ditto.
16379 (rs6000_dwarf_register_span): Delete.
16380 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
16381 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
16382 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
16384 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
16386 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
16387 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
16388 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
16389 mulv2sf3, divv2sf3): Delete expanders.
16391 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16393 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
16395 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16397 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
16398 * config/rs6000/rs6000.c: Ditto.
16400 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16402 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
16403 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
16404 comparison_operator.
16406 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16408 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
16409 * config/rs6000/rs6000.opt: Ditto.
16410 * config/rs6000/t-rtems: Ditto.
16412 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16414 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
16415 TARGET_E500_SINGLE by 0, simplify.
16416 * config/rs6000/rs6000.c: Ditto.
16417 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
16418 (spe_build_register_parallel): Delete.
16419 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
16420 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
16421 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
16422 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
16423 (E500_CONVERT): Delete.
16424 * config/rs6000/spe.md: Remove many patterns and all define_constants.
16426 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16428 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
16429 * config/rs6000/dfp.md: Ditto.
16430 (negdd2, *negdd2_fpr): Merge.
16431 (absdd2, *absdd2_fpr): Merge.
16432 (negtd2, *negtd2_fpr): Merge.
16433 (abstd2, *abstd2_fpr): Merge.
16434 * config/rs6000/e500.h: Delete file.
16435 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
16436 TARGET_FPRS by 1 and simplify.
16437 * config/rs6000/rs6000-c.c: Ditto.
16438 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
16439 TARGET_DF_SPE by 0.
16440 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
16442 * config/rs6000/rs6000.md: Ditto.
16443 (floatdidf2, *floatdidf2_fpr): Merge.
16444 (move_from_CR_gt_bit): Delete.
16445 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
16446 (E500_CR_IOR_COMPARE): Delete.
16447 (All patterns that require !TARGET_FPRS): Delete.
16448 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
16450 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16452 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
16454 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16456 * graphds.c (add_edge): Intitialize edge's attached data.
16457 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
16458 pointer parameter. Call pointed function on each edge during
16459 graph traversing. Skip traversing the edge when the function
16461 (graphds_dfs, graphds_scc): Ditto.
16462 (for_each_edge): New parameter. Pass the new parameter to callback
16464 * graphds.h (skip_edge_callback): New function pointer type.
16465 (graphds_dfs, graphds_scc): New function pointer parameter.
16466 (graphds_edge_callback, for_each_edge): New parameter.
16468 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16470 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
16471 out code checking if runtime alias check is possible to below ...
16472 Call the new function.
16473 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
16474 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
16476 2017-06-07 Marek Polacek <polacek@redhat.com>
16479 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
16480 TYPE_OVERFLOW_WRAPS check.
16482 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16484 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
16485 if versioning is required.
16486 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
16487 peeling with the check for versioning.
16489 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16491 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
16492 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
16493 Set true to new parameter if new ssa variable is defined.
16494 (vect_gen_vector_loop_niters): Refactor. Set range information
16495 for the new vector loop bound variable.
16496 (vect_do_peeling): Ditto.
16498 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16500 * tree-affine.c (ssa.h): Include header file.
16501 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
16502 has wrapping overflow behavior.
16504 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16506 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
16508 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16510 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
16511 (tree_to_aff_combination): ... here.
16513 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16515 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
16516 reg_pressure model function.
16517 (ivopts_global_cost_for_size): Delete.
16518 (determine_set_costs, iv_ca_recount_cost): Call new model function
16519 ivopts_estimate_reg_pressure.
16521 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16523 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
16524 expensive than udiv. Remove floating point cases from mod.
16526 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16528 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
16529 Increase idiv cost.
16531 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16533 * config/aarch64/aarch64.md
16534 (copysignsf3): Fix mask generation.
16536 2017-06-07 Jakub Jelinek <jakub@redhat.com>
16538 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
16540 (class dump_manager): Add register_dumps method.
16541 * dumpfile.c: Include langhooks.h.
16542 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
16543 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
16544 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
16545 (dump_manager::dump_register): Start with 512 entries instead of 32.
16546 (dump_manager::register_dumps): New method.
16547 * toplev.c (general_init): Instead of invoking register_dumps
16548 langhook, invoke register_dumps method on the dump manager.
16549 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
16552 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
16554 * doc/md.texi: Clarify the restrictions on a define_insn condition.
16555 Say that # requires an associated define_split to exist, and that
16556 the define_split must be suitable for use after register allocation.
16558 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16560 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
16561 (compute_outgoing_frequencies): Also initialize zero counts.
16562 (find_many_sub_basic_blocks): Do not produce uninitialized profile
16563 around loops; preserve more of profile when nothing changes.
16565 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
16567 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
16569 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
16570 * config/arm/arm-cpu-cdata.h: Regenerate.
16571 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
16572 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
16573 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
16574 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
16575 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
16577 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
16579 * config/arm/t-rmprofile: Likewise.
16580 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
16582 2017-06-06 David S. Miller <davem@davemloft.net>
16585 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
16586 blockage if function uses alloca.
16588 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16590 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
16591 New "uid" fields to hold pretty-print IDs of group and ref.
16592 Memory references are now identified as <group_id>:<ref_id>
16593 instead of using [random] addresses.
16594 (dump_mem_details): Simplify, no functional change.
16595 (dump_mem_ref): Simplify and make output more concise.
16596 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
16597 (find_or_create_group): Initialize group uid.
16598 (record_ref): Initialize ref uid. Improve debug output.
16599 (prune_group_by_reuse, should_issue_prefetch_p,)
16600 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
16601 (mark_nontemporal_store, determine_loop_nest_reuse):
16602 Improve debug output.
16604 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16606 * dbgcnt.def (prefetch): New debug counter.
16607 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
16608 (schedule_prefetches): Stop issueing prefetches if debug counter
16611 2017-06-06 Tom de Vries <tom@codesourcery.com>
16613 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
16614 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
16616 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16618 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
16619 Use aarch64_reg_or_zero predicate for operand 4.
16620 (aarch64_compare_and_swap<mode> define_insn_and_split):
16621 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
16622 (aarch64_store_exclusive<mode>): Likewise for operand 2.
16624 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
16626 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
16627 (arm_compute_save_core_reg_mask): This.
16628 (thumb1_compute_save_reg_mask): Rename into ...
16629 (thumb1_compute_save_core_reg_mask): This.
16630 (arm_compute_save_reg0_reg12_mask): Adapt comment.
16631 (arm_compute_frame_layout): Likewise.
16633 2017-06-06 Richard Biener <rguenther@suse.de>
16635 PR tree-optimization/80974
16636 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
16637 keep or clear leaders SSA info.
16639 2017-06-06 Tom de Vries <tom@codesourcery.com>
16641 * config/nvptx/nvptx.c (split_mode_p): New function.
16642 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
16644 2017-06-06 Tom de Vries <tom@codesourcery.com>
16646 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
16648 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16651 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
16654 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16656 * shrink-wrap.c (handle_simple_exit): Update profile.
16657 (try_shrink_wrapping): Upate profile.
16659 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16661 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
16662 (tree_guess_outgoing_edge_probabilities): New.
16663 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
16664 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
16666 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16668 * ipa-split.c (split_function): Initialize return bb profile.
16670 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
16672 * profile.c (compute_branch_probabilities): Also initialize
16673 EXIT_BLOCK profile.
16675 2017-06-06 Richard Biener <rguenther@suse.de>
16677 PR tree-optimization/80928
16678 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
16679 (vect_analyze_loop_operations): Properly guard analysis for
16681 (vect_transform_loop): Likewise.
16682 (vect_analyze_loop_2): Also reset SLP type on PHIs.
16683 (vect_model_induction_cost): Do not cost for pure SLP.
16684 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
16685 of induction in inner loop vectorization.
16686 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
16687 (vect_get_and_check_slp_defs): Handle vect_induction_def.
16688 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
16690 (vect_analyze_slp_cost_1): Cost induction.
16691 (vect_detect_hybrid_slp_stmts): Handle PHIs.
16692 (vect_get_slp_vect_defs): Likewise.
16693 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
16694 (vect_transform_stmt): Handle SLP reductions.
16695 * tree-vectorizer.h (vectorizable_induction): Adjust.
16697 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16699 * config/rs6000/rs6000.c (make_resolver_func): Update
16700 init_lowered_empty_function call.
16702 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
16704 * doc/invoke.texi: Document the -fprofile-abs-path option.
16705 * common.opt (fprofile-abs-path): New option.
16706 * gcov-io.h (gcov_write_filename): Declare.
16707 * gcov-io.c (gcov_write_filename): New function.
16708 * coverage.c (coverage_begin_function): Use gcov_write_filename.
16709 * profile.c (output_location): Likewise.
16711 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16713 * shring-wrap.c: Revert accidental commit.
16715 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
16717 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
16719 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16721 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
16723 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
16724 profile in callgraph edge.
16725 * profile-count.h (apply_probability): If THIS is 0, then result is 0
16726 (apply_scale): Likewise.
16727 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
16728 Also scale profile when inlining function with zero profile.
16729 (initialize_cfun): Update exit block profile even when it is zero.
16730 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
16731 when profile is read.
16733 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
16735 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
16736 (CLONE_*): New constants to define the processors we can generate
16737 code for with the target_clone attribute.
16738 (rs6000_clone_map): New array to identify which clone processors
16739 the current program is running on.
16740 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
16741 target_clone attribute.
16742 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
16743 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
16744 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
16745 (cpu_expand_builtin): Add support for target_clone attribute.
16746 (rs6000_valid_attribute_p): Allow "default" attribute.
16747 (get_decl_name): New debug function to simplify printing the
16748 current function name in debugging statements.
16749 (rs6000_clone_priority): New functions to support the target_clone
16750 attribute, and be able to generate code to switch between ISA 2.05
16751 through ISA 3.0 (power6 through power9).
16752 (rs6000_compare_version_priority): Likewise.
16753 (rs6000_get_function_versions_dispatcher): Likewise.
16754 (make_resolver_func): Likewise.
16755 (add_condition_to_bb): Likewise.
16756 (dispatch_function_versions): Likewise.
16757 (rs6000_generate_version_dispatcher_body): Likewise.
16758 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
16759 (fusion_gpr_load_p): Fix a spacing issue.
16760 * doc/extend.texi (Common Function Attributes): Document that the
16761 PowerPC supports the target_clone attribute.
16763 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
16765 * config/arm/arm.h: explain F symbol found in description of ARM
16766 register allocation in its legend.
16768 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16770 * config/mips/frame-header-opt.c: Include profile-count.h.
16771 * config/riscv/riscv.c: Include profile-count.h
16773 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
16775 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
16777 (sm_set_flag_if_changed): Add bbs field.
16778 (execute_sm_if_changed_flag_set): Pass BBS.
16779 (execute_sm): Update.
16781 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16783 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
16786 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16788 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
16789 (peephole2): New peephole2 to emit the above.
16790 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
16792 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16794 * config/aarch64/aarch64.c (define_peephole2 above
16795 *sub_<shift>_<mode>): New peephole.
16797 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
16799 * config/i386/i386.c (make_resolver_func): Update.
16800 * Makefile.in: Add profile-count.h and profile-count.o
16801 * auto-profile.c (afdo_indirect_call): Update to new API.
16802 (afdo_set_bb_count): Update.
16803 (afdo_propagate_edge): Update.
16804 (afdo_propagate_circuit): Update.
16805 (afdo_calculate_branch_prob): Update.
16806 (afdo_annotate_cfg): Update.
16807 * basic-block.h: Include profile-count.h
16808 (struct edge_def): Turn count to profile_count.
16809 (struct basic_block_def): Likewie.
16810 (REG_BR_PROB_BASE): Move to profile-count.h
16811 (RDIV): Move to profile-count.h
16812 * bb-reorder.c (max_entry_count): Turn to profile_count.
16813 (find_traces): Update.
16814 (rotate_loop):Update.
16815 (connect_traces):Update.
16816 (sanitize_hot_paths):Update.
16817 * bt-load.c (migrate_btr_defs): Update.
16818 * cfg.c (RDIV): Remove.
16819 (init_flow): Use alloc_block.
16820 (alloc_block): Uninitialize count.
16821 (unchecked_make_edge): Uninitialize count.
16822 (check_bb_profile): Update.
16823 (dump_edge_info): Update.
16824 (dump_bb_info): Update.
16825 (update_bb_profile_for_threading): Update.
16826 (scale_bbs_frequencies_int): Update.
16827 (scale_bbs_frequencies_gcov_type): Update.
16828 (scale_bbs_frequencies_profile_count): New.
16829 * cfg.h (update_bb_profile_for_threading): Update.
16830 (scale_bbs_frequencies_profile_count): Declare.
16831 * cfgbuild.c (compute_outgoing_frequencies): Update.
16832 (find_many_sub_basic_blocks): Update.
16833 * cfgcleanup.c (try_forward_edges): Update.
16834 (try_crossjump_to_edge): Update.
16835 * cfgexpand.c (expand_gimple_tailcall): Update.
16836 (construct_exit_block): Update.
16837 * cfghooks.c (verify_flow_info): Update.
16838 (dump_bb_for_graph): Update.
16839 (split_edge): Update.
16840 (make_forwarder_block): Update.
16841 (duplicate_block): Update.
16842 (account_profile_record): Update.
16843 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
16844 (get_estimated_loop_iterations): Update.
16845 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
16846 (single_likely_exit): Update.
16847 * cfgloopmanip.c (scale_loop_profile): Update.
16849 (set_zero_probability): Update.
16850 (lv_adjust_loop_entry_edge): Update.
16851 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16852 (purge_dead_edges): Update.
16853 (rtl_account_profile_record): Update.
16854 * cgraph.c (cgraph_node::create): Uninitialize count.
16855 (symbol_table::create_edge): Uninitialize count.
16856 (cgraph_update_edges_for_call_stmt_node): Update.
16857 (cgraph_edge::dump_edge_flags): Update.
16858 (cgraph_node::dump): Update.
16859 (cgraph_edge::maybe_hot_p): Update.
16860 * cgraph.h: Include profile-count.h
16861 (create_clone), create_edge, create_indirect_edge): Update.
16862 (cgraph_node): Turn count to profile_count.
16863 (cgraph_edge0: Likewise.
16864 (make_speculative, clone): Update.
16865 (create_edge): Update.
16866 (init_lowered_empty_function): Update.
16867 * cgraphclones.c (cgraph_edge::clone): Update.
16868 (duplicate_thunk_for_node): Update.
16869 (cgraph_node::create_clone): Update.
16870 * cgraphunit.c (cgraph_node::analyze): Update.
16871 (cgraph_node::expand_thunk): Update.
16872 * final.c (dump_basic_block_info): Update.
16873 * gimple-streamer-in.c (input_bb): Update.
16874 * gimple-streamer-out.c (output_bb): Update.
16875 * graphite.c (print_global_statistics): Update.
16876 (print_graphite_scop_statistics): Update.
16877 * hsa-brig.c: Include basic-block.h.
16878 * hsa-dump.c: Include basic-block.h.
16879 * hsa-gen.c (T sum_slice): Update.
16880 (convert_switch_statements):Update.
16881 * hsa-regalloc.c: Include basic-block.h.
16882 * ipa-chkp.c (chkp_produce_thunks): Update.
16883 * ipa-cp.c (struct caller_statistics): Update.
16884 (init_caller_stats): Update.
16885 (gather_caller_stats): Update.
16886 (ipcp_cloning_candidate_p): Update.
16887 (good_cloning_opportunity_p): Update.
16888 (get_info_about_necessary_edges): Update.
16889 (dump_profile_updates): Update.
16890 (update_profiling_info): Update.
16891 (update_specialized_profile): Update.
16892 (perhaps_add_new_callers): Update.
16893 (decide_about_value): Update.
16894 (ipa_cp_c_finalize): Update.
16895 * ipa-devirt.c (struct odr_type_warn_count): Update.
16896 (struct decl_warn_count): Update.
16897 (struct final_warning_record): Update.
16898 (possible_polymorphic_call_targets): Update.
16899 (ipa_devirt): Update.
16900 * ipa-fnsummary.c (redirect_to_unreachable): Update.
16901 * ipa-icf.c (sem_function::merge): Update.
16902 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
16903 * ipa-inline.c (compute_uninlined_call_time): Update.
16904 (compute_inlined_call_time): Update.
16905 (want_inline_small_function_p): Update.
16906 (want_inline_self_recursive_call_p): Update.
16907 (edge_badness): Update.
16908 (lookup_recursive_calls): Update.
16909 (recursive_inlining): Update.
16910 (inline_small_functions): Update.
16911 (dump_overall_stats): Update.
16912 (dump_inline_stats): Update.
16913 * ipa-profile.c (ipa_profile_generate_summary): Update.
16914 (ipa_propagate_frequency): Update.
16915 (ipa_profile): Update.
16916 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
16917 * ipa-utils.c (ipa_merge_profiles): Update.
16918 * loop-doloop.c (doloop_modify): Update.
16919 * loop-unroll.c (report_unroll): Update.
16920 (unroll_loop_runtime_iterations): Update.
16921 * lto-cgraph.c (lto_output_edge): Update.
16922 (lto_output_node): Update.
16923 (input_node): Update.
16924 (input_edge): Update.
16925 (merge_profile_summaries): Update.
16926 * lto-streamer-in.c (input_cfg): Update.
16927 * lto-streamer-out.c (output_cfg): Update.
16928 * mcf.c (create_fixup_graph): Update.
16929 (adjust_cfg_counts): Update.
16930 (sum_edge_counts): Update.
16931 * modulo-sched.c (sms_schedule): Update.
16932 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
16933 * predict.c (maybe_hot_count_p): Update.
16934 (probably_never_executed): Update.
16935 (dump_prediction): Update.
16936 (combine_predictions_for_bb): Update.
16937 (propagate_freq): Update.
16938 (handle_missing_profiles): Update.
16939 (counts_to_freqs): Update.
16940 (rebuild_frequencies): Update.
16941 (force_edge_cold): Update.
16942 * predict.h: Include profile-count.h
16943 (maybe_hot_count_p, counts_to_freqs): UPdate.
16944 * print-rtl-function.c: Do not include cfg.h
16945 * print-rtl.c: Include basic-block.h
16946 * profile-count.c: New file.
16947 * profile-count.h: New file.
16948 * profile.c (is_edge_inconsistent): Update.
16949 (correct_negative_edge_counts): Update.
16950 (is_inconsistent): Update.
16951 (set_bb_counts): Update.
16952 (read_profile_edge_counts): Update.
16953 (compute_frequency_overlap): Update.
16954 (compute_branch_probabilities): Update; Initialize and deinitialize
16956 (branch_prob): Update.
16957 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
16958 (edge_gcov_count): New.
16959 (bb_gcov_count): New.
16960 * shrink-wrap.c (try_shrink_wrapping): Update.
16961 * tracer.c (better_p): Update.
16962 * trans-mem.c (expand_transaction): Update.
16963 (ipa_tm_insert_irr_call): Update.
16964 (ipa_tm_insert_gettmclone_call): Update.
16965 * tree-call-cdce.c: Update.
16966 * tree-cfg.c (gimple_duplicate_sese_region): Update.
16967 (gimple_duplicate_sese_tail): Update.
16968 (gimple_account_profile_record): Update.
16969 (execute_fixup_cfg): Update.
16970 * tree-inline.c (copy_bb): Update.
16971 (copy_edges_for_bb): Update.
16972 (initialize_cfun): Update.
16973 (freqs_to_counts): Update.
16974 (copy_cfg_body): Update.
16975 (expand_call_inline): Update.
16976 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
16977 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
16978 (try_unroll_loop_completely): Update.
16979 (try_peel_loop): Update.
16980 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16981 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
16982 * tree-ssa-loop-split.c (connect_loops): Update.
16983 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
16984 * tree-ssa-reassoc.c (branch_fixup): Update.
16985 * tree-ssa-tail-merge.c (replace_block_by): Update.
16986 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
16987 (compute_path_counts): Update.
16988 (update_profile): Update.
16989 (recompute_probabilities): Update.
16990 (update_joiner_offpath_counts): Update.
16991 (estimated_freqs_path): Update.
16992 (freqs_to_counts_path): Update.
16993 (clear_counts_path): Update.
16994 (ssa_fix_duplicate_block_edges): Update.
16995 (duplicate_thread_path): Update.
16996 * tree-switch-conversion.c (case_bit_test_cmp): Update.
16997 (struct switch_conv_info): Update.
16998 * tree-tailcall.c (decrease_profile): Update.
16999 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
17000 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17001 * value-prof.c (check_counter): Update.
17002 (gimple_divmod_fixed_value): Update.
17003 (gimple_mod_pow2): Update.
17004 (gimple_mod_subtract): Update.
17005 (gimple_ic_transform): Update.
17006 (gimple_stringop_fixed_value): Update.
17007 * value-prof.h (gimple_ic): Update.
17009 2017-06-02 Carl Love <cel@us.ibm.com>
17011 * config/rs6000/rs6000-c: Add support for built-in functions
17012 vector double vec_doublee (vector signed int);
17013 vector double vec_doublee (vector unsigned int);
17014 vector double vec_doublee (vector float);
17015 vector double vec_doubleh (vector signed int);
17016 vector double vec_doubleh (vector unsigned int);
17017 vector double vec_doubleh (vector float);
17018 vector double vec_doublel (vector signed int);
17019 vector double vec_doublel (vector unsigned int);
17020 vector double vec_doublel (vector float);
17021 vector double vec_doubleo (vector signed int);
17022 vector double vec_doubleo (vector unsigned int);
17023 vector double vec_doubleo (vector float);.
17024 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
17025 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
17027 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
17028 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
17029 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
17031 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
17032 vec_doublel, vec_doubleh.
17033 * doc/extend.texi: Update the built-in documentation file for the
17034 new built-in functions.
17036 2017-06-02 David Malcolm <dmalcolm@redhat.com>
17039 * ipa-inline-analysis.c (free_growth_caches): Set
17040 edge_removal_hook_holder to NULL after removing it.
17042 2017-06-02 Sudakshina Das <sudi.das@arm.com>
17044 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
17045 comparision with zero.
17047 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
17048 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17049 for early expansion of vec_min and vec_max builtins.
17050 (builtin_function_type): Add min/max unsigned variants to those
17051 identified as having unsigned arguments.
17053 2017-06-02 Olivier Hainque <hainque@adacore.com>
17055 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
17057 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17059 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
17060 Use VALL_F16 iterator rather than VALL.
17062 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17064 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
17065 Emit CBNZ inside loop when doing a strong exchange and comparing
17066 against zero. Generate the CC flags after the loop.
17068 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
17070 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
17071 (dl_section_ref): New.
17072 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
17073 On AIX, append an expression to subtract the size of the
17074 section length to dl_section_ref.
17076 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
17078 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17079 for early expansion of vector absolute builtins.
17081 2017-06-02 Richard Biener <rguenther@suse.de>
17083 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
17084 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
17086 2017-06-02 Richard Biener <rguenther@suse.de>
17088 PR tree-optimization/80948
17089 * tree-tailcall.c (find_tail_calls): Track stmts to move in
17090 stmt order as well.
17092 2017-06-02 Richard Biener <rguenther@suse.de>
17094 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
17096 * tree-vect-stmts.c (process_use): Do not mark backedge defs
17097 for inductions as relevant.
17099 2017-06-02 Richard Biener <rguenther@suse.de>
17101 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
17102 (vectorizable_induction): ... this. Remove dead code.
17104 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
17106 * builtins. (expand_builtin_alloca): Remove second parameter and
17107 infer its value from the first parameter instead.
17108 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
17110 2017-06-02 Jakub Jelinek <jakub@redhat.com>
17112 PR rtl-optimization/80903
17113 * loop-doloop.c (add_test): Unshare sequence.
17115 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
17117 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
17119 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
17121 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
17123 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
17124 xlogue_layout::get_instance, logue_layout::xlogue_layout,
17125 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
17126 (xlogue_layout::get_stub_rtx): Make static.
17127 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
17128 (xlogue_layout::compute_stub_managed_regs): Rename to...
17129 (xlogue_layout::count_stub_managed_regs): ...this.
17130 (xlogue_layout::is_stub_managed_reg): New function.
17131 (xlogue_layout::m_stub_names): Rename to...
17132 (xlogue_layout::s_stub_names): ...this, make static.
17133 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
17134 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
17135 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
17136 xlogue_layout::s_stub_names): Instantiate statics.
17137 (stub_managed_regs): Remove.
17138 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
17139 (disable_call_ms2sysv_xlogues): Rename to...
17140 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
17141 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
17143 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
17144 change after reload_completed.
17145 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
17147 (ix86_expand_prologue): Likewise.
17148 (ix86_expand_epilogue): Likewise.
17149 (ix86_expand_split_stack_prologue): Likewise.
17150 (ix86_compute_frame_layout): Remove frame parameter ...
17151 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
17152 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
17154 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
17155 (ix86_frame): Move from here ...
17156 * config/i386/i386.h (ix86_frame): ... to here.
17157 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
17158 complete ix86_frame data structure instead. Remove some_ld_name.
17160 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
17162 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
17163 symbols that hold a DECL_VALUE_EXPR.
17165 2017-06-01 Martin Jambor <mjambor@suse.cz>
17167 PR tree-optimization/80898
17168 * tree-sra.c (process_subtree_disqualification): Removed.
17169 (disqualify_candidate): Do not acll
17170 process_subtree_disqualification.
17171 (subtree_mark_written_and_enqueue): New function.
17172 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
17173 RHS has been disqualified and re-queue LHS if necessary. Apart
17174 from that, ignore disqualified RHS.
17176 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17178 * config/s390/s390.c (s390_emit_epilogue): Disable early return
17179 address fetch for z10 or later.
17181 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17183 * config/arc/arc.md (tst_movb): Add guard when splitting.
17185 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17187 * config/arc/arc.c (arc_can_eliminate): Test against
17188 arc_frame_pointer_needed.
17190 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17192 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
17193 to prevent store reordering.
17194 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
17195 (type): Add block type.
17196 (stack_tie): Define special instruction to be used in
17199 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17201 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
17202 constraint. It is not valid for the pattern.
17203 (noncommutative_binary_comparison): Likewise.
17205 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17207 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
17210 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17212 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
17213 be used by the reg-alloc.
17215 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17217 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
17218 reg-alloc when having mul64 or mul32x16 instructions.
17219 (mulsidi3): Likewise.
17220 (umulsidi3): Likewise.
17221 (mulsi32x16): New pattern.
17222 (mulsi64): Likewise.
17223 (mulsidi64): Likewise.
17224 (umulsidi64): Likewise.
17225 (MUL32x16_REG): Define.
17226 (mul64_600): Use MUL32x16_REG.
17227 (mac64_600): Likewise.
17228 (umul64_600): Likewise.
17229 (umac64_600): Likewise.
17231 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17233 * config/arc/arc.md (mulsi3_700): Make it commutative.
17235 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
17237 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
17239 (*sign_extendsidi2_insn): Likewise for movstosw.
17241 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
17243 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
17244 the type of the input discriminant value. Convert the
17245 discriminant value of signedness vary.
17247 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
17249 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
17250 Add to -Wall section.
17252 2017-06-01 Richard Biener <rguenther@suse.de>
17254 PR middle-end/66313
17255 * fold-const.c (fold_plusminus_mult_expr): If the factored
17256 factor may be zero use a wrapping type for the inner operation.
17257 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
17258 and handle moved defs.
17259 (process_assignment): Properly guard the unary op case. Return a
17260 tri-state indicating that moving the stmt before the call may allow
17261 to continue. Pass through to_move.
17262 (find_tail_calls): Handle moving unrelated defs before
17265 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
17268 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
17269 splitter result in the canonical way.
17271 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
17273 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
17274 also for 32bit target. Update insn attributes.
17275 (zero-extendsidi2 splitter): Allow all registers for operand 1.
17277 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
17279 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
17280 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
17281 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
17282 (_mm_maskz_min_ss): New intrinsics.
17284 2017-05-31 Martin Liska <mliska@suse.cz>
17286 * tree-vect-loop.c (vect_create_epilog_for_reduction):
17287 Change comment style to one we normally use.
17288 (vectorizable_reduction): Likewise.
17289 (vectorizable_induction): Likewise.
17290 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
17291 (vectorizable_call): Likewise.
17292 (vectorizable_simd_clone_call): Likewise.
17293 (vectorizable_conversion): Likewise.
17294 (vectorizable_assignment): Likewise.
17295 (vectorizable_shift): Likewise.
17296 (vectorizable_operation): Likewise.
17297 (vectorizable_store): Likewise.
17298 (vectorizable_load): Likewise.
17299 * tree-vectorizer.h: Likewise.
17301 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
17303 * passes.c (emergency_dump_function): New.
17304 * tree-pass.h (emergency_dump_function): Declare.
17305 * plugin.c (plugins_internal_error_function): Remove.
17306 * plugin.h (plugins_internal_error_function): Remove declaration.
17307 * toplev.c (internal_error_function): New static function. Use it...
17308 (general_init): ...here.
17310 2017-05-31 Graham Markall <graham.markall@embecosm.com>
17312 * config/arc/arc.c (arc_print_operand): Handle constant operands.
17313 (arc_rtx_costs): Add costs for new patterns.
17314 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
17315 * config/arc/predicates.md: Add _1_2_3_operand predicate.
17317 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
17319 * tree-ssa-strlen.c (get_next_strinfo): New function.
17320 (get_stridx_plus_constant): Use it.
17321 (zero_length_string): Likewise.
17322 (adjust_related_strinfos): Likewise.
17323 (adjust_last_stmt): Likewise.
17325 2017-05-31 Richard Biener <rguenther@suse.de>
17328 * config/i386/i386.c (ix86_expand_builtin): Remove assert
17329 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
17331 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
17333 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
17334 loop_vinfo argument and use of dependence distance vectors.
17335 Check instead whether the two references differ only in their
17336 initial value and assume that they have the same alignment if the
17337 difference is a multiple of the vector alignment.
17338 (vect_analyze_data_refs_alignment): Update call accordingly.
17340 2017-05-31 Martin Liska <mliska@suse.cz>
17343 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
17345 2017-05-31 Bin Cheng <bin.cheng@arm.com>
17347 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
17348 (create_intersect_range_checks): Move from ...
17349 * tree-data-ref.c (create_intersect_range_checks_index)
17350 (create_intersect_range_checks): ... to here.
17351 (create_runtime_alias_checks): New function factored from ...
17352 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
17353 here. Call above function.
17354 * tree-data-ref.h (create_runtime_alias_checks): New function.
17356 2017-05-31 Bin Cheng <bin.cheng@arm.com>
17358 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
17359 segment length for dr_b and compute it in wide_int.
17361 2017-05-31 Richard Biener <rguenther@suse.de>
17363 PR tree-optimization/80906
17364 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
17365 and pass through iv_map.
17366 (copy_bb_and_scalar_dependences): Adjust.
17367 (translate_pending_phi_nodes): Likewise.
17368 (copy_loop_close_phi_args): Handle code-generating IVs instead
17371 2017-05-30 David Malcolm <dmalcolm@redhat.com>
17373 * diagnostic-color.c (color_dict): Add "type-diff".
17374 (parse_gcc_colors): Update comment.
17375 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
17376 -fdiagnostics-show-template-tree and -fno-elide-type.
17377 (GCC_COLORS): Add type-diff to example.
17379 (-fdiagnostics-show-template-tree): New.
17380 (-fno-elide-type): New.
17381 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
17382 the pp_format_decoder callback. Call any m_format_postprocessor's
17384 (pretty_printer::pretty_printer): Initialize
17385 m_format_postprocessor.
17386 (pretty_printer::~pretty_printer): Delete any
17387 m_format_postprocessor.
17388 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
17389 (class format_postprocessor): New class.
17390 (struct pretty_printer::format_decoder): Document the new parameters.
17391 (struct pretty_printer::m_format_postprocessor): New field.
17392 * tree-diagnostic.c (default_tree_printer): Update for new
17393 bool and const char ** params.
17394 * tree-diagnostic.h (default_tree_printer): Likewise.
17396 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
17398 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
17399 (lwa_operand): Delete rs6000_gen_cell_microcode test.
17400 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17401 rs6000_gen_cell_microcode code.
17402 (rs6000_final_prescan_insn): Delete.
17403 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
17404 "warn-cell-microcode" entries.
17405 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
17406 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
17407 throughout. Change cc_reg_not_micro_cr0_operand to
17408 cc_reg_not_cr0_operand throughout.
17409 (*extendhi<mode>2_noload): Delete.
17410 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
17411 (mwarn-cell-microcode): Delete.
17412 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
17413 -mgen-cell-microcode and -mwarn-cell-microcode.
17415 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
17418 * config/i386/constraints.md (Yd): New constraint.
17420 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
17421 and (?Yd, r) alternatives. Update insn attributes.
17422 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
17423 and (?*Yd, r) alternatives. Update insn attributes.
17424 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
17426 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
17428 * gimplify.c (gimplify_modify_expr): Don't create a
17429 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
17432 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
17434 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
17436 2017-05-30 Richard Biener <rguenther@suse.de>
17438 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
17439 and reduc_def fields.
17440 (STMT_VINFO_REDUC_TYPE): New define.
17441 (STMT_VINFO_REDUC_DEF): Likewise.
17442 (vect_force_simple_reduction): Adjust prototype.
17443 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
17444 (vect_is_simple_reduction): Remove check_reduction argument.
17445 (vect_force_simple_reduction): Adjust and set
17446 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
17447 (vectorizable_reduction): Do not re-do reduction analysis
17448 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
17449 * tree-parloops.c (gather_scalar_reductions): Adjust.
17451 2017-05-30 Richard Biener <rguenther@suse.de>
17453 PR middle-end/80901
17454 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
17457 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
17459 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
17460 Introduce unknown_misalignment parameter and remove vf.
17461 (vect_peeling_hash_get_lowest_cost):
17462 Pass unknown_misalignment parameter.
17463 (vect_enhance_data_refs_alignment):
17464 Fix unsupportable data ref treatment.
17466 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17468 * tree-vect-data-refs.c (vect_get_data_access_cost):
17469 Workaround for SLP handling.
17470 (vect_enhance_data_refs_alignment):
17471 Compute costs for doing no peeling at all, compare to the best
17472 peeling costs so far and avoid peeling if cheaper.
17474 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17476 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
17477 Return peeling info and set costs to zero for unlimited cost
17479 (vect_enhance_data_refs_alignment): Also inspect all datarefs
17480 with unknown misalignment. Compute and costs for unknown
17481 misalignment, compare them to the costs for known misalignment
17482 and choose the cheapest for peeling.
17484 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17486 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
17487 (vect_get_peeling_costs_all_drs): Create function.
17488 (vect_peeling_hash_get_lowest_cost):
17489 Use vect_get_peeling_costs_all_drs.
17490 (vect_peeling_supportable): Create function.
17491 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
17493 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17495 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
17496 DR_HAS_NEGATIVE_STEP.
17497 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
17498 (vect_enhance_data_refs_alignment): Use.
17499 (vect_duplicate_ssa_name_ptr_info): Use.
17500 * tree-vectorizer.h (dr_misalignment): Use.
17501 (known_alignment_for_access_p): Use.
17503 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17506 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
17508 (has_section_name): New function.
17510 2017-05-30 Martin Liska <mliska@suse.cz>
17513 * auto-profile.c (get_function_decl_from_block): Fix
17516 2017-05-30 Richard Biener <rguenther@suse.de>
17518 PR middle-end/80876
17519 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
17521 2017-05-30 Martin Liska <mliska@suse.cz>
17523 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
17524 * dumpfile.h (struct dump_file_info): Remove ctors.
17526 2017-05-30 Martin Liska <mliska@suse.cz>
17528 * predict.def: Fix GNU coding style.
17530 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
17532 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
17533 Mark 'to' argument with ATTRIBUTE_UNUSED.
17535 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
17537 * config/xtensa/xtensa.c (xtensa_emit_call): Use
17538 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
17539 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
17542 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
17544 * doc/install.texi (Options specification): Restore entry of
17545 --enable-sjlj-exceptions.
17547 2017-05-27 Michael Eager <eager@eagercon.com>
17550 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
17552 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
17554 * config/microblaze/microblaze.h
17555 (FIXED_REGISTERS): Update in macro.
17556 (CALL_USED_REGISTERS): Update in macro.
17558 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
17560 * doc/install.texi: Add links to macOS binary distributions.
17562 2017-05-27 Jakub Jelinek <jakub@redhat.com>
17566 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17568 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
17570 2017-05-26 Martin Liska <mliska@suse.cz>
17572 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
17574 2017-05-26 Martin Liska <mliska@suse.cz>
17576 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
17577 always leading ';; '.
17578 (dump_bb_info): Likewise.
17579 (brief_dump_cfg): Likewise.
17580 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
17581 * dumpfile.c: Remove usage of TDF_VERBOSE.
17582 * dumpfile.h (enum dump_kind): Likewise.
17583 (dump_gimple_bb_header): Do not use TDF_COMMENT.
17584 * print-tree.c (debug_verbose): Remove.
17585 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
17586 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
17587 * tree-diagnostic.c (default_tree_printer): Replace
17588 TDF_DIAGNOSTIC with TDF_SLIM.
17590 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17592 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
17593 in parameter loop, rather than loop_vinfo.
17594 (create_intersect_range_checks): Ditto.
17595 (vect_create_cond_for_alias_checks): Update call to above functions.
17597 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17599 PR tree-optimization/80815
17600 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
17601 for merging runtime alias checks. Handle negative DR_STEPs.
17603 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17605 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
17607 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
17608 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
17609 out code pruning runtime alias checks.
17610 * tree-data-ref.c (prune_runtime_alias_test_list): New function
17611 factored out from above.
17612 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17614 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
17616 (prune_runtime_alias_test_list): New decalaration.
17618 2017-05-26 Bin Cheng <bin.cheng@arm.com>
17620 * tree-vect-data-refs.c (compare_tree): Rename and move ...
17621 * tree-data-ref.c (data_ref_compare_tree): ... to here.
17622 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
17623 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
17624 (operator==, comp_dr_with_seg_len_pair): Ditto.
17625 (vect_prune_runtime_alias_test_list): Ditto.
17627 2017-05-26 Martin Liska <mliska@suse.cz>
17630 * params.def: Bound partial-inlining-entry-probability param.
17632 2017-05-26 Marek Polacek <polacek@redhat.com>
17635 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
17638 2017-05-26 Richard Biener <rguenther@suse.de>
17640 PR tree-optimization/80842
17641 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
17644 2017-05-26 Richard Biener <rguenther@suse.de>
17646 PR tree-optimization/80844
17647 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
17649 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
17651 * doc/md.texi (Machine Constraints): Update x86 family
17652 machine constraints section to match 'config/i386/constraints.md'.
17654 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
17656 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
17658 2017-05-25 Nathan Sidwell <nathan@acm.org>
17660 * doc/invoke.texi (--enable-languages): Update documentation.
17662 2017-05-25 Martin Liska <mliska@suse.cz>
17664 * dumpfile.c: Add TDF_FOLDING.
17665 * dumpfile.h (enum dump_kind): Likewise.
17666 * genmatch.c (dt_simplify::gen_1): Use it.
17668 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17670 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
17672 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17674 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
17675 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
17677 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
17679 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
17680 * match.pd (X == C): Rewrite it here.
17681 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
17682 with_certain_nonzero_bits2): New predicates.
17683 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
17685 2017-05-24 Nathan Sidwell <nathan@acm.org>
17687 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
17690 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
17693 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17695 * config/powerpcspe: New port. Files are copied from the rs6000
17696 port, with "rs6000" in filenames replaced by "powerpcspe".
17698 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
17700 PR rtl-optimization/80754
17701 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
17703 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
17705 * config/sparc/sparc.md (length): Return the correct value for -mflat
17706 sibcalls to match output_sibcall.
17708 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
17712 * config/rs6000/rs6000.c (struct machine_function): Add new field
17714 (rs6000_get_separate_components): Init that field, use it.
17715 (rs6000_components_for_bb): Use the field.
17717 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
17719 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
17721 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
17723 PR middle-end/80823
17724 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
17726 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17729 * config/s390/s390.c (s390_check_qrst_address): Check incoming
17730 address against address_operand predicate.
17731 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
17733 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
17735 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
17736 parameters passed indirectly.
17738 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
17740 * config/i386/i386.md (*movdi_internal): Remove SSE4
17741 alternative 18 (?r, *v). Update insn attributes.
17742 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
17743 Update insn attributes.
17744 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
17745 Update insn attributes.
17746 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
17747 alternative 1 (r, v). Remove isa attribute.
17748 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
17749 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
17750 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
17752 2017-05-23 Tom de Vries <tom@codesourcery.com>
17754 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
17757 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17759 * cgraphunit.c (symbol_table::process_new_functions): Update.
17760 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
17761 (inline_generate_summary): Rename to ...
17762 (ipa_fn_summary_generate): ... this one.
17763 (inline_read_summary): Rename to ...
17764 (ipa_fn_summary_read): ... this one.
17765 (inline_write_summary): Rename to ...
17766 (ipa_fn_summary_write): ... this one.
17767 (inline_free_summary): Rename to ...
17768 (ipa_free_fn_summary): ... this one.
17769 (pass_data_local_fn_summary, pass_local_fn_summary,
17770 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
17771 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
17772 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
17773 make_pass_ipa_fn_summary): New.
17774 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
17775 inline_write_summary, inline_free_summary): Remove.
17776 (ipa_free_fn_summary) : New.
17777 * ipa-inline.c (ipa_inline): Update.
17778 (pass_ipa_inline): Do not generate summaries.
17779 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
17781 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
17782 and add pass_ipa_fn_summary.
17783 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
17785 (make_pass_inline_parameters): Remove.
17787 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
17789 * omp-low.c (struct omp_context): Remove "default_kind" member.
17792 * omp-offload.c (execute_oacc_device_lower): Remove the
17793 parallelism dimensions function attributes for unparallelized
17794 OpenACC kernels constructs.
17796 2017-05-23 Martin Liska <mliska@suse.cz>
17798 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
17800 (cgraph_edge::make_speculative): Likewise.
17801 (cgraph_edge::resolve_speculation): Likewise.
17802 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
17803 (cgraph_node::dump): Likewise.
17804 * cgraph.h: Likewise.
17805 * cgraphunit.c (analyze_functions): Likewise.
17806 (symbol_table::compile): Likewise.
17807 * ipa-cp.c (print_all_lattices): Likewise.
17808 (determine_versionability): Likewise.
17809 (initialize_node_lattices): Likewise.
17810 (ipcp_verify_propagated_values): Likewise.
17811 (estimate_local_effects): Likewise.
17812 (update_profiling_info): Likewise.
17813 (create_specialized_node): Likewise.
17814 (perhaps_add_new_callers): Likewise.
17815 (decide_about_value): Likewise.
17816 (decide_whether_version_node): Likewise.
17817 (identify_dead_nodes): Likewise.
17818 (ipcp_store_bits_results): Likewise.
17819 * ipa-devirt.c (dump_targets): Likewise.
17820 (ipa_devirt): Likewise.
17821 * ipa-icf.c (sem_item::dump): Likewise.
17822 (sem_function::equals): Likewise.
17823 (sem_variable::equals): Likewise.
17824 (sem_item_optimizer::read_section): Likewise.
17825 (sem_item_optimizer::execute): Likewise.
17826 (congruence_class::dump): Likewise.
17827 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
17828 (dump_inline_summary): Likewise.
17829 (estimate_node_size_and_time): Likewise.
17830 (inline_analyze_function): Likewise.
17831 * ipa-inline-transform.c (inline_call): Likewise.
17832 * ipa-inline.c (report_inline_failed_reason): Likewise.
17833 (want_early_inline_function_p): Likewise.
17834 (edge_badness): Likewise.
17835 (update_edge_key): Likewise.
17836 (inline_small_functions): Likewise.
17837 * ipa-profile.c (ipa_profile): Likewise.
17838 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
17839 (ipa_make_edge_direct_to_target): Likewise.
17840 (remove_described_reference): Likewise.
17841 (ipa_impossible_devirt_target): Likewise.
17842 (propagate_controlled_uses): Likewise.
17843 (ipa_print_node_params): Likewise.
17844 (ipcp_transform_function): Likewise.
17845 * ipa-pure-const.c (pure_const_read_summary): Likewise.
17846 (propagate_pure_const): Likewise.
17847 * ipa-reference.c (generate_summary): Likewise.
17848 (read_write_all_from_decl): Likewise.
17849 (propagate): Likewise.
17850 (ipa_reference_read_optimization_summary): Likewise.
17851 * ipa-utils.c (ipa_merge_profiles): Likewise.
17852 * ipa.c (walk_polymorphic_call_targets): Likewise.
17853 (symbol_table::remove_unreachable_nodes): Likewise.
17854 (ipa_single_use): Likewise.
17855 * passes.c (execute_todo): Likewise.
17856 * predict.c (drop_profile): Likewise.
17857 * symtab.c (symtab_node::get_dump_name): New function.
17858 (symtab_node::dump_name): Likewise.
17859 (symtab_node::dump_asm_name): Likewise.
17860 (symtab_node::dump_references): Likewise.
17861 (symtab_node::dump_referring): Likewise.
17862 (symtab_node::dump_base): Likewise.
17863 (symtab_node::debug_symtab): Likewise.
17864 * tree-sra.c (convert_callers_for_node): Likewise.
17865 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
17866 * value-prof.c (init_node_map): Likewise.
17868 2017-05-23 Martin Liska <mliska@suse.cz>
17870 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
17871 and symtab_node::debug_symtab to symbol_table::debug.
17872 * cgraphunit.c (analyze_functions): Use the renamed function.
17873 (symbol_table::compile): Likewise.
17874 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
17875 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
17876 * passes.c (execute_todo): Likewise.
17877 * symtab.c (symbol_table::dump): New function.
17878 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
17880 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17882 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
17883 that nonconst implies exec.
17885 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17887 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
17888 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
17889 (inline_edge_summary_vec): Turn into ...
17890 (ipa_call_summaries): ... this one.
17891 (redirect_to_unreachable, edge_set_predicate,
17892 evaluate_properties_for_edge, inline_summary_alloc,
17893 reset_ipa_call_summary, reset_inline_summary,
17894 inline_summary_t::duplicate): Update.
17895 (inline_edge_duplication_hook): Turn to ...
17896 (ipa_call_summary_t::duplicate): ... this one.
17897 (inline_edge_removal_hook): Turn to ...
17898 (ipa_call_summary_t::remove): ... this one.
17899 (dump_inline_edge_summary): Turn to ...
17900 (dump_ipa_call_summary): ... this one.
17901 (estimate_function_body_sizes): Update.
17902 (inline_update_callee_summaries): Update.
17903 (remap_edge_change_prob): Update.
17904 (remap_edge_summaries): Update.
17905 (inline_merge_summary): Update.
17906 (do_estimate_edge_time): Update.
17907 (inline_generate_summary): Update.
17908 (inline_read_section): Update.
17909 (inline_read_summary): Update.
17910 (inline_free_summary): Update.
17911 * ipa-inline.c (can_inline_edge_p): Update.
17912 (compute_inlined_call_time): Update.
17913 (want_inline_small_function_p): Update.
17914 (edge_badness): Update.
17915 (early_inliner): Update.
17916 * ipa-inline.h (inline_edge_summary): Turn to ...
17917 (ipa_call_summary): ... this one.
17918 (ipa_call_summary_t): New class.
17919 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
17920 (ipa_call_summaries): New.
17921 (inline_edge_summary): Remove.
17922 (estimate_edge_growth): Update.
17923 * ipa-profile.c (ipa_propagate_frequency_1): Update.
17924 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17925 * ipa-split.c (execute_split_functions): Update.
17926 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
17928 2017-05-23 Tom de Vries <tom@codesourcery.com>
17930 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17931 attributes): Document rdrand effective target.
17933 2017-05-23 Tom de Vries <tom@codesourcery.com>
17935 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
17936 attributes): Sort alphabetically.
17938 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
17940 * config/avr/genmultilib.awk: Use gsub instead of gensub.
17942 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
17945 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
17946 V2DF/V2DI splat into two separate patterns, one that handles
17947 registers, and the other that only handles memory. Drop support
17948 for splatting from a GPR on ISA 2.07 and then splitting the
17949 splat into direct move and splat.
17950 (vsx_splat_<mode>_reg): Likewise.
17951 (vsx_splat_<mode>_mem): Likewise.
17953 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
17955 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
17957 2017-05-22 Jakub Jelinek <jakub@redhat.com>
17959 PR middle-end/80809
17960 * omp-low.c (finish_taskreg_remap): New function.
17961 (finish_taskreg_scan): If unit size of ctx->record_type
17962 is non-constant, unshare the size expression and replace
17963 decls in it with possible outer var refs.
17965 PR middle-end/80809
17966 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
17967 GOVD_SHARED rather than GOVD_PRIVATE with it.
17968 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
17969 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
17971 PR middle-end/80853
17972 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
17973 as last argument to build_outer_var_ref for pointer bases of array
17974 section reductions.
17976 2017-05-19 Martin Sebor <msebor@redhat.com>
17978 * print-tree.c (print_node): Print DECL_READ_P flag.
17980 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
17982 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
17983 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
17984 * cgraph.c: Likewise.
17985 * cgraphunit.c: Likewise.
17986 * gengtype.c: Likewise.
17987 * ipa-cp.c: Likewise.
17988 * ipa-devirt.c: Likewise.
17989 * ipa-icf.c: Likewise.
17990 * ipa-predicate.c: Likewise.
17991 * ipa-profile.c: Likewise.
17992 * ipa-prop.c: Likewise.
17993 * ipa-split.c: Likewise.
17995 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
17996 edge_predicate_pool, dump_inline_hints,
17997 inline_summary::account_size_time, redirect_to_unreachable,
17998 edge_set_predicate, set_hint_predicate,
17999 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
18000 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
18001 inline_summary_t::remove, remap_hint_predicate_after_duplication,
18002 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
18003 ipa_call_summary_t::remove, initialize_growth_caches,
18004 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
18005 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
18006 mark_modified, unmodified_parm_1, unmodified_parm,
18007 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
18008 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
18009 compute_bb_predicates, will_be_nonconstant_expr_predicate,
18010 will_be_nonconstant_predicate, record_modified_bb_info,
18011 get_minimal_bb, record_modified, param_change_prob,
18012 phi_result_unknown_predicate, predicate_for_phi_result,
18013 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
18014 estimate_function_body_sizes, compute_inline_parameters,
18015 compute_inline_parameters_for_curren, pass_data_inline_parameters,
18016 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
18017 inline_update_callee_summaries, remap_edge_change_prob,
18018 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
18019 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
18020 inline_analyze_function, inline_summary_t::insert,
18021 inline_generate_summary, read_ipa_call_summary, inline_read_section,
18022 inline_read_summary, write_ipa_call_summary, inline_write_summary,
18023 inline_free_summary): Move to ipa-fnsummary.h
18024 (predicate_t): Remove.
18025 * ipa-fnsummary.c: New file.
18026 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
18027 (enum inline_hints_vals, inline_hints, agg_position_info,
18028 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
18029 inline_summaries, ipa_call_summary, ipa_call_summary_t,
18030 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
18031 dump_inline_summary, dump_inline_hints, inline_generate_summary,
18032 inline_read_summary, inline_write_summary, inline_free_summary,
18033 inline_analyze_function, initialize_inline_failed,
18034 inline_merge_summary, inline_update_overall_summary,
18035 compute_inline_parameters): Move to ipa-fnsummary.h
18036 * ipa-fnsummary.h: New file.
18037 * ipa-inline-transform.h: Include ipa-inline.h.
18038 * ipa-inline.c: LIkewise.
18040 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18042 * ipa-inline.c (edge_badness): Use inlined_time instead of
18043 inline_summaries->get.
18045 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18047 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
18049 2017-05-22 Nathan Sidwell <nathan@acm.org>
18051 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
18052 (fdump-lang): Document 'raw' option.
18053 * dumpfile.h (TDI_tu): Delete.
18054 * dumpfile.c (dump_files): Remove translation-unit.
18055 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
18057 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
18059 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
18060 command option from $(AWK) call.
18061 * config/avr/genmultilib.awk: Simplify and rewrite so that it
18062 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
18063 [FORMAT]: Remove handling of variable.
18064 * config/avr/t-multilib: Regenerate.
18066 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18068 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
18070 (dump_inline_summary): Do not print self_time.
18071 (estimate_function_body_sizes): Do not set self_time.
18072 (compute_inline_parameters): Likewise.
18073 (inline_read_section, inline_write_summary): Do not stream self_time.
18074 * ipa-inline.h (inline_summary): Drop self_time.
18076 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18078 * ipa-inline-analysis.c (account_size_time): Rename to ...
18079 (inline_summary::account_size_time): ... this one.
18080 (reset_ipa_call_summary): Turn to ...
18081 (ipa_call_summary::reset): ... this one.
18082 (reset_inline_summary): Turn to ...
18083 (inline_summary::reset): ... this one.
18084 (inline_summary_t::remove): Update.
18085 (inline_summary_t::duplicate): Update.
18086 (ipa_call_summary_t::remove): Update.
18087 (dump_inline_summary): Update.
18088 (estimate_function_body_sizes): Update.
18089 (compute_inline_parameters): Update.
18090 (estimate_node_size_and_time): Update.
18091 (inline_merge_summary): Update.
18092 (inline_update_overall_summary): Update.
18093 (inline_read_section): Update.
18094 (inline_write_summary): Update.
18095 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
18096 add account_size_time and reset member functions.
18097 (ipa_call_summary): Add reset function.
18098 * ipa-predicate.h (predicate::operator &): Constify.
18100 2017-05-22 Richard Biener <rguenther@suse.de>
18102 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
18104 2017-05-19 Jason Merrill <jason@redhat.com>
18106 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
18108 2017-05-19 Marek Polacek <polacek@redhat.com>
18111 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
18112 TYPE_OVERFLOW_WRAPS checks.
18114 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
18116 * tree-core.h (enum omp_clause_default_kind): Add
18117 "OMP_CLAUSE_DEFAULT_PRESENT".
18118 * tree-pretty-print.c (dump_omp_clause): Handle it.
18119 * gimplify.c (enum gimplify_omp_var_data): Add
18120 "GOVD_MAP_FORCE_PRESENT".
18121 (gimplify_adjust_omp_clauses_1): Map it to
18122 "GOMP_MAP_FORCE_PRESENT".
18123 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
18125 * gimplify.c (oacc_default_clause): Clarify.
18127 2017-05-19 Nathan Sidwell <nathan@acm.org>
18129 LANG_HOOK_REGISTER_DUMPS
18130 * toplev.c (general_init): Call register dump lang hook.
18131 * doc/invoke.texi: Document -fdump-lang option family.
18132 * dumpfile.c (dump_files): Remove class dump here.
18133 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
18134 * dumpfile.h (tree_dump_index): Remove TDI_class.
18135 * langhooks-def.h (lhd_register_dumps): Declare.
18136 (LANG_HOOKS_REGISTER_DUMPS): Define.
18137 (LANG_HOOKS_INITIALIZER): Add it.
18138 * langhooks.c (lhd_register_dumps): Define.
18139 * langhooks.h (struct lang_hooks): Add register_dumps.
18141 2017-05-19 Nathan Sidwell <nathan@acm.org>
18143 * context.h (context::set_passes): New.
18144 * context.c (context::context): Do not create pass manager.
18145 * toplev.c (general_init): Create pass manager here.
18147 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
18149 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
18150 use this splitter if two add or or instructions would also work for
18151 the constant we want to generate.
18153 2017-05-19 Richard Biener <rguenther@suse.de>
18156 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
18157 predicate evaluation.
18159 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18161 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
18163 * ipa-inline.c (want_inline_small_function_p): Do not cast to
18166 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18168 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
18169 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
18170 (inline_edge_summary_vec): Turn into ...
18171 (ipa_call_summaries): ... this one.
18172 (redirect_to_unreachable, edge_set_predicate,
18173 evaluate_properties_for_edge, inline_summary_alloc,
18174 reset_ipa_call_summary, reset_inline_summary,
18175 inline_summary_t::duplicate): Update.
18176 (inline_edge_duplication_hook): Turn to ...
18177 (ipa_call_summary_t::duplicate): ... this one.
18178 (inline_edge_removal_hook): Turn to ...
18179 (ipa_call_summary_t::remove): ... this one.
18180 (dump_inline_edge_summary): Turn to ...
18181 (dump_ipa_call_summary): ... this one.
18182 (estimate_function_body_sizes): Update.
18183 (inline_update_callee_summaries): Update.
18184 (remap_edge_change_prob): Update.
18185 (remap_edge_summaries): Update.
18186 (inline_merge_summary): Update.
18187 (do_estimate_edge_time): Update.
18188 (inline_generate_summary): Update.
18189 (inline_read_section): Update.
18190 (inline_read_summary): Update.
18191 (inline_free_summary): Update.
18192 * ipa-inline.c (can_inline_edge_p): Update.
18193 (compute_inlined_call_time): Update.
18194 (want_inline_small_function_p): Update.
18195 (edge_badness): Update.
18196 (early_inliner): Update.
18197 * ipa-inline.h (inline_edge_summary): Turn to ...
18198 (ipa_call_summary): ... this one.
18199 (ipa_call_summary_t): New class.
18200 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
18201 (ipa_call_summaries): New.
18202 (inline_edge_summary): Remove.
18203 (estimate_edge_growth): Update.
18204 * ipa-profile.c (ipa_propagate_frequency_1): Update.
18205 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18206 * ipa-split.c (execute_split_functions): Update.
18207 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
18209 2017-05-19 Richard Biener <rguenther@suse.de>
18211 PR middle-end/80764
18212 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
18214 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
18216 * config/rs6000/rs6000.c (struct machine_function): Add field
18217 fpr_is_wrapped_separately.
18218 (rs6000_get_separate_components): Use 64 components. Handle the
18219 new FPR components.
18220 (rs6000_components_for_bb): Handle the FPR components.
18221 (rs6000_emit_prologue_components): Handle the FPR components.
18222 (rs6000_emit_epilogue_components): Handle the FPR components.
18223 (rs6000_set_handled_components): Handle the FPR components.
18224 (rs6000_emit_prologue): Don't output prologue code for those FPRs
18225 that are already separately shrink-wrapped.
18226 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
18227 that are already separately shrink-wrapped.
18229 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18232 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
18235 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
18236 (define_peephole2 for Altivec d-form load): Add peepholes to catch
18237 cases where the register allocator uses a move and an offsettable
18238 memory operation to/from a FPR register on ISA 2.06/2.07.
18239 (define_peephole2 for Altivec d-form store): Likewise.
18241 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
18244 * config/i386/mmx.md (*mov<mode>_internal): Enable
18245 alternatives 11, 12, 13 and 14 also for 32bit targets.
18246 Remove alternatives 15, 16, 17 and 18.
18247 * config/i386/sse.md (vec_concatv2di): Change
18248 alternative (!x, *y) to (x, ?!*Yn).
18250 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
18252 * dumpfile.h (enum dump_kind): Remove stray comma.
18254 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
18256 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
18257 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
18258 predicate::num_conditions
18259 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
18260 (CHANGED): turn into predicate::changed.
18261 (agg_position_info): Move to ipa-predicate.h
18262 (add_condition, predicate::add_clause, predicate::operator &=,
18263 predicate::or_with, predicate::evaluate, predicate::probability,
18264 dump_condition, dump_clause, predicate::dump,
18265 predicate::remap_after_duplication, predicate::remap_after_inlining,
18266 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
18267 (evaluate_conditions_for_known_args): Update.
18268 (set_cond_stmt_execution_predicate): Update.
18269 * ipa-inline.h: Include ipa-predicate.h
18270 (condition, inline_param_summary, conditions, agg_position_info,
18271 predicate): Move to ipa-predicate.h
18272 * ipa-predicate.c: New file.
18273 * ipa-predicate.h: New file.
18275 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
18277 * final.c (leaf_function_p): Check we are not in a sequence.
18279 2017-05-18 Martin Liska <mliska@suse.cz>
18281 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
18282 * dumpfile.c (dump_register): Use new enum dump_kind.
18283 (get_dump_file_name): Likewise.
18284 (dump_enable_all): Likewise.
18285 (dump_switch_p_1): Likewise.
18286 (enable_rtl_dump_file): Remove usage of TDF_RTL.
18287 * dumpfile.h (enum dump_kind): New enum type.
18288 (struct dump_file_info): Create constructor and
18289 format fields and comments.
18290 * passes.c (pass_manager::register_one_dump_file):
18292 * statistics.c (statistics_early_init): Likewise.
18293 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
18294 TDF_TREE with TDF_SLIM.
18295 (gather_memory_references_ref): Likewise.
18297 2017-05-18 Martin Liska <mliska@suse.cz>
18299 * vec.h (struct vnull): Use it.
18301 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
18303 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
18304 (true_predicate, false_predicate, true_predicate_p,
18305 false_predicate_p): Remove.
18306 (single_cond_predicate, not_inlined_predicate): Turn to member function
18308 (add_condition): Update.
18309 (add_clause): Turn to...
18310 (predicate::add_clause): ... this one; update; allow passing NULL
18312 (and_predicates): Turn to ...
18313 (predicate::operator &=): ... this one.
18314 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
18315 (or_predicates): Turn to ...
18316 (predicate::or_with): ... this one.
18317 (evaluate_predicate): Turn to ...
18318 (predicate::evaluate): ... this one.
18319 (predicate_probability): Turn to ...
18320 (predicate::probability): ... this one.
18321 (dump_condition): Update.
18322 (dump_predicate): Turn to ...
18323 (predicate::dump): ... this one.
18324 (account_size_time): Update.
18325 (edge_set_predicate): Update.
18326 (set_hint_predicate): UPdate.
18327 (evaluate_conditions_for_known_args): Update.
18328 (evaluate_properties_for_edge): Update.
18329 (remap_predicate_after_duplication): Turn to...
18330 (predicate::remap_after_duplication): ... this one.
18331 (remap_hint_predicate_after_duplication): Update.
18332 (inline_summary_t::duplicate): UPdate.
18333 (dump_inline_edge_summary): Update.
18334 (dump_inline_summary): Update.
18335 (set_cond_stmt_execution_predicate): Update.
18336 (set_switch_stmt_execution_predicate): Update.
18337 (compute_bb_predicates): Update.
18338 (will_be_nonconstant_expr_predicate): Update.
18339 (will_be_nonconstant_predicate): Update.
18340 (phi_result_unknown_predicate): Update.
18341 (predicate_for_phi_result): Update.
18342 (array_index_predicate): Update.
18343 (estimate_function_body_sizes): Update.
18344 (estimate_node_size_and_time): Update.
18345 (estimate_ipcp_clone_size_and_time): Update.
18346 (remap_predicate): Rename to ...
18347 (predicate::remap_after_inlining): ... this one.
18348 (remap_hint_predicate): Update.
18349 (inline_merge_summary): Update.
18350 (inline_update_overall_summary): Update.
18351 (estimate_size_after_inlining): Update.
18352 (read_predicate): Rename to ...
18353 (predicate::stream_in): ... this one.
18354 (read_inline_edge_summary): Update.
18355 (write_predicate): Rename to ...
18356 (predicate::stream_out): ... this one.
18357 (write_inline_edge_summary): Update.
18358 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
18359 (clause_t): Turn to uint32_t
18360 (predicate): Turn to class; implement constructor and operators
18362 (size_time_entry): Update.
18363 (inline_summary): Update.
18364 (inline_edge_summary): Update.
18366 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
18368 * fold-const.c (fold_binary_loc): Move transformation...
18369 * match.pd (C - X CMP X): ... here.
18371 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
18373 * config/sparc/sparc.c (sparc_option_override): Set function
18374 alignment for -mcpu=niagara7 to 64 to match the I$ line.
18375 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
18377 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
18379 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
18381 2017-05-18 Marek Polacek <polacek@redhat.com>
18384 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
18385 (pass_ubsan::execute): Call gimple_assign_single_p instead of
18386 gimple_assign_load_p.
18388 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
18390 PR middle-end/80692
18391 * real.c (do_compare): Give decimal_do_compare preference over
18392 comparing just the signs.
18394 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
18396 * doc/md.texi (Canonicalization of Instructions): Describe the
18397 canonical form of instructions that inherently set a condition
18400 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
18402 PR middle-end/80775
18403 * tree-cfg.c: Move deletion of unreachable case statements to after
18404 the merging of consecutive case labels.
18406 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
18408 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
18409 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
18410 restoring of callee-saved registers.
18412 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
18414 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
18415 * config/visium/visium.c (single_set_and_flags): Likewise.
18416 * config/visium/visium.md (Substitutions): Likewise.
18418 2017-05-17 Martin Liska <mliska@suse.cz>
18420 * cfg.c: Introduce dump_flags_t type and
18421 use it instead of int type.
18423 * cfghooks.c: Likewise.
18424 * cfghooks.h (struct cfg_hooks): Likewise.
18425 * cfgrtl.c: Likewise.
18426 * cfgrtl.h: Likewise.
18427 * cgraph.c (cgraph_node::get_body): Likewise.
18428 * coretypes.h: Likewise.
18429 * domwalk.c: Likewise.
18430 * domwalk.h: Likewise.
18431 * dumpfile.c (struct dump_option_value_info): Likewise.
18432 (dump_enable_all): Likewise.
18433 (dump_switch_p_1): Likewise.
18434 (opt_info_switch_p): Likewise.
18435 * dumpfile.h (enum tree_dump_index): Likewise.
18436 (struct dump_file_info): Likewise.
18437 * genemit.c: Likewise.
18438 * generic-match-head.c: Likewise.
18439 * gengtype.c (open_base_files): Likewise.
18440 * gimple-pretty-print.c: Likewise.
18441 * gimple-pretty-print.h: Likewise.
18442 * graph.c (print_graph_cfg): Likewise.
18443 * graphite-scop-detection.c (dot_all_sese): Likewise.
18444 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
18445 * loop-unroll.c (report_unroll): Likewise.
18446 * passes.c (pass_manager::register_one_dump_file): Likewise.
18447 * print-tree.c: Likewise.
18448 * statistics.c: Likewise.
18449 * tree-cfg.c: Likewise.
18450 * tree-cfg.h: Likewise.
18451 * tree-dfa.c: Likewise.
18452 * tree-dfa.h: Likewise.
18453 * tree-dump.c (dump_function): Likewise.
18454 * tree-dump.h (struct dump_info): Likewise.
18455 * tree-pretty-print.c: Likewise.
18456 * tree-pretty-print.h: Likewise.
18457 * tree-ssa-live.c: Likewise.
18458 * tree-ssa-live.h: Likewise.
18459 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
18460 * tree-vect-loop.c: Likewise.
18461 * tree-vect-slp.c: Likewise.
18463 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
18464 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18466 PR tree-optimization/80457
18467 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
18468 of all arguments to a statement as scalar_to_vec operations.
18469 (vectorizable_call): Adjust call to vect_model_simple_cost for
18471 (vectorizable_conversion): Likewise.
18472 (vectorizable_assignment): Likewise.
18473 (vectorizable_shift): Likewise.
18474 (vectorizable_operation): Likewise.
18475 (vectorizable_comparison): Likewise.
18476 (vect_is_simple_cond): Record the def types for operands.
18477 (vectorizable_condition): Likewise, call vect_model_simple_cost.
18478 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
18479 for statement argument count.
18481 2017-05-16 Carl Love <cel@us.ibm.com>
18483 * config/rs6000/rs6000-c: Add support for built-in functions
18484 vector unsigned long long vec_bperm (vector unsigned long long,
18485 vector unsigned char)
18486 vector signed long long vec_mule (vector signed int,
18488 vector unsigned long long vec_mule (vector unsigned int,
18489 vector unsigned int)
18490 vector signed long long vec_mulo (vector signed int,
18492 vector unsigned long long vec_mulo (vector unsigned int,
18493 vector unsigned int)
18494 vector signed char vec_sldw (vector signed char,
18495 vector signed char,
18497 vector unsigned char vec_sldw (vector unsigned char,
18498 vector unsigned char,
18500 vector signed short vec_sldw (vector signed short,
18501 vector signed short,
18503 vector unsigned short vec_sldw (vector unsigned short,
18504 vector unsigned short,
18506 vector signed int vec_sldw (vector signed int,
18509 vector unsigned int vec_sldw (vector unsigned int,
18510 vector unsigned int,
18512 vector signed long long vec_sldw (vector signed long long,
18513 vector signed long long,
18515 vector unsigned long long vec_sldw (vector unsigned long long,
18516 vector unsigned long long,
18518 * config/rs6000/rs6000-c: Add support for built-in functions
18519 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
18520 * config/rs6000/altivec.h: Add defintion for vec_sldw.
18521 * doc/extend.texi: Update the built-in documentation for the
18522 new built-in functions.
18524 2017-05-16 Marek Polacek <polacek@redhat.com>
18528 * tree.c (save_expr): Don't fold the expression.
18530 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
18532 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
18533 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
18534 and (?*y,m). Update insn attributes.
18536 2017-05-16 Martin Liska <mliska@suse.cz>
18538 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
18539 flags argument of print_gimple_stmt, print_gimple_expr,
18540 print_generic_stmt and print_generic_expr.
18541 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
18542 * coretypes.h: Likewise.
18543 * except.c (dump_eh_tree): Likewise.
18544 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
18545 * gimple-pretty-print.h: Likewise.
18546 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
18547 (backprop::push_to_worklist): Likewise.
18548 (backprop::pop_from_worklist): Likewise.
18549 (backprop::process_use): Likewise.
18550 (backprop::intersect_uses): Likewise.
18551 (note_replacement): Likewise.
18552 * gimple-ssa-store-merging.c
18553 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
18554 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
18555 (pass_store_merging::execute): Likewise.
18556 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
18557 (ssa_base_cand_dump_callback): Likewise.
18558 (dump_incr_vec): Likewise.
18559 (replace_refs): Likewise.
18560 (replace_mult_candidate): Likewise.
18561 (create_add_on_incoming_edge): Likewise.
18562 (create_phi_basis): Likewise.
18563 (insert_initializers): Likewise.
18564 (all_phi_incrs_profitable): Likewise.
18565 (introduce_cast_before_cand): Likewise.
18566 (replace_one_candidate): Likewise.
18567 * gimplify.c (gimplify_expr): Likewise.
18568 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
18569 (set_rename): Likewise.
18570 (rename_uses): Likewise.
18571 (copy_loop_phi_nodes): Likewise.
18572 (add_close_phis_to_merge_points): Likewise.
18573 (copy_loop_close_phi_args): Likewise.
18574 (copy_cond_phi_args): Likewise.
18575 (graphite_copy_stmts_from_block): Likewise.
18576 (translate_pending_phi_nodes): Likewise.
18577 * graphite-poly.c (print_pdr): Likewise.
18578 (dump_gbb_cases): Likewise.
18579 (dump_gbb_conditions): Likewise.
18580 (print_scop_params): Likewise.
18581 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
18582 (build_cross_bb_scalars_use): Likewise.
18583 (gather_bbs::before_dom_children): Likewise.
18584 * hsa-dump.c (dump_hsa_immed): Likewise.
18585 * ipa-cp.c (print_ipcp_constant_value): Likewise.
18586 (get_replacement_map): Likewise.
18587 * ipa-inline-analysis.c (dump_condition): Likewise.
18588 (estimate_function_body_sizes): Likewise.
18589 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
18590 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
18591 * ipa-prop.c (ipa_dump_param): Likewise.
18592 (ipa_print_node_jump_functions_for_edge): Likewise.
18593 (ipa_modify_call_arguments): Likewise.
18594 (ipa_modify_expr): Likewise.
18595 (ipa_dump_param_adjustments): Likewise.
18596 (ipa_dump_agg_replacement_values): Likewise.
18597 (ipcp_modif_dom_walker::before_dom_children): Likewise.
18598 * ipa-pure-const.c (check_stmt): Likewise.
18599 (pass_nothrow::execute): Likewise.
18600 * ipa-split.c (execute_split_functions): Likewise.
18601 * omp-offload.c (dump_oacc_loop_part): Likewise.
18602 (dump_oacc_loop): Likewise.
18603 * trans-mem.c (tm_log_emit): Likewise.
18604 (tm_memopt_accumulate_memops): Likewise.
18605 (dump_tm_memopt_set): Likewise.
18606 (dump_tm_memopt_transform): Likewise.
18607 * tree-cfg.c (gimple_verify_flow_info): Likewise.
18608 (print_loop): Likewise.
18609 * tree-chkp-opt.c (chkp_print_addr): Likewise.
18610 (chkp_gather_checks_info): Likewise.
18611 (chkp_get_check_result): Likewise.
18612 (chkp_remove_check_if_pass): Likewise.
18613 (chkp_use_outer_bounds_if_possible): Likewise.
18614 (chkp_reduce_bounds_lifetime): Likewise.
18615 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
18616 (chkp_mark_completed_bounds): Likewise.
18617 (chkp_register_incomplete_bounds): Likewise.
18618 (chkp_mark_invalid_bounds): Likewise.
18619 (chkp_maybe_copy_and_register_bounds): Likewise.
18620 (chkp_build_returned_bound): Likewise.
18621 (chkp_get_bound_for_parm): Likewise.
18622 (chkp_build_bndldx): Likewise.
18623 (chkp_get_bounds_by_definition): Likewise.
18624 (chkp_generate_extern_var_bounds): Likewise.
18625 (chkp_get_bounds_for_decl_addr): Likewise.
18626 * tree-chrec.c (chrec_apply): Likewise.
18627 * tree-data-ref.c (dump_data_reference): Likewise.
18628 (dump_subscript): Likewise.
18629 (dump_data_dependence_relation): Likewise.
18630 (analyze_overlapping_iterations): Likewise.
18631 * tree-inline.c (expand_call_inline): Likewise.
18632 (tree_function_versioning): Likewise.
18633 * tree-into-ssa.c (dump_defs_stack): Likewise.
18634 (dump_currdefs): Likewise.
18635 (dump_names_replaced_by): Likewise.
18636 (dump_update_ssa): Likewise.
18637 (update_ssa): Likewise.
18638 * tree-object-size.c (pass_object_sizes::execute): Likewise.
18639 * tree-parloops.c (build_new_reduction): Likewise.
18640 (try_create_reduction_list): Likewise.
18641 (ref_conflicts_with_region): Likewise.
18642 (oacc_entry_exit_ok_1): Likewise.
18643 (oacc_entry_exit_single_gang): Likewise.
18644 * tree-pretty-print.h: Likewise.
18645 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
18646 (get_scalar_evolution): Likewise.
18647 (add_to_evolution): Likewise.
18648 (get_loop_exit_condition): Likewise.
18649 (analyze_evolution_in_loop): Likewise.
18650 (analyze_initial_condition): Likewise.
18651 (analyze_scalar_evolution): Likewise.
18652 (instantiate_scev): Likewise.
18653 (number_of_latch_executions): Likewise.
18654 (gather_chrec_stats): Likewise.
18655 (final_value_replacement_loop): Likewise.
18656 (scev_const_prop): Likewise.
18657 * tree-sra.c (dump_access): Likewise.
18658 (disqualify_candidate): Likewise.
18659 (create_access): Likewise.
18660 (reject): Likewise.
18661 (maybe_add_sra_candidate): Likewise.
18662 (create_access_replacement): Likewise.
18663 (analyze_access_subtree): Likewise.
18664 (analyze_all_variable_accesses): Likewise.
18665 (sra_modify_assign): Likewise.
18666 (initialize_constant_pool_replacements): Likewise.
18667 (find_param_candidates): Likewise.
18668 (decide_one_param_reduction): Likewise.
18669 (replace_removed_params_ssa_names): Likewise.
18670 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
18671 * tree-ssa-copy.c (dump_copy_of): Likewise.
18672 (copy_prop_visit_cond_stmt): Likewise.
18673 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
18674 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
18675 (record_equivalences_from_stmt): Likewise.
18676 * tree-ssa-dse.c (compute_trims): Likewise.
18677 (delete_dead_call): Likewise.
18678 (delete_dead_assignment): Likewise.
18679 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
18680 (forward_propagate_into_cond): Likewise.
18681 (pass_forwprop::execute): Likewise.
18682 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
18683 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
18685 (move_computations_worker): Likewise.
18686 (execute_sm): Likewise.
18687 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
18688 (remove_exits_and_undefined_stmts): Likewise.
18689 (remove_redundant_iv_tests): Likewise.
18690 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
18691 (adjust_iv_update_pos): Likewise.
18692 * tree-ssa-math-opts.c (bswap_replace): Likewise.
18693 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
18694 (value_replacement): Likewise.
18695 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
18696 * tree-ssa-pre.c (print_pre_expr): Likewise.
18697 (get_representative_for): Likewise.
18698 (create_expression_by_pieces): Likewise.
18699 (insert_into_preds_of_block): Likewise.
18700 (eliminate_insert): Likewise.
18701 (eliminate_dom_walker::before_dom_children): Likewise.
18702 (eliminate): Likewise.
18703 (remove_dead_inserted_code): Likewise.
18704 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
18705 * tree-ssa-reassoc.c (get_rank): Likewise.
18706 (eliminate_duplicate_pair): Likewise.
18707 (eliminate_plus_minus_pair): Likewise.
18708 (eliminate_not_pairs): Likewise.
18709 (undistribute_ops_list): Likewise.
18710 (eliminate_redundant_comparison): Likewise.
18711 (update_range_test): Likewise.
18712 (optimize_range_tests_var_bound): Likewise.
18713 (optimize_vec_cond_expr): Likewise.
18714 (rewrite_expr_tree): Likewise.
18715 (rewrite_expr_tree_parallel): Likewise.
18716 (linearize_expr): Likewise.
18717 (break_up_subtract): Likewise.
18718 (linearize_expr_tree): Likewise.
18719 (attempt_builtin_powi): Likewise.
18720 (attempt_builtin_copysign): Likewise.
18721 (transform_stmt_to_copy): Likewise.
18722 (transform_stmt_to_multiply): Likewise.
18723 (dump_ops_vector): Likewise.
18724 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
18725 (print_scc): Likewise.
18726 (set_ssa_val_to): Likewise.
18727 (visit_reference_op_store): Likewise.
18728 (visit_use): Likewise.
18729 (sccvn_dom_walker::before_dom_children): Likewise.
18730 (run_scc_vn): Likewise.
18731 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18733 (expr_hash_elt::print): Likewise.
18734 (const_and_copies::pop_to_marker): Likewise.
18735 (const_and_copies::record_const_or_copy_raw): Likewise.
18736 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
18737 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
18738 (dump_predicates): Likewise.
18739 (find_uninit_use): Likewise.
18740 (warn_uninitialized_phi): Likewise.
18741 (pass_late_warn_uninitialized::execute): Likewise.
18742 * tree-ssa.c (verify_vssa): Likewise.
18743 (verify_ssa): Likewise.
18744 (maybe_optimize_var): Likewise.
18745 * tree-vrp.c (dump_value_range): Likewise.
18746 (dump_all_value_ranges): Likewise.
18747 (dump_asserts_for): Likewise.
18748 (register_edge_assert_for_2): Likewise.
18749 (vrp_visit_cond_stmt): Likewise.
18750 (vrp_visit_switch_stmt): Likewise.
18751 (vrp_visit_stmt): Likewise.
18752 (vrp_visit_phi_node): Likewise.
18753 (simplify_cond_using_ranges_1): Likewise.
18754 (fold_predicate_in): Likewise.
18755 (evrp_dom_walker::before_dom_children): Likewise.
18756 (evrp_dom_walker::push_value_range): Likewise.
18757 (evrp_dom_walker::pop_value_range): Likewise.
18758 (execute_early_vrp): Likewise.
18760 2017-05-16 Richard Biener <rguenther@suse.de>
18762 * dwarf2out.c (loc_list_from_tree_1): Do not create
18763 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
18765 2017-05-16 Richard Biener <rguenther@suse.de>
18767 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
18769 (note_variable_value_in_expr): If we resolved the decl ref
18770 do not push to the stack.
18772 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
18774 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
18775 operations in fast-math mode.
18776 (vaddq_f16): Likewise.
18777 (vmul_f16): Likewise.
18778 (vmulq_f16): Likewise.
18779 (vsub_f16): Likewise.
18780 (vsubq_f16): Likewise.
18781 * config/arm/neon.md (add<mode>3): New.
18783 (fma:<VH:mode>3): New. Also remove outdated comment.
18786 2017-05-16 Martin Liska <mliska@suse.cz>
18790 * ipa-devirt.c (warn_types_mismatch): Fix typo.
18791 (odr_types_equivalent_p): Likewise.
18793 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
18795 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
18797 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
18800 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
18801 non-interunit SSE move alternatives with '?'.
18802 (zero-extendsidi peephole2): New peephole to skip intermediate
18803 general register in SSE zero-extend sequence.
18805 2017-05-15 Jeff Law <law@redhat.com>
18807 * reorg.c (relax_delay_slots): Create a new variable to hold
18808 the temporary target rather than clobbering TARGET_LABEL.
18810 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
18811 missing argument to extract_bit_field call.
18812 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
18814 2017-05-15 Martin Liska <mliska@suse.cz>
18817 * gcc.c (process_command): Do not allow empty argument of -o option.
18819 2017-05-15 Renlin Li <renlin.li@arm.com>
18821 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
18822 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
18823 * config/aarch64/constraints.md (Usf): Add long call check.
18824 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
18825 (call_value): Likewise.
18826 (sibcall): Likewise.
18827 (sibcall_value): Likewise.
18829 (call_value_insn): New.
18830 (sibcall_insn): Update rtx pattern.
18831 (sibcall_value_insn): Likewise.
18832 (call_internal): Remove.
18833 (call_value_internal): Likewise.
18834 (sibcall_internal): Likewise.
18835 (sibcall_value_internal): Likewise.
18836 (call_reg): Likewise.
18837 (call_symbol): Likewise.
18838 (call_value_reg): Likewise.
18839 (call_value_symbol): Likewise.
18841 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
18844 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
18846 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
18848 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
18849 compatible with CCGOCmode and with CCZmode.
18851 2017-05-14 Martin Sebor <msebor@redhat.com>
18853 PR middle-end/77671
18854 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
18855 (gimple_fold_builtin_snprintf): Same.
18856 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
18857 (gimple_fold_builtin_snprintf): Same.
18858 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
18859 of character types.
18860 (is_call_safe): New function.
18861 (try_substitute_return_value): Call it.
18862 (try_simplify_call): New function.
18863 (pass_sprintf_length::handle_gimple_call): Call it.
18865 2017-05-14 Martin Sebor <msebor@redhat.com>
18867 PR middle-end/80669
18868 * builtins.c (expand_builtin_stpncpy): Simplify.
18870 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
18872 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
18873 * config/i386/i386.h
18874 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
18875 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
18876 (struct machine_function): Add new members call_ms2sysv,
18877 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
18878 (struct machine_frame_state): New fields sp_realigned and
18879 sp_realigned_offset.
18880 * config/i386/i386.c
18881 (enum xlogue_stub): New enum.
18882 (enum xlogue_stub_sets): New enum.
18883 (class xlogue_layout): New class.
18884 (struct ix86_frame): New fields stack_realign_allocate_offset,
18885 stack_realign_offset and outlined_save_offset. Modify comments to
18886 detail stack layout when using out-of-line stubs.
18887 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
18888 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
18889 -mcall-ms2sysv-xlogues.
18890 (stub_managed_regs): New static variable.
18891 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
18892 registers managed by out-of-line stub.
18893 (disable_call_ms2sysv_xlogues): New function.
18894 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
18895 m->call_ms2sysv when appropriate and compute frame layout for
18897 (sp_valid_at, fp_valid_at): New inline functions.
18898 (choose_basereg): New function.
18899 (choose_baseaddr): Add align parameter, use choose_basereg and modify
18901 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
18902 Use align parameter of choose_baseaddr to generated aligned SSE movs
18904 (pro_epilogue_adjust_stack): Modify to track
18905 machine_frame_state::sp_realigned.
18906 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
18907 (ix86_nsaved_sseregs): Likewise.
18908 (ix86_emit_save_regs): Likewise.
18909 (ix86_emit_save_regs_using_mov): Likewise.
18910 (ix86_emit_save_sse_regs_using_mov): Likewise.
18911 (get_scratch_register_on_entry): Likewise.
18912 (gen_frame_set): New function.
18913 (gen_frame_load): Likewise.
18914 (gen_frame_store): Likewise.
18915 (emit_outlined_ms2sysv_save): Likewise.
18916 (emit_outlined_ms2sysv_restore): Likewise.
18917 (ix86_expand_prologue): Modify stack re-alignment code and call
18918 emit_outlined_ms2sysv_save when appropriate.
18919 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
18920 parameter rtx_insn *insn, which allows the function to be used to only
18921 generate the notes.
18922 (ix86_expand_epilogue): Modify validity checks of frame and stack
18923 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
18924 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
18925 * config/i386/predicates.md
18926 (save_multiple): New predicate.
18927 (restore_multiple): Likewise.
18928 * config/i386/sse.md
18929 (save_multiple<mode>): New pattern.
18930 (save_multiple_realign<mode>): Likewise.
18931 (restore_multiple<mode>): Likewise.
18932 (restore_multiple_and_return<mode>): Likewise.
18933 (restore_multiple_leave_return<mode>): Likewise.
18934 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
18936 2017-05-14 Julia Koval <julia.koval@intel.com>
18938 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
18939 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
18940 (__builtin_ia32_xsetbv): New builtins.
18941 * config/i386/i386.c (ix86_expand_special_args_builtin):
18943 (ix86_expand_builtin): Special expand for new intrinsics.
18944 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
18945 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
18946 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
18948 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18950 * cfganal.c (inverted_post_order_compute): Change argument type
18952 * cfganal.h (inverted_post_order_compute): Adjust prototype.
18953 * df-core.c (rest_of_handle_df_initialize): Adjust.
18954 (rest_of_handle_df_finish): Likewise.
18955 (df_analyze_1): Likewise.
18956 (df_analyze): Likewise.
18957 (loop_inverted_post_order_compute): Change argument to be a vec *.
18958 (df_analyze_loop): Adjust.
18959 (df_get_n_blocks): Likewise.
18960 (df_get_postorder): Likewise.
18961 * df.h (struct df_d): Change field to be a vec.
18962 * lcm.c (compute_laterin): Adjust.
18963 (compute_available): Likewise.
18964 * lra-lives.c (lra_create_live_ranges_1): Likewise.
18965 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
18966 * tree-ssa-pre.c (compute_antic): Likewise.
18968 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18970 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
18971 (depth_first_search::depth_first_search): Change structure init
18972 function to this constructor.
18973 (depth_first_search::add_bb): Rename function to this member.
18974 (depth_first_search::execute): Likewise.
18975 (flow_dfs_compute_reverse_finish): Adjust.
18977 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18979 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
18980 (longest_simple_path): Likewise.
18981 * shrink-wrap.c (spread_components): Likewise.
18982 (disqualify_problematic_components): Likewise.
18983 (emit_common_heads_for_components): Likewise.
18984 (emit_common_tails_for_components): Likewise.
18985 (insert_prologue_epilogue_for_components): Likewise.
18987 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18989 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
18992 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
18994 * df-core.c (df_set_blocks): Start using auto_bitmap.
18995 (df_compact_blocks): Likewise.
18996 * df-problems.c (df_rd_confluence_n): Likewise.
18997 * df-scan.c (df_insn_rescan_all): Likewise.
18998 (df_process_deferred_rescans): Likewise.
18999 (df_update_entry_block_defs): Likewise.
19000 (df_update_exit_block_uses): Likewise.
19001 (df_entry_block_bitmap_verify): Likewise.
19002 (df_exit_block_bitmap_verify): Likewise.
19003 (df_scan_verify): Likewise.
19004 * lra-constraints.c (lra_constraints): Likewise.
19005 (undo_optional_reloads): Likewise.
19006 (lra_undo_inheritance): Likewise.
19007 * lra-remat.c (calculate_gen_cands): Likewise.
19008 (do_remat): Likewise.
19009 * lra-spills.c (assign_spill_hard_regs): Likewise.
19010 (spill_pseudos): Likewise.
19011 * tree-ssa-pre.c (bitmap_set_and): Likewise.
19012 (bitmap_set_subtract_values): Likewise.
19014 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19016 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
19017 management with auto_bitmap.
19018 (fix_inter_tick): Likewise.
19019 (fix_recovery_deps): Likewise.
19020 * ira.c (add_store_equivs): Likewise.
19021 (find_moveable_pseudos): Likewise.
19022 (split_live_ranges_for_shrink_wrap): Likewise.
19023 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
19024 (rtx_reuse_manager::seen_def_p): Likewise.
19025 (rtx_reuse_manager::set_seen_def): Likewise.
19026 * print-rtl.h (class rtx_reuse_manager): Likewise.
19028 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19030 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
19032 (migrate_btr_def): Likewise.
19033 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
19034 * df-core.c (loop_post_order_compute): Likewise.
19035 (loop_inverted_post_order_compute): Likewise.
19036 * hsa-common.h: Likewise.
19037 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
19038 * init-regs.c (initialize_uninitialized_regs): Likewise.
19039 * ipa-inline.c (resolve_noninline_speculation): Likewise.
19040 (inline_small_functions): Likewise.
19041 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
19042 * ira.c (combine_and_move_insns): Likewise.
19043 (build_insn_chain): Likewise.
19044 * loop-invariant.c (find_invariants): Likewise.
19045 * lower-subreg.c (propagate_pseudo_copies): Likewise.
19046 * predict.c (tree_predict_by_opcode): Likewise.
19047 (predict_paths_leading_to): Likewise.
19048 (predict_paths_leading_to_edge): Likewise.
19049 (estimate_loops_at_level): Likewise.
19050 (estimate_loops): Likewise.
19051 * shrink-wrap.c (try_shrink_wrapping): Likewise.
19052 (spread_components): Likewise.
19053 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
19054 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
19055 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
19056 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
19057 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
19058 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
19059 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
19060 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
19061 (mark_threaded_blocks): Likewise.
19062 (thread_through_all_blocks): Likewise.
19063 * tree-ssa.c (verify_ssa): Likewise.
19064 (execute_update_addresses_taken): Likewise.
19065 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
19067 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19069 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
19071 (post_order_compute): Likewise.
19072 (inverted_post_order_compute): Likewise.
19073 (pre_and_rev_post_order_compute_fn): Likewise.
19075 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19077 * genrecog.c (int_set::int_set): Explicitly construct our
19078 auto_vec base class.
19079 * vec.h (auto_vec::auto_vec): New constructor.
19081 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19083 * bitmap.h (class auto_bitmap): New constructor taking
19084 bitmap_obstack * argument.
19086 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19088 * bitmap.h (class auto_bitmap): Change type of m_bits to
19089 bitmap_head, and adjust ctor / dtor and member operators.
19091 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
19093 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
19094 when returned register mode doesn't match original mode.
19096 2017-05-12 Jeff Law <law@redhat.com>
19097 Jakub Jelinek <jakub@redhat.com>
19099 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
19100 we look for cc setter after the compare-elim changes.
19101 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
19102 within the vector to match what compare-elim now expects.
19103 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
19104 (xorsi3_flags, one_cmplsi2_flags): Likewise.
19106 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
19107 after the compare-elim changes.
19108 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
19109 the vector to match what compare-elim now expects.
19110 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
19111 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
19112 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
19113 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
19114 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
19116 * config/visium/visium.c (single_set_and_flags): Fix where
19117 we look for cc setter after the compare-elim changes.
19118 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
19119 with the vector to match what compare-elim now expects.
19120 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
19121 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
19122 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
19123 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
19124 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
19125 (neg<mode>2_insn_set_overflow): Likewise.
19127 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
19129 PR middle-end/79794
19130 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
19131 maybe_expand_insn call, set ops[0].target. If still set after call,
19132 set alt_rtl. Add extra arg to recursive calls.
19133 (extract_bit_field): Add alt_rtl argument. Pass to
19135 * expmed.h (extract_bit_field): Fix prototype.
19136 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
19137 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
19138 to extract_bit_field_calls.
19139 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
19140 Pass alt_rtl to extract_bit_field calls.
19141 * calls.c (store_unaligned_arguments_into_psuedos)
19142 load_register_parameters): Pass extra NULL to extract_bit_field calls.
19143 * optabs.c (maybe_legitimize_operand): Clear op->target when call
19145 * optabs.h (struct expand_operand): Add target bitfield.
19147 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19149 * compare-elim.c (try_eliminate_compare): Canonicalize
19150 operation with embedded compare to
19151 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
19152 (set (reg) (operation)].
19154 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
19156 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19159 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
19160 cost of adding a carry flag for ADC instruction.
19161 [case MINUS]: Ignore the cost of subtracting a carry flag
19162 for SBB instruction.
19164 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
19166 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
19168 * config/rs6000/bmiintrin.h: New file.
19169 * config/rs6000/bmi2intrin.h: New file.
19170 * config/rs6000/x86intrin.h: New file.
19172 2017-05-12 Jeff Law <law@redhat.com>
19174 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
19177 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
19179 PR middle-end/80707
19180 * tree-cfg.c: Remove cfg edges of unreachable case statements.
19182 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19184 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
19185 early expansion of vector divide builtins.
19186 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
19187 builtins identified as having unsigned arguments.
19189 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19191 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
19192 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
19193 expansion of vector logical operations (and, andc, or, xor,
19196 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19198 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
19199 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
19201 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19203 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
19204 early GIMPLE expansion of vector multiplies.
19206 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19208 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
19209 TARGET_HAVE_MOVT conditional.
19210 (movt splitter): Likewise.
19212 2017-05-12 Richard Biener <rguenther@suse.de>
19214 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
19215 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
19216 Fold all stmts not inplace.
19218 2017-05-12 Richard Biener <rguenther@suse.de>
19220 PR tree-optimization/80713
19221 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
19222 inserted_exprs bit for not removed stmts.
19224 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
19226 PR middle-end/69921
19227 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
19228 parallelized" attribute for parallelized OpenACC kernels.
19229 * omp-offload.c (execute_oacc_device_lower): Use it.
19231 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
19232 Set "oacc kernels" attribute.
19233 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
19234 parameter. Adjust all users.
19235 (oacc_fn_attrib_kernels_p): Remove function.
19236 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
19237 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
19238 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
19239 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
19240 assert "oacc kernels" attribute is set.
19242 2017-05-11 Carl Love <cel@us.ibm.com>
19244 * config/rs6000/rs6000-c: Add support for built-in functions
19245 vector unsigned char vec_popcnt (vector signed char)
19246 vector unsigned char vec_popcnt (vector unsigned char)
19247 vector unsigned short vec_popcnt (vector signed short)
19248 vector unsigned short vec_popcnt (vector unsigned short)
19249 vector unsigned int vec_popcnt (vector signed int)
19250 vector unsigned int vec_popcnt (vector unsigned int)
19251 vector unsigned long long vec_popcnt (vector signed long long)
19252 vector unsigned long long vec_popcnt (vector unsigned long long)
19253 vector signed long long vec_slo (vector signed long long,
19254 vector signed char)
19255 vector signed long long vec_slo (vector signed long long,
19256 vector unsigned char)
19257 vector unsigned long long vec_slo (vector unsigned long long,
19258 vector signed char)
19259 vector unsigned long long vec_slo (vector unsigned long long,
19260 vector unsigned char)
19261 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
19262 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
19263 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
19264 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
19265 * doc/extend.texi: Update the built-in documentation file for the
19266 new built-in functions.
19268 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
19270 * attribs.h (sorted_attr_string): Move machine independent
19271 functions for target clone support from the i386 port to common
19272 code. Rename ix86_function_versions to common_function_versions.
19273 Rename make_name to make_unique_name.
19274 (common_function_versions): Likewise.
19275 (make_unique_name): Likewise.
19276 (make_dispatcher_decl): Likewise.
19277 (is_function_default_version): Likewise.
19278 * attribs.c (attr_strcmp): Likewise.
19279 (sorted_attr_string): Likewise.
19280 (common_function_versions): Likewise.
19281 (make_unique_name): Likewise.
19282 (make_dispatcher_decl): Likewise.
19283 (is_function_default_version): Likewise.
19284 * config/i386/i386.c (attr_strcmp): Likewise.
19285 (sorted_attr_string): Likewise.
19286 (ix86_function_versions): Likewise.
19287 (make_name): Likewise.
19288 (make_dispatcher_decl): Likewise.
19289 (is_function_default_version): Likewise.
19290 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
19292 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19295 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
19296 Account for direct move costs for vec_construct of integer
19299 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
19302 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
19303 (UNSPEC_STX_ATOMIC): Ditto.
19304 (loaddi_via_sse): New insn.
19305 (storedi_via_sse): Ditto.
19306 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
19307 Update corresponding peephole2 patterns.
19308 (atomic_storedi_fpu): Ditto.
19310 2017-05-11 Julia Koval <julia.koval@intel.com>
19312 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
19313 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
19315 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
19316 (__builtin_ia32_rsqrt14ss_mask): New builtins.
19317 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
19319 2017-05-11 Nathan Sidwell <nathan@acm.org>
19321 * graphite-poly.c: Include dumpfile.h.
19323 * dumpfle.h (dump_function): Declare here ...
19324 * tree-dump.h (dump_function): ... not here.
19325 * dumpfile.c: #include tree-cfg.h.
19326 (dump_function): Move here from ...
19327 * tree-dump.c (dump_function): ... here.
19328 * gimplify.c: #include splay-tree.h, not tree-dump.h.
19329 * graphite-poly.c: Don't include tree-dump.h.
19330 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
19331 * print-tree.c: Likewise.
19332 * stor-layout.c: Likewise.
19333 * tree-nested.c: Likewise.
19335 * dumpfile.c (dump_start): Use TDF_FLAGS.
19336 (dump_enable_all): Fix TDF_KIND check thinko.
19338 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
19340 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19341 array entries to represent two legal parameterizations of the
19342 overloaded __builtin_cmpb function, as represented by the
19343 P6_OV_BUILTIN_CMPB constant.
19344 (altivec_resolve_overloaded_builtin): Add special case handling
19345 for the __builtin_cmpb function, as represented by the
19346 P6_OV_BUILTIN_CMPB constant.
19347 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
19348 (BU_P6_64BIT_2): New macro.
19349 (BU_P6_OVERLOAD_2): New macro
19350 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
19351 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
19352 (CMPB): Add overload support to represent both 32-bit and 64-bit
19353 compare-bytes function.
19354 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
19355 support for TARGET_CMPB.
19356 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
19357 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19358 documentation of the __builtin_cmpb overloaded built-in function.
19360 2017-05-11 Richard Biener <rguenther@suse.de>
19362 PR tree-optimization/80705
19363 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
19364 bases are not vectorizable.
19366 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19368 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
19369 when counting register pressure.
19371 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19373 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
19374 (iv_ca_more_deps): Renamed to ...
19375 (iv_ca_compare_deps): ... this.
19376 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
19378 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19380 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
19382 (determine_group_iv_costs): ... here.
19383 (find_inv_vars_cb): Record inv var if it's not recorded before.
19385 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19387 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
19388 (get_shiftadd_cost): Ditto.
19390 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19392 * tree-ssa-address.c: Include header file.
19393 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
19395 (add_to_parts): Refactor.
19396 (addr_to_parts): New parameter. Update use of move_hint_to_base.
19397 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
19400 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19402 PR tree-optimization/53090
19403 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
19405 (extract_cond_operands): Detect condition with IV on both sides
19406 and return COMP_IV_EXPR_2.
19407 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
19408 (rewrite_use_compare): Simplify by removing call to function
19409 extract_cond_operands.
19411 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19413 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
19414 (extract_cond_operands): Detect condition comparing against non-
19415 invariant bound and return appropriate enum value.
19416 (find_interesting_uses_cond): Update use of extract_cond_operands.
19417 Handle its return value accordingly.
19418 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
19420 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19422 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
19423 nonlinear iv_use computation in loop invariant sensitive way.
19425 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19427 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
19428 (find_iv_candidates): Call relate_compare_use_with_all_cands.
19430 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19432 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
19433 (dump_cand): Support iv_cand.inv_exprs.
19434 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
19436 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
19439 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19441 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
19443 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
19444 as local function. Include necessary header files.
19445 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
19447 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19449 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
19451 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19453 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
19454 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
19455 RSHIFT_EXPR and BIT_NOT_EXPR.
19457 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19459 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
19460 (adjust_setup_cost): New parameter supporting round up adjustment.
19461 (struct address_cost_data): Delete.
19462 (force_expr_to_var_cost): Don't bound cost with spill_cost.
19463 (split_address_cost, ptr_difference_cost): Delete.
19464 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
19465 (struct ainc_cost_data): New struct.
19466 (get_address_cost_ainc): New function.
19467 (get_address_cost, get_computation_cost): Reimplement.
19468 (determine_group_iv_cost_address): Record inv_expr for all uses of
19470 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
19471 (iv_ca_has_deps): Reimplemented to ...
19472 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
19474 (iv_ca_extend): Call iv_ca_more_deps.
19476 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19478 * tree-ssa-address.c (struct mem_address): Move to header file.
19479 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
19480 * tree-ssa-address.h (struct mem_address): Move from C file.
19481 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
19483 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19485 * tree-affine.h (aff_combination_type): New interface.
19486 (aff_combination_zero_p): Remove static.
19487 (aff_combination_const_p): New interface.
19488 (aff_combination_singleton_var_p): New interfaces.
19490 2017-05-11 Richard Biener <rguenther@suse.de>
19492 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
19493 Skip unreachable blocks and destinations.
19494 (eliminate): Move stmt removal and fixup ...
19495 (fini_eliminate): ... here. Skip inserted exprs.
19496 (pass_pre::execute): Move fini_pre after fini_eliminate.
19497 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
19498 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
19499 PRE to get rid of dead code that has invalid SSA form and
19500 split critical edges again.
19502 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19504 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19506 2017-05-11 Richard Biener <rguenther@suse.de>
19508 * passes.c (execute_function_todo): Verify loops if they are
19509 said to be up-to-date.
19510 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
19511 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
19513 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
19516 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
19517 handle calling assemble_external ourself.
19520 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
19521 modes with zero size. Enhance comment.
19523 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19525 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
19526 built-ins for vec_xl and vec_xst with short and char pointer
19529 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
19531 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
19532 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
19533 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
19534 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
19535 (_mm_maskz_min_round_ss): New intrinsics.
19536 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
19537 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
19538 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
19539 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
19540 (__builtin_ia32_minss_mask_round): New builtins.
19541 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19542 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19543 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
19545 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
19546 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
19548 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
19551 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
19553 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
19554 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
19555 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
19556 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
19557 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
19558 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
19559 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
19560 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19561 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
19562 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
19563 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
19564 (__builtin_ia32_mulss_mask_round): New builtins.
19565 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
19566 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
19567 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
19569 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
19570 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
19572 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
19575 2017-05-10 Julia Koval <julia.koval@intel.com>
19577 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
19578 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
19579 (_mm256_setr_m128i): New intrinsics.
19581 2017-05-10 Julia Koval <julia.koval@intel.com>
19583 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
19584 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
19585 (_mm_maskz_rcp14_ss): New intrinsics.
19586 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
19587 (__builtin_ia32_rcp14ss_mask): New builtins.
19588 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
19590 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
19592 PR tree-optimization/51513
19593 * tree-cfg.c (gimple_seq_unreachable_p): New function.
19594 (assert_unreachable_fallthru_edge_p): Use it.
19595 (group_case_labels_stmt): Likewise.
19596 * tree-cfg.h: Prototype it.
19597 * stmt.c: Include cfghooks.h and tree-cfg.h.
19598 (emit_case_dispatch_table) <gap_label>: New local variable.
19599 Use it to fill dispatch table gaps.
19600 Test for default_label before updating probabilities.
19601 (expand_case) <default_label>: Remove unneeded initialization.
19602 Test for unreachable default case statement and remove its edge.
19603 Set default_label accordingly.
19604 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
19606 2017-05-10 Carl Love <cel@us.ibm.com>
19608 * config/rs6000/rs6000-c: Add support for built-in functions
19609 vector signed char vec_neg (vector signed char)
19610 vector signed short int vec_neg (vector short int)
19611 vector signed int vec_neg (vector signed int)
19612 vector signed long long vec_neg (vector signed long long)
19613 vector float vec_neg (vector float)
19614 vector double vec_neg (vector double)
19615 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
19617 * config/rs6000/altivec.h: Add define for vec_neg
19618 * doc/extend.texi: Update the built-in documentation for the
19619 new built-in functions.
19621 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19623 PR tree-optimization/77644
19624 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
19626 2017-05-10 Nathan Sidwell <nathan@acm.org>
19628 * dumpfile.h (TDI_lang_all): New.
19629 (TDF_KIND): New. Renumber others
19630 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
19632 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
19634 (get_dump_file_name): Adjust suffix generation.
19635 (dump_enable_all): Use TDF_KIND.
19636 * doc/invoke.texi (-fdump-lang-all): Document.
19638 * dumpfile.h: Tabify.
19640 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
19643 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
19644 Move member access before delete.
19646 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
19648 * tree-inline.c (expand_call_inline): Split block at stmt
19651 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
19654 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
19655 are now unused after splitting mov{sf,sd}_hardfloat.
19656 (f32_lr2): Likewise.
19657 (f32_lm): Likewise.
19658 (f32_lm2): Likewise.
19659 (f32_li): Likewise.
19660 (f32_li2): Likewise.
19661 (f32_lv): Likewise.
19662 (f32_sr): Likewise.
19663 (f32_sr2): Likewise.
19664 (f32_sm): Likewise.
19665 (f32_sm2): Likewise.
19666 (f32_si): Likewise.
19667 (f32_si2): Likewise.
19668 (f32_sv): Likewise.
19669 (f32_dm): Likewise.
19670 (f32_vsx): Likewise.
19671 (f32_av): Likewise.
19672 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
19673 For movsf, order stores so the VSX stores occur before the GPR
19674 store which encourages the register allocator to use a traditional
19675 FPR instead of a GPR. For movsd, order the stores so that the GPR
19676 store comes before the VSX stores to allow the power6 to work.
19677 This is due to the power6 not having a 32-bit integer store
19678 instruction from a FPR.
19679 (movsf_hardfloat): Likewise.
19680 (movsd_hardfloat): Likewise.
19682 2017-05-09 Martin Sebor <msebor@redhat.com>
19684 PR translation/80280
19685 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
19688 PR translation/80280
19689 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
19690 data member added in r247778.
19691 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
19693 2017-05-09 Nathan Sidwell <nathan@acm.org>
19695 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
19697 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
19700 2017-05-09 Marek Polacek <polacek@redhat.com>
19702 * doc/invoke.texi: Fix typo.
19704 2017-05-09 Richard Biener <rguenther@suse.de>
19706 * tree-vrp.c (vrp_val_is_max): Adjust comment.
19707 (vrp_val_is_min): Likewise.
19708 (set_value_range_to_value): Likewise.
19709 (set_value_range_to_nonnegative): Likewise.
19710 (gimple_assign_nonzero_p): Likewise.
19711 (gimple_stmt_nonzero_p): Likewise.
19712 (vrp_int_const_binop): Likewise. Remove unreachable case.
19713 (adjust_range_with_scev): Adjust comments.
19714 (compare_range_with_value): Likewise.
19715 (extract_range_from_phi_node): Likewise.
19716 (test_for_singularity): Likewise.
19718 2017-05-09 Richard Biener <rguenther@suse.de>
19720 * tree-vrp.c (get_single_symbol): Add assert that we don't
19721 get overflowed constants as invariant part.
19722 (compare_values_warnv): Add comment before the TREE_NO_WARNING
19723 checks. Use wi::cmp instead of recursing for integer constants.
19724 (compare_values): Just ignore whether we assumed undefined
19725 overflow instead of failing the compare.
19726 (extract_range_for_var_from_comparison_expr): Add comment before the
19727 TREE_NO_WARNING sets.
19728 (test_for_singularity): Likewise.
19729 (extract_range_from_comparison): Do not disable optimization
19730 when we assumed undefined overflow.
19731 (extract_range_basic): Remove init of unused var.
19733 2017-05-09 Richard Biener <rguenther@suse.de>
19735 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
19736 (extract_range_from_multiplicative_op_1): Adjust.
19737 (extract_range_from_binary_expr_1): Use int_const_binop.
19739 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
19742 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
19743 rs6000_store_data_bypass_p in seven define_bypass directives and
19744 in several comments.
19745 * config/rs6000/rs6000-protos.h: Add prototype for
19746 rs6000_store_data_bypass_p function.
19747 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
19748 function implements slightly different (rs6000-specific) semantics
19749 than store_data_bypass_p, returning false rather than aborting
19750 with assertion error when arguments do not satisfy the
19751 requirements of store data bypass.
19752 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
19753 rs6000_store_data_bypass_p.
19755 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
19757 * config/xtensa/xtensa-protos.h
19758 (xtensa_initial_elimination_offset): New declaration.
19759 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
19760 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
19761 macro definition, add case for FRAME_POINTER_REGNUM when
19762 FRAME_GROWS_DOWNWARD.
19763 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
19764 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
19765 xtensa_initial_elimination_offset.
19767 2017-05-08 Nathan Sidwell <nathan@acm.org>
19769 * doc/invoke.texi: Alphabetize -fdump options.
19771 2017-05-08 Martin Sebor <msebor@redhat.com>
19773 PR translation/80280
19774 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
19776 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
19778 * target.def (compute_frame_layout): New optional target hook.
19779 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
19780 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
19781 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
19783 * reload1.c (verify_initial_elim_offsets): Likewise.
19784 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
19785 (use_simple_return_p): Call arm_compute_frame_layout if needed.
19786 (arm_get_frame_offsets): Split up into this ...
19787 (arm_compute_frame_layout): ... and this function.
19789 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
19791 * config/aarch64/constraints.md (Usa): New constraint.
19792 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
19794 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
19796 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
19797 with_multilib_list after it has been checked.
19799 2017-05-08 Richard Biener <rguenther@suse.de>
19801 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
19802 (bitmap_set_subtract_values): Likewise.
19804 2017-05-08 Richard Biener <rguenther@suse.de>
19806 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
19807 (gimple_assign_nonzero): ... this and remove strict_overflow_p
19809 (gimple_stmt_nonzero_warnv_p): Rename to ...
19810 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
19812 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
19813 (extract_range_basic): Adjust, do not disable propagation on
19814 strict overflow sensitive simplification.
19815 (vrp_visit_cond_stmt): Likewise.
19817 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
19819 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
19820 body size unconditionally.
19822 2017-05-07 Jeff Law <law@redhat.com>
19825 2017-05-06 Jeff Law <law@redhat.com>
19826 PR tree-optimization/78496
19827 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19830 PR tree-optimization/78496
19831 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19832 (simplify_stmt_using_ranges): Call it.
19833 (vrp_dom_walker::before_dom_children): Extract equivalences
19834 from an ASSERT_EXPR with an equality comparison against a
19837 2017-05-06 Jeff Law <law@redhat.com>
19839 PR tree-optimization/78496
19840 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
19843 PR tree-optimization/78496
19844 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
19845 (simplify_stmt_using_ranges): Call it.
19846 (vrp_dom_walker::before_dom_children): Extract equivalences
19847 from an ASSERT_EXPR with an equality comparison against a
19850 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19852 * lra-constraints.c (lra_copy_reg_equiv): New function.
19853 (split_reg): Use it to copy equivalence information from the
19854 original register to the spill register.
19856 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
19858 PR rtl-optimization/75964
19859 * simplify-rtx.c (simplify_const_relational_operation): Remove
19860 invalid handling of comparisons of integer ABS.
19862 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
19864 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
19865 initialize to zero.
19866 (init_regs): Remove declaration.
19867 (function_arg_advance_32): Initialize error_p as boolean variable.
19869 2017-05-05 Nathan Sidwell <nathan@acm.org>
19871 * store-motion.c (remove_reachable_equiv_notes): Reformat long
19872 lines. Use for (;;).
19874 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19876 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
19877 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
19878 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
19879 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
19880 VF=2 that require versioning.
19882 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19884 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
19887 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19889 * diagnostic.h (diagnostic_override_option_index): Convert from
19890 macro to inline function.
19892 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19894 * diagnostic.c (last_module_changed_p): New function.
19895 (set_last_module): New function.
19896 (diagnostic_report_current_module): Convert macro usage to
19897 the above functions.
19898 * diagnostic.h (diagnostic_context::last_module): Strengthen
19899 from const line_map * to const line_map_ordinary *.
19900 (diagnostic_last_module_changed): Delete macro.
19901 (diagnostic_set_last_module): Delete macro.
19903 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19905 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
19906 with diagnostic_report_diagnostic.
19907 (diagnostic_n_impl_richloc): Likewise.
19908 * diagnostic.h (report_diagnostic): Delete macro.
19909 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
19910 with diagnostic_report_diagnostic.
19911 * substring-locations.c (format_warning_va): Likewise.
19913 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19915 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
19916 save/restor of format_spec. Move option-printing code to...
19917 (print_option_information): ...this new function, and
19918 reimplement by simply printing to the pretty_printer,
19919 rather than appending to the format string.
19921 2017-05-05 David Malcolm <dmalcolm@redhat.com>
19923 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
19924 handling logic into...
19925 (update_effective_level_from_pragmas): ...this new function.
19927 2017-05-04 Andrew Waterman <andrew@sifive.com>
19929 * config/riscv/riscv.opt (mstrict-align): New option.
19930 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
19931 (SLOW_UNALIGNED_ACCESS): Define.
19932 (riscv_slow_unaligned_access): Declare.
19933 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
19935 (riscv_slow_unaligned_access): New variable.
19936 (rocket_tune_info): Set slow_unaligned_access to true.
19937 (optimize_size_tune_info): Set slow_unaligned_access to false.
19938 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
19939 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
19940 (riscv_option_override): Set riscv_slow_unaligned_access.
19941 * doc/invoke.texi: Add -mstrict-align to RISC-V.
19943 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
19945 * config/riscv/riscv.md: Unify indentation.
19947 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19952 * config/rs6000/rs6000.md (u code attribute): Add FIX and
19954 (extendsi<mode>2): Add support for doing sign extension via
19955 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
19956 don't have ISA 3.0 instructions.
19957 (extendsi<mode>2 splitter): Likewise.
19958 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
19959 generate the normal insns since SImode can now go in vector
19960 registers. Disallow the special UNSPECs needed for previous
19961 machines to hide SImode being used. Add new insns
19962 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
19963 (fix_trunc<mode>si2_stfiwx): Likewise.
19964 (fix_trunc<mode>si2_internal): Likewise.
19965 (fixuns_trunc<mode>si2): Likewise.
19966 (fixuns_trunc<mode>si2_stfiwx): Likewise.
19967 (fctiw<u>z_<mode>_smallint): Likewise.
19968 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
19969 of floating point to 32-bit integer from doing a direct move to
19970 the GPR registers to do a store.
19971 (fctiwz_<mode>): Break long line.
19973 2017-05-05 Bin Cheng <bin.cheng@arm.com>
19975 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
19976 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
19977 (addr_list, addr_offset_valid_p): New.
19978 (split_address_groups): Check offset validity with above function.
19979 (gt-tree-ssa-loop-ivopts.h): Include header file.
19981 2017-05-05 Nathan Sidwell <nathan@acm.org>
19983 * config.gcc (arm*-*-*): Add missing 'fi'.
19985 2017-05-05 Steve Ellcey <sellcey@cavium.com>
19987 * doc/invoke.texi (-fopt-info): Explicitly say order of options
19988 included in -fopt-info does not matter.
19989 * doc/optinfo.texi (-fopt-info): Fix description of default
19990 behavour. Explicitly say order of options included in -fopt-info
19993 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19995 * config.gcc: Allow combinations of aprofile and rmprofile values for
19996 --with-multilib-list.
19997 * config/arm/t-multilib: New file.
19998 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
19999 variables. Remove setting of ISA and floating-point ABI in
20000 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
20001 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
20002 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
20003 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
20005 * config/arm/t-rmprofile: Likewise except for the matches changes.
20006 * doc/install.texi (--with-multilib-list): Document the combination of
20007 aprofile and rmprofile values and warn about pitfalls in doing that.
20009 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
20011 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
20012 (movdi_aarch64): Likewise.
20014 2017-05-05 Jakub Jelinek <jakub@redhat.com>
20016 PR tree-optimization/80632
20017 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
20019 (build_arrays): Initialize it for virtual phis.
20020 (fix_phi_nodes): Use it for virtual phis.
20022 PR tree-optimization/80558
20023 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
20024 [x, y] op z into [x op, y op z] for op & or | if conditions
20027 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
20028 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20031 * config/arm/arm.md (use_literal_pool): Remove.
20032 (64-bit immediate split): No longer takes cost into consideration
20033 if arm_disable_literal_pool is enabled.
20034 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
20035 used when arm_disable_literal_pool is enabled.
20036 (arm_max_const_double_inline_cost): Remove use of
20037 arm_disable_literal_pool.
20038 (push_minipool_fix): Add assert.
20039 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
20040 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
20041 (no_literal_pool_sf_immediate): New.
20043 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20045 PR tree-optimization/80613
20046 * tree-ssa-dce.c (propagate_necessity): Remove cases for
20047 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
20049 2017-05-05 Richard Biener <rguenther@suse.de>
20051 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
20053 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
20055 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
20056 of this flag from insn conditions due to removal from r247495.
20058 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
20060 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
20062 (arm_early_store_addr_dep_ptr): Likewise.
20063 * config/arm/aarch-common-protos.h
20064 (arm_early_load_addr_dep_ptr): Add prototype.
20065 (arm_early_store_addr_dep_ptr): Likewise.
20066 * config/arm/cortex-a53.md: Add new bypasses.
20068 2017-05-05 Jakub Jelinek <jakub@redhat.com>
20070 * tree.c (next_type_uid): Change type to unsigned.
20071 (type_hash_canon): Decrement back next_type_uid if
20072 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
20073 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
20076 2017-05-04 Martin Sebor <msebor@redhat.com>
20078 * builtins.c: Fix a trivial typo in a comment.
20080 PR middle-end/79234
20081 * builtins.c (check_sizes): Adjust to handle reading past the end.
20082 Avoid printing excessive upper bound of ranges. Use %E to print
20083 tree nodes instead of converting them to %wu.
20084 (expand_builtin_memchr): New function.
20085 (compute_dest_size): Rename...
20086 (compute_objsize): ...to this.
20087 (expand_builtin_memcpy): Adjust.
20088 (expand_builtin_mempcpy): Adjust.
20089 (expand_builtin_strcat): Adjust.
20090 (expand_builtin_strcpy): Adjust.
20091 (check_strncat_sizes): Adjust.
20092 (expand_builtin_strncat): Adjust.
20093 (expand_builtin_strncpy): Adjust and simplify.
20094 (expand_builtin_memset): Adjust.
20095 (expand_builtin_bzero): Adjust.
20096 (expand_builtin_memcmp): Adjust.
20097 (expand_builtin): Handle memcmp.
20098 (maybe_emit_chk_warning): Check strncat just once.
20100 2017-05-04 Martin Sebor <msebor@redhat.com>
20102 PR preprocessor/79214
20103 PR middle-end/79222
20104 PR middle-end/79223
20105 * builtins.c (check_sizes): Add inlining context and issue
20106 warnings even when -Wno-system-headers is set.
20107 (check_strncat_sizes): Same.
20108 (expand_builtin_strncat): Same.
20109 (expand_builtin_memmove): New function.
20110 (expand_builtin_stpncpy): Same.
20111 (expand_builtin): Handle memmove and stpncpy.
20113 2017-05-04 Bin Cheng <bin.cheng@arm.com>
20115 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
20116 which is not used any more.
20118 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20120 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
20122 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20124 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
20125 (cortexa53_tunings): Likewise.
20126 (cortexa57_tunings): Likewise.
20127 (cortexa72_tunings): Likewise.
20128 (cortexa73_tunings): Likewise.
20130 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20132 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
20133 Set loop alignment to 8.
20135 2017-05-04 Martin Sebor <msebor@redhat.com>
20137 PR translation/80280
20138 * builtins.c (expand_builtin_object_size): Add missing quoting to
20139 %D and like directives.
20140 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
20141 (hsa_type_for_tree_type): Same.
20142 (verify_function_arguments): Same.
20143 * symtab.c (symbol_table::change_decl_assembler_name): Same.
20144 * varasm.c (get_section): Same.
20147 2017-05-04 Martin Sebor <msebor@redhat.com>
20149 PR translation/80280
20150 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
20152 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20154 * config/aarch64/aarch64.c (generic_addrcost_table):
20155 Change HI/TI mode setting.
20157 2017-05-04 Martin Jambor <mjambor@suse.cz>
20159 PR tree-optimization/80622
20160 * tree-sra.c (comes_initialized_p): New function.
20161 (build_accesses_from_assign): Only set write lazily when
20162 comes_initialized_p is false.
20163 (analyze_access_subtree): Use comes_initialized_p.
20164 (propagate_subaccesses_across_link): Assert !comes_initialized_p
20165 instead of testing for PARM_DECL.
20167 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20169 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
20170 constraint on operand 0 to allow more general addressing modes.
20171 Adjust output template.
20172 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
20174 * config/aarch64/aarch64-protos.h
20175 (aarch64_address_valid_for_prefetch_p): Declare prototype.
20176 * config/aarch64/constraints.md (Dp): New address constraint.
20177 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
20180 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
20182 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
20183 update use of estimate_ipcp_clone_size_and_time.
20184 (estimate_local_effects): Update use of
20185 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
20186 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
20187 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
20188 Return nonspecialized time.
20190 2017-05-04 Richard Biener <rguenther@suse.de>
20192 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
20193 for the last VUSE which def dominates the PHI. Directly call
20195 (get_continuation_for_phi_1): Remove.
20197 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
20199 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
20200 to explain the use of truncating division. Cap the number of
20201 iterations to the maximum given by nb_iterations_upper_bound,
20204 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20206 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
20207 * configure: Regenerate.
20208 * config.in: Regenerate.
20209 * config/i386/driver-mingw32.c: new file.
20210 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
20211 * config.host: Link driver-mingw32.o on MinGW host.
20212 * doc/install.texi: Document new --enable-mingw-wildcard configure
20215 2017-05-04 Marek Polacek <polacek@redhat.com>
20217 PR tree-optimization/80612
20218 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
20220 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20221 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
20223 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
20224 (movt splitter): Likewise.
20225 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
20226 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
20227 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
20228 block for Thumb-1 with MOVT.
20229 (thumb2_legitimate_address_p): Move code block ...
20230 (can_avoid_literal_pool_for_label_p): ... into this new function.
20231 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
20233 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
20234 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
20235 "M-profile targets with the MOVT instruction".
20237 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20239 * config/arm/arm-builtins.c (arm_init_builtins): Rename
20240 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
20241 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
20243 2017-05-04 Martin Liska <mliska@suse.cz>
20245 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
20246 variable cond_code.
20248 2017-05-04 Richard Biener <rguenther@suse.de>
20250 * tree.c (array_at_struct_end_p): Handle arrays at struct
20251 end with flexarrays more conservatively. Refactor and treat
20252 arrays of arrays or aggregates more strict. Fix
20253 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
20254 * tree.c (array_at_struct_end_p): Adjust prototype.
20255 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
20256 * gimple-fold.c (get_range_strlen): Likewise.
20257 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
20259 2017-05-04 Richard Biener <rguenther@suse.de>
20261 PR tree-optimization/31130
20262 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
20264 (supports_overflow_infinity): Likewise.
20265 (is_negative_overflow_infinity): Likewise.
20266 (is_positive_overflow_infinity): Likewise.
20267 (is_overflow_infinity): Likewise.
20268 (stmt_overflow_infinity): Likewise.
20269 (overflow_infinity_range_p): Likewise.
20270 (usable_range_p): Remove as always returning true.
20271 (make_overflow_infinity): Remove.
20272 (negative_overflow_infinity): Likewise.
20273 (positive_overflow_infinity): Likewise.
20274 (avoid_overflow_infinity): Likewise.
20275 (set_value_range): Adjust accordingly.
20276 (set_value_range_to_nonnegative): Likewise, remove now unused
20277 overflow_infinity arg.
20278 (vrp_operand_equal_p): Adjust.
20279 (update_value_range): Likewise.
20280 (range_int_cst_singleton_p): Likewise.
20281 (operand_less_p): Likewise.
20282 (compare_values_warnv): Likewise.
20283 (extract_range_for_var_from_comparison_expr): Likewise.
20284 (vrp_int_const_binop): Likewise.
20285 (zero_nonzero_bits_from_vr): Likewise.
20286 (extract_range_from_multiplicative_op_1): Likewise.
20287 (extract_range_from_binary_expr_1): Likewise.
20288 (extract_range_from_unary_expr): Likewise.
20289 (extract_range_from_comparison): Likewise.
20290 (extract_range_basic): Likewise.
20291 (adjust_range_with_scev): Likewise.
20292 (compare_ranges): Likewise.
20293 (compare_range_with_value): Likewise.
20294 (dump_value_range): Likewise.
20295 (test_for_singularity): Likewise, remove strict_overflow_p parameter
20297 (simplify_cond_using_ranges): Adjust.
20299 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20301 * brig-builtins.def: Added a builtin for class_f64.
20302 * builtin-types.def: Added a builtin type needed by class_f64.
20304 2017-05-03 Jason Merrill <jason@redhat.com>
20306 * timevar.def: Add TV_CONSTEXPR.
20308 2017-05-03 David Malcolm <dmalcolm@redhat.com>
20310 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
20312 2017-05-03 Martin Jambor <mjambor@suse.cz>
20314 * ipa-prop.c (ipa_update_after_lto_read): Removed.
20315 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
20316 * ipa-cp.c (ipcp_propagate_stage): Do not call
20317 ipa_update_after_lto_read.
20318 * ipa-inline.c (ipa_inline): Likewise.
20320 2017-05-03 Martin Jambor <mjambor@suse.cz>
20322 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
20323 tag. Added a default constructor and a destructor.
20324 (ipa_edge_args_sum_t): New class;
20325 (ipa_edge_args_sum): Declare.
20326 (ipa_edge_args_vector): Remove declaration.
20327 (IPA_EDGE_REF): Use ipa_edge_args_sum.
20328 (ipa_free_edge_args_substructures): Remove declaration.
20329 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
20330 (ipa_edge_args_info_available_for_edge_p): Likewise.
20331 * ipa-prop.c (ipa_edge_args_vector): Removed.
20332 (edge_removal_hook_holder): Likewise.
20333 (edge_duplication_hook_holder): Likewise.
20334 (ipa_edge_args_sum): New variable.
20335 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
20336 ipa_edge_args_vector.
20337 (ipa_free_edge_args_substructures): Likewise.
20338 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
20339 ipa_edge_args_vector.
20340 (ipa_edge_removal_hook): Turned into method
20341 ipa_edge_args_sum_t::remove.
20342 (ipa_edge_duplication_hook): Turned into method
20343 ipa_edge_args_sum_t::duplicate.
20344 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
20345 registering edge hooks.
20346 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
20347 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
20348 ipa_edge_args_sum instead of ipa_edge_args_vector.
20349 * ipa-profile.c (ipa_profile): Likewise.
20351 2017-05-03 Martin Jambor <mjambor@suse.cz>
20353 * symbol-summary.h (function_summary): New method exists.
20354 (function_summary::symtab_removal): Deallocate through release.
20355 (call_summary): New class.
20356 (gt_ggc_mx): New overload.
20357 (gt_pch_nx): Likewise.
20358 (gt_pch_nx): Likewise.
20360 2017-05-03 Jeff Law <law@redhat.com>
20362 PR tree-optimization/78496
20363 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
20364 from simplify_cond_using_ranges. Split off code to walk
20365 backwards through casts into ...
20366 (simplify_cond_using_ranges_2): New function.
20367 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
20368 (execute_vrp): After identifying jump threads, call
20369 simplify_cond_using_ranges_2.
20371 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
20374 * ipa-inline.h (inline_summary): Add ctor.
20375 (create_ggc): Do not use ggc_cleared_alloc.
20377 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
20378 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20380 * gcc.c (handle_braces): Support escaping in switch matching
20382 * doc/invoke.texi (Spec Files): Document it.
20383 Remove superfluous @code markup in items.
20385 2017-05-03 David Malcolm <dmalcolm@redhat.com>
20387 * diagnostic-show-locus.c (struct column_range): New struct.
20388 (get_affected_columns): New function.
20389 (get_printed_columns): New function.
20390 (struct correction): New struct.
20391 (correction::ensure_capacity): New function.
20392 (correction::ensure_terminated): New function.
20393 (struct line_corrections): New struct.
20394 (line_corrections::~line_corrections): New dtor.
20395 (line_corrections::add_hint): New function.
20396 (layout::print_trailing_fixits): Reimplement in terms of the new
20398 (selftest::test_overlapped_fixit_printing): New function.
20399 (selftest::diagnostic_show_locus_c_tests): Call it.
20401 2017-05-03 Nathan Sidwell <nathan@acm.org>
20403 Canonicalize canonical type hashing
20404 * tree.h (type_hash_canon_hash): Declare.
20405 * tree.c (type_hash_list, attribute_hash_list): Move into
20406 type_hash_canon_hash.
20407 (build_type_attribute_qual_variant): Break out hash code calc into
20408 type_hash_canon_hash.
20409 (type_hash_canon_hash): New. Generic type hash computation.
20410 (build_range_type_1, build_array_type_1, build_function_type,
20411 build_method_type_directly, build_offset_type, build_complex_type,
20412 make_vector_type): Call it.
20414 2017-05-03 Richard Biener <rguenther@suse.de>
20416 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20417 When all DRs have unknown misaligned do not always peel
20418 when there is a store but apply the same costing model as if
20419 there were only loads.
20421 2017-05-03 Richard Biener <rguenther@suse.de>
20424 PR tree-optimization/80492
20425 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20426 compare_base_decls returning dont-know properly.
20428 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
20430 * config/arm/iterators.md (CCSI): New mode iterator.
20431 (arch): New mode attribute.
20432 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
20433 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
20434 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
20435 code iterator for success result mode.
20436 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
20437 the corresponding new insn generators.
20439 2017-05-03 Bin Cheng <bin.cheng@arm.com>
20442 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20443 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20445 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
20447 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
20448 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
20449 (DDR_A): Wrap DDR argument in brackets.
20450 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
20451 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
20452 (DDR_REVERSED_P): Likewise.
20454 2017-05-03 Jakub Jelinek <jakub@redhat.com>
20456 PR tree-optimization/79472
20457 * tree-switch-conversion.c (struct switch_conv_info): Add
20458 contiguous_range and default_case_nonstandard fields.
20459 (collect_switch_conv_info): Compute contiguous_range and
20460 default_case_nonstandard fields, don't clear final_bb if
20461 contiguous_range and only the default case doesn't have the required
20463 (check_all_empty_except_final): Set default_case_nonstandard instead
20464 of failing if contiguous_range and the default case doesn't have empty
20466 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
20467 and only the default case doesn't have the required constants. Skip
20469 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
20470 if default_case_nonstandard.
20471 (build_constructors): Build constant 1 just once. Assert that default
20472 values aren't inserted in between cases if contiguous_range. Skip
20474 (build_arrays): Skip virtual phis.
20475 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
20476 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
20477 Handle virtual phis.
20478 (gen_inbound_check): Handle default_case_nonstandard case.
20479 (process_switch): Adjust check_final_bb caller. Call
20480 gather_default_values with the first non-default case instead of
20481 default case if default_case_nonstandard.
20483 2017-05-02 Nathan Sidwell <nathan@acm.org>
20485 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
20486 check. Fix formatting.
20488 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
20490 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
20491 errors when comparing specialized and unspecialized times.
20493 2017-05-02 David Malcolm <dmalcolm@redhat.com>
20495 * diagnostic-show-locus.c
20496 (layout::should_print_annotation_line_p): Make private.
20497 (layout::print_annotation_line): Make private.
20498 (layout::annotation_line_showed_range_p): Make private.
20499 (layout::show_ruler): Make private.
20500 (layout::print_source_line): Make private. Pass in line and
20501 line_width, rather than calling location_get_source_line. Drop
20503 (layout::print_leading_fixits): New method.
20504 (layout::print_any_fixits): Rename to...
20505 (layout::print_trailing_fixits): ...this, and make private.
20506 Don't print newline fixits.
20507 (diagnostic_show_locus): Move logic for printing one row into...
20508 (layout::print_line): ...this new function. Move the
20509 location_get_source_line call and error-handling from
20510 print_source_line to here. Call print_leading_fixits, and rename
20511 print_any_fixits to print_trailing_fixits.
20512 (selftest::test_fixit_insert_containing_newline): Update now that
20513 newlines are partially supported.
20514 (selftest::test_fixit_insert_containing_newline_2): New test.
20515 (selftest::test_fixit_replace_containing_newline): Update comments.
20516 (selftest::diagnostic_show_locus_c_tests): Call the new test.
20517 * edit-context.c (class added_line): New class.
20518 (class edited_line): Describe newline handling in comment.
20519 (edited_line::actually_edited_p): New method.
20520 (edited_line::print_content): Delete redundant decl.
20521 (edited_line::m_predecessors): New field.
20522 (edited_file::print_content): Call edited_line::print_content.
20523 (edited_file::print_diff): Update to support newlines.
20524 (edited_file::print_diff_hunk): Likewise.
20525 (edited_file::print_run_of_changed_lines): New function.
20526 (edited_file::print_diff_line): Convert to...
20527 (print_diff_line): ...this.
20528 (edited_file::get_effective_line_count): New function.
20529 (edited_line::edited_line): Initialize new field m_predecessors.
20530 (edited_line::~edited_line): Clean up m_predecessors.
20531 (edited_line::apply_fixit): Handle newlines.
20532 (edited_line::get_effective_line_count): New function.
20533 (edited_line::print_content): New function.
20534 (edited_line::print_diff_lines): New function.
20535 (selftest::test_applying_fixits_insert_containing_newline): New
20537 (selftest::test_applying_fixits_replace_containing_newline): New
20539 (selftest::insert_line): New function.
20540 (selftest::test_applying_fixits_multiple_lines): Add example of
20542 (selftest::edit_context_c_tests): Call the new tests.
20544 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20546 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
20547 parameter cand. Update dump information.
20548 (get_computation_cost): Update uses.
20550 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20552 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
20553 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
20554 (get_computation_at, rewrite_use_address): Update use of
20555 get_computation_aff.
20557 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20559 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
20560 (get_computation): Delete.
20561 (get_computation_cost): Implement like get_computation_cost_at.
20562 Use get_computation_at.
20563 (get_computation_cost_at): Delete.
20564 (rewrite_use_nonlinear_expr): Use get_computation_at.
20565 (rewrite_use_compare, remove_unused_ivs): Ditto.
20567 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20569 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
20571 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20573 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
20574 (ivopts_global_cost_for_size): Rename parameter and update uses.
20575 (iv_ca_recount_cost): Update uses.
20576 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
20577 candidates seperately in n_invs and n_cands.
20578 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
20580 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20582 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
20583 (find_inv_vars_cb): New.
20584 (find_depends): Renamed to ...
20585 (find_inv_vars): ... this.
20586 (add_candidate_1, force_var_cost): Call find_inv_vars.
20587 (split_address_cost, determine_group_iv_cost_cond): Ditto.
20589 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20591 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
20592 inv_vars. Add inv_exprs.
20593 (struct iv_cand): Rename depends_on to inv_vars.
20594 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
20595 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
20596 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
20597 (dump_cand): Dump inv_vars.
20598 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
20599 (record_invariant, find_depends, add_candidate_1): Ditto.
20600 (set_group_iv_cost, force_var_cost): Ditto.
20601 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
20602 (get_computation_cost_at, get_computation_cost): Ditto.
20603 (determine_group_iv_cost_generic): Ditto.
20604 (determine_group_iv_cost_address): Ditto.
20605 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
20606 (determine_group_iv_costs): Ditto.
20607 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
20608 (iv_ca_set_remove_invariants): Renamed to ...
20609 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
20610 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
20611 (iv_ca_set_add_invariants): Renamed to ...
20612 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
20613 (iv_ca_set_cp): Use iv_ca_set_add_invs.
20614 (iv_ca_has_deps): Support inv_vars and inv_exprs.
20615 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
20616 (create_new_ivs): Remove useless dump.
20618 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20620 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
20622 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
20623 (iv_ca_set_no_cp, create_new_iv): Ditto.
20625 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20627 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20629 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20631 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
20632 function tree_check2.
20634 2017-05-02 Martin Liska <mliska@suse.cz>
20636 * doc/gcov.texi: Add missing preposition.
20637 * gcov.c (function_info::function_info): Properly fill up
20638 all member variables.
20640 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20642 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
20644 2017-05-02 Tamar Christina <tamar.christina@arm.com>
20646 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
20648 2017-05-02 Martin Liska <mliska@suse.cz>
20651 * lto-streamer-in.c (lto_read_tree_1): Remove
20652 LTO_STREAMER_DEBUG.
20653 * lto-streamer.c (struct tree_hash_entry): Likewise.
20654 (struct tree_entry_hasher): Likewise.
20655 (tree_entry_hasher::hash): Likewise.
20656 (tree_entry_hasher::equal): Likewise.
20657 (lto_streamer_init): Likewise.
20658 (lto_orig_address_map): Likewise.
20659 (lto_orig_address_get): Likewise.
20660 (lto_orig_address_remove): Likewise.
20661 * lto-streamer.h: Likewise.
20662 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
20663 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
20665 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
20667 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
20668 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
20669 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
20670 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
20671 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
20672 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
20673 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
20674 (mm_maskz_sub_ss): New intrinsics.
20675 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20676 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20677 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
20678 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
20679 (__builtin_ia32_subss_mask_round): New builtins.
20680 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20681 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20682 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
20684 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
20685 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20687 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20690 2017-05-02 Martin Jambor <mjambor@suse.cz>
20692 PR tree-optimization/78687
20693 * tree-sra.c (access): New field parent.
20694 (process_subtree_disqualification): New function.
20695 (disqualify_candidate): Call it.
20696 (build_accesses_from_assign): Reset write flag if creating an
20698 (build_access_subtree): Fill in parent field and also prpagate
20699 down grp_write flag.
20700 (create_artificial_child_access): New parameter set_grp_write, set
20701 grp_write to its value.
20702 (propagate_subaccesses_across_link): Also propagate grp_write flag
20704 (propagate_all_subaccesses): Push the closest parent back to work
20705 queue if add_access_to_work_queue returned true.
20707 2017-05-02 Richard Biener <rguenther@suse.de>
20709 * common.opt (fstrict-overflow): Alias negative to fwrapv.
20710 * doc/invoke.texi (fstrict-overflow): Remove all traces of
20711 -fstrict-overflow documentation.
20712 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
20713 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
20714 flag_strict_overflow.
20715 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
20716 * lto-opts.c (lto_write_options): Do not stream it.
20717 * lto-wrapper.c (merge_and_complain): Do not handle it.
20718 * opts.c (default_options_table): Do not set -fstrict-overflow.
20719 (finish_options): Likewise do not clear it when sanitizing.
20720 * simplify-rtx.c (simplify_const_relational_operation): Do not
20721 test flag_strict_overflow.
20723 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20725 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
20726 using enabled attribute.
20727 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
20728 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
20729 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
20730 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
20731 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
20732 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
20733 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
20734 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
20735 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
20736 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
20738 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
20740 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
20742 2017-05-02 Richard Biener <rguenther@suse.de>
20744 PR tree-optimization/80591
20746 2017-04-10 Richard Biener <rguenther@suse.de>
20748 * tree-ssa-structalias.c (find_func_aliases): Properly handle
20751 2017-05-02 Richard Biener <rguenther@suse.de>
20753 PR tree-optimization/80549
20754 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
20755 (cleanup_tree_cfg_noloop): Create forwarders to known loop
20756 headers if they do not have a preheader.
20758 2017-05-02 Martin Liska <mliska@suse.cz>
20761 * common.opt: Fix typo.
20762 * doc/invoke.texi: Likewise.
20764 2017-05-01 Jan Beulich <jbeulich@suse.com>
20766 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
20767 swapping, add (x,x,m,x,n) alternative.
20769 2017-05-01 Nathan Sidwell <nathan@acm.org>
20771 * calls.c (combine_pending_stack_adjustment_and_call): Remove
20772 unnecessary unadjusted_alignment check.
20774 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
20777 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
20779 * gimplify.c (gimplify_cilk_detach): New function.
20780 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
20781 * tree-core.h: Document EXPR_CILK_SPAWN.
20782 * tree.h (EXPR_CILK_SPAWN): Define.
20784 2017-05-01 David Malcolm <dmalcolm@redhat.com>
20786 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
20787 to use new fixit_hint representation, using the "replace" logic.
20788 (get_line_span_for_fixit_hint): Likewise.
20789 (layout::print_any_fixits): Likewise.
20790 (selftest::test_one_liner_many_fixits): Rename to...
20791 (selftest::test_one_liner_many_fixits_1): ...this, and update
20792 comment and expected output to reflect that the multiple fix-it
20793 hints are now consolidated into one insertion.
20794 (selftest::test_one_liner_many_fixits_2): New test.
20795 (selftest::test_diagnostic_show_locus_one_liner): Update for
20797 (selftest::test_fixit_consolidation): Update for fix-it API
20799 * diagnostic.c (print_parseable_fixits): Likewise.
20800 * edit-context.c (edited_line::m_line_events): Convert from
20801 auto_vec <line_event *> to auto_vec <line_event>.
20802 (class line_event): Convert from abstract base class to a concrete
20803 class, taking over the role of replace_event.
20804 (class insert_event): Delete.
20805 (class replace_event): Rename to class line_event. Convert to
20807 (edit_context::add_fixits): Reimplement.
20808 (edit_context::apply_insert): Delete.
20809 (edit_context::apply_replace): Rename to...
20810 (edit_context::apply_fixit): ...this. Convert to half-open range.
20811 (edited_file::apply_insert): Delete.
20812 (edited_file::apply_replace): Rename to...
20813 (edited_file::apply_fixit): ...this.
20814 (edited_line::~edited_line): Drop deletion of events.
20815 (edited_line::apply_insert): Delete.
20816 (edited_line::apply_replace): Rename to...
20817 (edited_line::apply_fixit): ...this. Convert to half-open range.
20818 Update for change to type of m_line_events.
20819 * edit-context.h (edit_context::apply_insert): Delete.
20820 (edit_context::apply_replace): Rename to...
20821 (edit_context::apply_fixit): ...this.
20823 2017-05-01 Martin Sebor <msebor@redhat.com>
20825 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
20828 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
20831 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
20832 __get_cpuid_max returns 0.
20833 (__get_cpuid_count): Ditto.
20835 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
20837 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
20838 replacement expression is another instance of one of its arguments.
20840 2017-05-01 Jakub Jelinek <jakub@redhat.com>
20843 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
20844 check for stack push/pop autoinc.
20845 * config/i386/i386.c (ix86_agi_dependent): Return false
20846 if the only reason why modified_in_p returned true is that
20847 addr is SP based and set_insn is a push or pop.
20849 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20851 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
20854 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
20857 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
20858 (account_size_time): Use two predicates - exec_pred and
20860 (evaluate_conditions_for_known_args): Compute both clause and
20862 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
20863 (inline_summary_t::duplicate): Update.
20864 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
20866 (compute_inline_parameters): Likewise.
20867 (estimate_edge_size_and_time): Update caluclation of time.
20868 (estimate_node_size_and_time): Compute both time and nonspecialized
20870 (estimate_ipcp_clone_size_and_time): Update.
20871 (inline_merge_summary): Update.
20872 (do_estimate_edge_time): Update.
20873 (do_estimate_edge_size): Update.
20874 (do_estimate_edge_hints): Update.
20875 (inline_read_section, inline_write_summary): Stream both new predicates.
20876 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
20878 (compute_inlined_call_time): Cleanup.
20879 (big_speedup_p): Update.
20880 (edge_badness): Update.
20881 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
20882 (size_time_entry): Replace predicate by exec_predicate and
20883 nonconst_predicate.
20884 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
20885 (estimate_edge_time): Return also nonspec_time.
20886 (reset_edge_growth_cache): Update.
20888 2017-04-29 Jakub Jelinek <jakub@redhat.com>
20890 PR rtl-optimization/80491
20891 * ifcvt.c (noce_process_if_block): When looking for x setter
20892 with missing else_bb, don't check only the insn right before
20893 cond_earliest, but look for the last insn that x is modified in
20894 within the same bb.
20896 PR rtl-optimization/80491
20897 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
20899 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
20901 PR tree-optimization/80487
20902 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
20904 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20906 PR tree-optimization/79697
20907 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
20908 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
20909 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
20911 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
20912 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
20914 2017-04-28 Martin Sebor <msebor@redhat.com>
20916 PR tree-optimization/80523
20917 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
20918 (init_target_to_host_charmap, target_to_host, target_strtol10): New
20920 (maybe_warn, format_directive, parse_directive): Use new functions.
20921 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
20923 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
20925 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
20927 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
20929 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
20930 target_header_dir): Set correctly.
20931 * configure: Regenerated.
20932 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20933 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20934 instead of SYSTEM_HEADER_DIR.
20936 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
20938 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
20939 (estimate_local_effects): Likewise.
20940 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
20941 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
20942 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
20943 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
20944 do_estimate_edge_time, estimate_edge_time): Likewise.
20945 * ipa-inline-analysis.c (estimate_node_size_and_time,
20946 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
20947 (estimate_time_after_inlining): Remove.
20949 2017-04-28 Martin Liska <mliska@suse.cz>
20951 * doc/gcov.texi: Enhance documentation of gcov.
20953 2017-04-28 Martin Liska <mliska@suse.cz>
20955 * doc/gcov.texi: Sort options in alphabetic order.
20956 * doc/gcov-dump.texi: Likewise.
20957 * doc/gcov-tool.texi: Likewise.
20958 * gcov.c (print_usage): Likewise.
20959 * gcov-dump.c (print_usage): Likewise.
20960 * gcov-tool.c (print_merge_usage_message): Likewise.
20961 (print_rewrite_usage_message): Likewise.
20962 (print_overlap_usage_message): Likewise.
20964 2017-04-28 Martin Liska <mliska@suse.cz>
20966 PR gcov-profile/53915
20967 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
20969 2017-04-28 Martin Liska <mliska@suse.cz>
20971 PR gcov-profile/79891
20972 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
20973 is marked by compiler as living on a line.
20974 (get_cycles_count): Remove usage of the union.
20975 (output_intermediate_file): Likewise.
20976 (find_source): Fix GNU coding style.
20977 (accumulate_line_counts): Remove old non-all block mode.
20978 (output_lines): Remove usage of the union.
20979 * profile.c (output_location): Include all BBs, even if
20980 belonging to a same line (and file) as a previous BB.
20982 2017-04-28 Martin Liska <mliska@suse.cz>
20984 * gcov.c (process_args): Handle new argument 'w'.
20985 (read_graph_file): Assign ID to BBs.
20986 (output_branch_count): Display BB # if verbose flag is set.
20987 (output_lines): Likewise for arcs.
20988 (print_usage): Add '--verbose' option help.
20989 * doc/gcov.texi: Document --verbose (-w) option.
20991 2017-04-28 Martin Liska <mliska@suse.cz>
20993 * gcov.c (struct block_location_info): New struct.
20994 (process_file): Fill up the new structure.
20995 (read_graph_file): Replace usage of encoding by the newly added
20997 (add_line_counts): Likewise.
20998 (accumulate_line_counts): Remove usage of the union.
20999 (function_info::function_info): New function.
21000 (function_info::~function_info): Likewise.
21001 (process_file): Call delete instead of release_function.
21002 (release_function): Release the function.
21003 (release_structures): Call delete instead of release_function.
21004 (solve_flow_graph): Replace usage of num_blocks.
21005 (find_exception_blocks): Likewise.
21006 (output_lines): Fix GNU coding style.
21008 2017-04-28 Martin Liska <mliska@suse.cz>
21011 * coverage.c (coverage_remove_note_file): New function.
21012 * coverage.h: Declare the function.
21013 * toplev.c (finalize): Clean if an error has been seen.
21015 2017-04-28 Martin Liska <mliska@suse.cz>
21017 PR gcov-profile/80031
21018 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
21019 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
21020 * gcov.c (read_graph_file): Read just number of blocks.
21021 * profile.c (branch_prob): Do not stream 0 flags per a basic
21024 2017-04-28 Martin Liska <mliska@suse.cz>
21026 * gcov-dump.c (tag_*): Add new argument to declarations.
21027 (dump_gcov_file): Likewise.
21028 (tag_blocks): Add and use new argument depth.
21029 (tag_arcs): Likewise.
21030 (tag_lines): Likewise.
21031 (tag_counters): Likewise.
21032 (tag_summary): Likewise.
21033 (dump_working_sets): Use depth to do a proper indentation.
21035 2017-04-28 Jakub Jelinek <jakub@redhat.com>
21038 * cgraph.h (symtab_node::debug_symtab): No longer inline.
21039 * symtab.c (symtab_node::debug_symtab): Move definition here.
21041 2017-04-28 Richard Biener <rguenther@suse.de>
21043 * lto-streamer.h (LTO_major_version): Bump to 7.
21045 2017-04-28 Richard Biener <rguenther@suse.de>
21047 * tree-vrp.c (assert_info): New struct.
21048 (add_assert_info): New helper.
21049 (register_edge_assert_for_2): Refactor to add asserts to a vector
21051 (register_edge_assert_for_1): Likewise.
21052 (register_edge_assert_for): Likewise.
21053 (finish_register_edge_assert_for): New helper actually registering
21054 asserts where live on edge.
21055 (find_conditional_asserts): Adjust.
21056 (find_switch_asserts): Likewise.
21057 (evrp_dom_walker::try_find_new_range): Generalize.
21058 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
21060 2017-04-27 Marek Polacek <polacek@redhat.com>
21063 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
21064 arg10 and arg11 to itype.
21066 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
21068 * doc/extend.texi (Object Size Checking): Improve grammar.
21070 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
21073 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
21074 that the logic for permitting reciprocal estimates matches that
21077 2017-04-27 Jakub Jelinek <jakub@redhat.com>
21080 * tree.c (type_cache_hasher::equal): Only compare
21081 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
21082 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
21083 non-aggregate element types.
21084 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
21085 about the flag on ARRAY_TYPEs in the comment, formatting fix.
21087 2017-04-27 Richard Biener <rguenther@suse.de>
21089 PR middle-end/80533
21090 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
21091 stripping ARRAY_REFs from MEM_EXPR make sure we're not
21092 keeping a reference to a trailing array.
21094 2017-04-27 Richard Biener <rguenther@suse.de>
21096 PR middle-end/80539
21097 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
21098 being in loop-closed SSA form conservatively.
21099 (chrec_fold_multiply_poly_poly): Likewise.
21101 2017-04-27 Tamar Christina <tamar.christina@arm.com>
21103 PR middle-end/79665
21104 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
21105 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
21107 2017-04-27 Jakub Jelinek <jakub@redhat.com>
21110 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
21111 (aarch64_function_arg_alignment): Return unsigned int again, but still
21112 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
21113 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
21114 Don't emit -Wpsabi note.
21115 (aarch64_function_arg_boundary): Likewise.
21116 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
21119 2017-04-26 Nathan Sidwell <nathan@acm.org>
21121 * tree.h (crc32_unsigned_n): Declare.
21122 (crc32_unsigned, crc32_unsigned): Make inline.
21123 * tree.c (crc32_unsigned_bits): Replace with ...
21124 (crc32_unsigned_n): ... this.
21125 (crc32_unsigned, crc32_byte): Remove.
21126 (crc32_string): Remove unnecessary braces.
21128 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21130 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
21131 * ipa-inline-analysis.c (MAX_TIME): Remove.
21132 (account_size_time): Use sreal for time.
21133 (dump_inline_summary): Update.
21134 (estimate_function_body_sizes): Update.
21135 (estimate_edge_size_and_time): Update.
21136 (estimate_calls_size_and_time): Update.
21137 (estimate_node_size_and_time): Update.
21138 (inline_merge_summary): Update.
21139 (inline_update_overall_summary): Update.
21140 (estimate_time_after_inlining): Update.
21141 (inline_read_section): Update.
21142 (inline_write_summary): Update.
21143 * ipa-inline.c (compute_uninlined_call_time): Update.
21144 (compute_inlined_call_time): Update.
21145 (recursive_inlining): Update.
21146 (inline_small_functions): Update.
21147 (dump_overall_stats): Update.
21148 * ipa-inline.h: Include sreal.h.
21149 (size_time_entry): Turn time to sreal.
21150 (inline_summary): Turn self_time nad time to sreal.
21152 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21154 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
21156 (sreal::stream_out, sreal::stream_in): New.
21157 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
21159 2017-04-25 Jakub Jelinek <jakub@redhat.com>
21161 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
21164 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
21167 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
21169 (dimode_scalar_chain::compute_convert_gain): Ditto.
21170 (dimode_scalar_chain::make_vector_copies): Ditto.
21171 (dimode_scalar_chain::convert_reg): Ditto.
21172 (dimode_scalar_chain::convert_insn): Ditto.
21173 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
21174 (VI248_AVX512BW_1): New mode iterator.
21175 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
21176 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
21179 2017-04-25 Martin Sebor <msebor@redhat.com>
21181 PR tree-optimization/80497
21182 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
21183 constants are representable in HOST_WIDE_INT.
21184 (parse_directive): Ditto.
21186 2017-04-25 Martin Sebor <msebor@redhat.com>
21189 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
21190 (new_zero_array): Adjust signature.
21191 (dom_info::dom_init): Used unsigned rather that size_t.
21192 (dom_info::dom_info): Same.
21194 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21195 Jakub Jelinek <jakub@redhat.com>
21198 * config/arm/arm.c: Include gimple.h.
21199 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
21200 returns negative, increment ncrn only if it returned positive.
21201 (arm_needs_doubleword_align): Return int instead of bool,
21202 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
21203 members, but if there is any such non-FIELD_DECL
21204 > PARM_BOUNDARY aligned decl, return -1 instead of false.
21205 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
21206 returns negative, increment nregs only if it returned positive.
21207 (arm_setup_incoming_varargs): Likewise.
21208 (arm_function_arg_boundary): Emit -Wpsabi note if
21209 arm_needs_doubleword_align returns negative, return
21210 DOUBLEWORD_ALIGNMENT only if it returned positive.
21212 2017-04-25 Marek Polacek <polacek@redhat.com>
21215 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21216 first argument to type.
21218 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
21221 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
21222 type checks to test for compatibility instead of equality.
21224 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21225 Jakub Jelinek <jakub@redhat.com>
21228 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
21230 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
21231 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
21232 the alignment computation, but return their maximum in warn_alignment.
21233 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
21234 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
21236 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
21237 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
21240 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21242 * config/arc/simdext.md (dmpyh): Fix typo.
21244 2017-04-25 Richard Biener <rguenther@suse.de>
21246 PR tree-optimization/80492
21247 * alias.c (compare_base_decls): Handle registers with asm
21248 specification conservatively.
21249 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
21250 compare_base_decls returning dont-know properly.
21252 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21254 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
21255 (legitimate_offset_address_p): New function.
21256 (arc_legitimate_address_p): Use above function.
21258 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21260 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
21262 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21264 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
21265 ACCH registers whenever they are available.
21267 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21269 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
21270 double regs fix when not used.
21272 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21274 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
21276 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
21277 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
21279 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21281 * config/arc/arc.c (arc_output_addsi): Check for h-register class
21282 when emitting short ADD instructions.
21284 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21286 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
21288 (cmpsi_cc_c_insn): Likewise.
21289 (cbranchsi4_scratch): Compute proper instruction length using
21290 compact_hreg_operand.
21291 * config/arc/predicates.md (compact_hreg_operand): New predicate.
21293 2017-04-25 Richard Biener <rguenther@suse.de>
21295 PR middle-end/80509
21296 * passes.c (pass_manager::pass_manager): Initialize
21297 m_name_to_pass_map.
21299 2017-04-25 Richard Biener <rguenther@suse.de>
21301 PR tree-optimization/79201
21302 * tree-ssa-sink.c (statement_sink_location): Handle calls.
21304 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21307 * config/s390/vector.md: Split MEM->GPR vector moves for
21308 non-s_operand addresses.
21310 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21313 * config/s390/predicates.md (reload_const_wide_int_operand): New
21315 * config/s390/s390.md ("movti"): Remove d/P alternative.
21316 ("movti_bigconst"): New pattern definition.
21318 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21321 * s390-protos.h (s390_expand_cs_hqi): Removed.
21322 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
21323 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
21324 modes as well as CCZ1mode and CCZmode.
21325 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
21326 signature of s390_emit_compare_and_swap.
21327 (s390_expand_cs_hqi): Likewise, make static.
21328 (s390_expand_cs_tdsi): Generate an explicit compare before trying
21329 compare-and-swap, in some cases.
21330 (s390_expand_cs): Wrapper function.
21331 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
21333 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
21334 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
21335 patterns for small and large integers. Forbid symref memory operands.
21336 Move expander to s390.c. Require cc register.
21337 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
21338 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
21339 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
21340 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
21341 symref memory operands. Remove CC mode and call s390_match_ccmode
21343 ("atomic_exchange<mode>"): Allow and implement all integer modes.
21345 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21347 * config/s390/s390.md (define_peephole2): New peephole to help
21348 combining the load-and-test pattern with volatile memory.
21350 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21352 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
21353 with CCZmode for TARGET_Z196.
21355 2017-04-25 Jakub Jelinek <jakub@redhat.com>
21357 PR rtl-optimization/80501
21358 * combine.c (make_compound_operation_int): Set subreg_code to SET
21359 even for AND with mask of the sign bit of mode.
21361 PR rtl-optimization/80500
21362 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
21363 sum's initial value.
21365 2017-04-25 Julian Brown <julian@codesourcery.com>
21366 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21368 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
21370 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
21372 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
21374 2017-04-25 Julian Brown <julian@codesourcery.com>
21375 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21377 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
21378 (thunderx2t99_sha): New Reservation.
21380 2017-04-25 Julian Brown <julian@codesourcery.com>
21381 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21383 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
21384 type for 1-element load.
21386 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
21388 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
21390 2017-04-24 Martin Jambor <mjambor@suse.cz>
21392 PR tree-optimization/80293
21393 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
21394 char arrays not totally scalarizable if it is false.
21395 (analyze_all_variable_accesses): Pass correct value in the new
21396 parameter. Add a statistics counter.
21398 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
21400 PR middle-end/79931
21401 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
21403 2017-04-24 Richard Biener <rguenther@suse.de>
21405 PR tree-optimization/80494
21406 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
21407 out for complex types.
21409 2017-04-24 Richard Biener <rguenther@suse.de>
21411 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
21412 * tree-ssa-sccvn.c (print_scc): Print SCC size.
21413 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
21414 (DFS): Adjust and never fail.
21415 (sccvn_dom_walker::fail): Remove.
21416 (sccvn_dom_walker::before_dom_children): Adjust.
21417 (run_scc_vn): Likewise and never fail.
21418 * tree-ssa-pre.c (pass_pre::execute): Adjust.
21419 (pass_fre::execute): Likewise.
21421 2017-04-24 Richard Biener <rguenther@suse.de>
21423 PR tree-optimization/79725
21424 * tree-ssa-sink.c (statement_sink_location): Return whether
21425 failure reason was zero uses. Move that check later.
21426 (sink_code_in_bb): Deal with zero uses by removing the stmt
21429 2017-04-24 Richard Biener <rguenther@suse.de>
21432 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
21433 pointer-based references.
21435 2017-04-24 Richard Biener <rguenther@suse.de>
21438 * pass_manager.h (pass_manager::operator new): Remove.
21439 (pass_manager::operator delete): Likewise.
21440 * passes.c (pass_manager::operator new): Remove.
21441 (pass_manager::operator delete): Likewise.
21442 (pass_manager::pass_manager): Zero individual pass members.
21444 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
21447 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
21448 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
21449 Check "XEXP (src, 1)" operand here.
21450 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
21451 Check "XEXP (src, 1)" operand here.
21452 (dimode_scalar_chain::make_vector_copies): Detect count register
21453 of a shift instruction. Zero extend count register from QImode
21454 to DImode to satisfy vector shift pattern count operand predicate.
21455 Substitute vector shift count operand with a DImode copy.
21456 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
21459 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
21461 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
21462 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21463 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
21464 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21465 (UNSPEC_NOREX_MEM): Remove definition.
21467 2017-04-21 Richard Biener <rguenther@suse.de>
21469 PR tree-optimization/79547
21470 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
21471 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
21472 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
21473 without any constraints.
21475 2017-04-21 Richard Biener <rguenther@suse.de>
21477 PR tree-optimization/78847
21478 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
21480 2017-04-21 Richard Biener <rguenther@suse.de>
21482 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
21483 (build_distinct_type_copy): Likewise.
21484 (build_variant_type_copy): Likewise.
21485 * tree.c (build_qualified_type): Pass down mem-stat info.
21486 (build_distinct_type_copy): Likewise.
21487 (build_variant_type_copy): Likewise.
21489 2017-04-21 Richard Biener <rguenther@suse.de>
21491 PR tree-optimization/80237
21492 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
21494 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
21495 for a simplified result.
21497 2016-04-21 Richard Biener <rguenther@suse.de>
21499 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
21500 sth as strict as a simple_iv but a chrec without symbols and an
21501 operand defined in the loop we are peeling (and not some subloop).
21502 (propagate_constants_for_unrolling): Propagate all constants.
21504 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21507 * config/i386/i386.c (print_reg): Remove assert for disalowed
21508 regno values, call output_operand_lossage instead.
21510 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21513 * config/i386/constraints.md (Yc): New register constraint.
21514 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
21515 Use Yc constraint for alternative 2 of operand 0. Remove
21516 preferred_for_speed attribute.
21518 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
21520 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
21521 lastprivate clauses in SIMT case.
21523 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
21525 * doc/invoke.texi (-Wextra-semi): Document new warning option.
21527 2017-04-20 Richard Biener <rguenther@suse.de>
21529 PR tree-optimization/57796
21530 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
21531 as N scalar stores.
21532 (vect_model_load_cost): Cost gathers as N scalar loads.
21534 2017-04-20 Richard Biener <rguenther@suse.de>
21536 * ggc-page.c (ggc_allocated_p): Rename to ...
21537 (safe_lookup_page_table_entry): ... this and return the lookup
21539 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
21541 2017-04-20 Richard Biener <rguenther@suse.de>
21543 PR tree-optimization/80453
21544 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
21545 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
21546 from the conditions.
21547 (vn_phi_eq): Pass them down.
21548 (vn_phi_lookup): Record them.
21549 (vn_phi_insert): Likewise.
21551 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
21553 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
21554 uninitialized variable warning to avoid buffer overrun.
21556 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
21559 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
21560 is suppressed for '{ 0 }' in C.
21562 2017-04-20 Jakub Jelinek <jakub@redhat.com>
21564 * BASE-VER: Set to 8.0.0.
21566 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
21568 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
21569 priority .init_array and .fini_array section with SECTION_NOTYPE
21572 2017-04-20 Jakub Jelinek <jakub@redhat.com>
21574 PR middle-end/80423
21575 * tree.h (build_array_type): Add typeless_storage default argument.
21576 * tree.c (type_cache_hasher::equal): Also compare
21577 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
21578 (build_array_type): Add typeless_storage argument, set
21579 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
21581 (build_nonshared_array_type): Adjust build_array_type_1 caller.
21582 (build_array_type): Likewise. Add typeless_storage argument.
21584 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21585 Jakub Jelinek <jakub@redhat.com>
21587 PR tree-optimization/80426
21588 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
21589 operation on symbolic operands, also compute the overflow for the
21590 invariant part when the operation degenerates into a negation.
21592 2017-04-19 Jakub Jelinek <jakub@redhat.com>
21595 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
21596 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
21599 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
21601 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
21604 * config/avr/avr.c (tree.h): Include it.
21605 (cgraph.h): Include it.
21606 (avr_encode_section_info): Don't warn for uninitialized progmem
21607 variable if it's just an alias.
21609 2017-04-19 Richard Biener <rguenther@suse.de>
21612 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
21613 when needed by AutoPGO.
21615 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
21618 * doc/lto.texi: Remove an extra 'that'.
21620 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
21622 PR rtl-optimization/80429
21623 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
21624 are only used in debug insns.
21626 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
21627 Vladimir Makarov <vmakarov@redhat.com>
21629 * config/sparc/predicates.md (input_operand): Add comment. Return
21630 true for any memory operand when LRA is in progress.
21631 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
21633 2017-04-18 Jeff Law <law@redhat.com>
21636 * mips.md ({return,simple_return}_internal): Do not overwrite
21639 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21641 PR tree-optimization/80443
21642 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
21643 instead of adding 1, subtract -1 and similarly instead of subtracting
21646 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
21648 PR rtl-optimization/80357
21649 * haifa-sched.c (tmp_bitmap): New variable.
21650 (model_recompute): Handle duplicate use records.
21651 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
21652 (free_global_sched_pressure_data): Free it.
21654 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
21657 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
21658 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21659 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21660 instead of SYSTEM_HEADER_DIR.
21662 2017-04-18 Jeff Law <law@redhat.com>
21664 PR middle-end/80422
21665 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
21666 predecessors after walking up the insn chain.
21668 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21671 * dwarf2out.c (modified_type_die): Try harder not to emit internal
21672 sizetype type into debug info.
21674 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21677 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
21678 unneeded test for TARGET_UPPER_REGS_SF.
21679 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
21681 2017-04-18 Jakub Jelinek <jakub@redhat.com>
21684 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
21685 instead of gsi_after_labels.
21687 2017-04-18 Jeff Law <law@redhat.com>
21689 * regcprop.c (maybe_mode_change): Avoid creating copies of the
21693 2017-04-13 Jeff Law <law@redhat.com>
21694 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21695 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21697 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
21700 * config/avr/avr.c (intl.h): Include it.
21701 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
21703 2017-04-18 Martin Liska <mliska@suse.cz>
21705 PR gcov-profile/78783
21706 * gcov-tool.c (gcov_output_files): Validate that destination
21707 file is either removed by the tool or by a user.
21709 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21710 Guy Benyei <guybe@mellanox.com>
21712 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
21713 block, and do not negate it, the stored id is already negative.
21715 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
21717 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
21719 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
21722 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
21723 masks of options that should be turned off if the VSX vector
21724 options are turned off.
21725 (OTHER_P8_VECTOR_MASKS): Likewise.
21726 (OTHER_VSX_VECTOR_MASKS): Likewise.
21727 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
21728 rs6000_disable_incompatible_switches to validate no type switches
21730 (rs6000_incompatible_switch): New function to disallow turning on
21731 other vector options if -mno-vsx, -mno-power8-vector, or
21732 -mno-power9-vector are specified.
21734 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21736 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
21738 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21740 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
21741 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
21742 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
21743 (ARG_POINTER_CFA_OFFSET): Likewise.
21745 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
21747 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
21748 conditions to take advantage of various optimizations.
21750 2017-04-13 Jeff Law <law@redhat.com>
21752 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
21753 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
21754 (zero_extendsidi2_dext): Likewise.
21756 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21759 * fold-const.c (fold_ternary_loc): Revert
21760 use op0 instead of fold_convert_loc (loc, type, arg0) part of
21763 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
21765 PR rtl-optimization/80343
21766 * lra-remat.c (update_scratch_ops): Assign original hard reg to
21767 new scratch pseudo.
21769 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
21772 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
21773 to ubsan_encode_value.
21775 2017-04-13 Jeff Law <law@redhat.com>
21777 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
21778 appearing in DEBUG_INSNs.
21780 2017-04-13 Martin Liska <mliska@suse.cz>
21782 PR gcov-profile/80413
21783 * gcov-io.c (gcov_write_string): Copy to buffer just when
21784 allocated size is greater than zero.
21786 2017-04-13 Jakub Jelinek <jakub@redhat.com>
21789 * dwarf2out.c (decls_for_scope): Ignore declarations of
21790 current_function_decl in BLOCK_NONLOCALIZED_VARS.
21792 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
21795 * ipa-visibility.c (non_local_p): Fix typos.
21796 (localize_node): When localizing symbol in same comdat group,
21797 dissolve the group only when we know external symbols are going
21799 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
21801 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21803 PR tree-optimization/79390
21804 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
21805 order does not result in usable sequence, retry with reversed operand
21811 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
21812 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
21813 op0 instead of fold_convert_loc (loc, type, arg0).
21815 2017-04-12 Jeff Law <law@redhat.com>
21817 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
21818 has a delay slot in the generated code.
21820 * config/cris/cris.md (cris_preferred_reload_class): Return
21821 GENNONACR_REGS rather than GENERAL_REGS.
21823 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21826 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
21827 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
21828 signedness of the result type.
21830 2017-04-12 Richard Biener <rguenther@suse.de>
21831 Jeff Law <law@redhat.com>
21833 PR tree-optimization/80359
21834 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
21835 trim stores to TARGET_MEM_REFs.
21837 2017-04-12 Richard Biener <rguenther@suse.de>
21839 PR tree-optimization/79390
21840 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
21841 threading case even more.
21843 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
21846 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
21847 for quad_address_p for TImode, instead of just not indexed_address.
21849 2017-04-12 Richard Biener <rguenther@suse.de>
21850 Bernd Edlinger <bernd.edlinger@hotmail.de>
21852 PR middle-end/79671
21853 * alias.c (component_uses_parent_alias_set_from): Handle
21854 TYPE_TYPELESS_STORAGE.
21855 (get_alias_set): Likewise.
21856 * tree-core.h (tree_type_common): Add typeless_storage flag.
21857 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
21858 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
21859 for types containing members with TYPE_TYPELESS_STORAGE.
21860 (place_field): Likewise.
21861 (layout_type): Likewise for ARRAY_TYPE.
21862 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
21863 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
21864 TYPE_TYPELESS_STORAGE.
21865 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
21867 2017-04-12 Jakub Jelinek <jakub@redhat.com>
21870 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21871 first argument to type.
21873 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21877 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
21878 CONST0_RTX (mode) rather than const0_rtx where appropriate.
21879 (rs6000_expand_binop_builtin): Likewise.
21880 (rs6000_expand_ternop_builtin): Likewise; also add missing
21881 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
21882 vshasigma built-ins.
21883 * doc/extend.texi: Document that vec_xxpermdi's third argument
21884 must be a constant.
21886 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
21888 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
21889 Use shift_const cost parameter when calculating gain of STV shifts.
21891 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
21893 PR rtl-optimization/70478
21894 * lra-constraints.c (process_alt_operands): Check memory for
21895 disfavoring memory insn operand.
21897 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21899 PR middle-end/80100
21900 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
21901 left shift in unsigned HOST_WIDE_INT type.
21903 PR rtl-optimization/80385
21904 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
21905 (not (neg X)) into (plus X -1) for complex or non-integral modes.
21908 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
21909 if they have any depend clauses.
21911 2017-04-11 Martin Liska <mliska@suse.cz>
21914 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
21915 * ipa-split.c (split_function): Create a local comdat symbol
21916 if caller is in a comdat group.
21918 2017-04-11 Martin Liska <mliska@suse.cz>
21921 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
21924 2017-04-11 Martin Sebor <msebor@redhat.com>
21926 PR middle-end/80364
21927 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
21928 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
21930 (directive::set_width, directive::set_precision, format_character):
21932 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
21935 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
21938 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
21939 conflict, set target->arch_name instead of target->cpu_name.
21941 2017-04-11 Richard Biener <rguenther@suse.de>
21943 PR tree-optimization/80374
21944 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
21945 build_zero_cst, remove fold_convertible_p check again.
21947 2017-04-11 Martin Liska <mliska@suse.cz>
21950 * ubsan.c (instrument_object_size): Do not instrument register
21953 2017-04-11 Jakub Jelinek <jakub@redhat.com>
21956 * config/i386/i386-builtin-types.def
21957 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
21958 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
21959 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
21960 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
21961 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
21962 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
21963 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
21964 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
21965 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
21966 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
21967 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
21968 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
21969 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
21970 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
21971 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
21972 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
21973 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
21974 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
21975 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
21976 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
21977 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
21978 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
21979 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
21980 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
21981 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
21982 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
21983 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
21984 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
21985 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
21986 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
21987 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
21988 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
21989 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
21990 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
21991 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
21992 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
21993 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
21994 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
21995 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
21996 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
21997 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
21998 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
21999 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
22000 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
22001 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
22002 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
22004 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
22005 flag to second_arg_count, handle 4 argument function type _COUNT
22006 aliases, handle second_arg_count on second argument rather than last.
22008 2017-04-10 Jeff Law <law@redhat.com>
22010 PR tree-optimization/80374
22011 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
22012 record anything if we can not convert integer_zero_node to the
22015 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22018 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22019 Enhance special handling given to the TARGET_P9_MINMAX option in
22020 relation to certain other options.
22022 2017-04-10 Bin Cheng <bin.cheng@arm.com>
22024 PR tree-optimization/80153
22025 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
22026 remove POINTER_PLUS_EXPR's base part directly, rather than through
22029 2017-04-10 Richard Biener <rguenther@suse.de>
22030 Bin Cheng <bin.cheng@arm.com>
22032 PR tree-optimization/80153
22033 * tree-affine.c (aff_combination_to_tree): Get base pointer from
22034 the first element of pointer type aff_tree. Build result expr in
22036 (add_elt_to_tree): Convert to type unconditionally. Remove other
22037 fold_convert calls.
22038 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
22039 (rewrite_use_nonlinear_expr): Check invariant using iv information.
22041 2017-04-10 Richard Biener <rguenther@suse.de>
22043 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22046 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
22048 PR rtl-optimization/70478
22049 * lra-constraints.c (curr_small_class_check): New.
22050 (update_and_check_small_class_inputs): New.
22051 (process_alt_operands): Update curr_small_class_check. Disfavor
22052 alternative insn memory operands. Check available regs for small
22055 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
22058 * config/mips/mips.opt (-mvirt): Update description.
22059 * doc/invoke.texi (-mvirt): Likewise.
22061 2017-04-10 Richard Biener <rguenther@suse.de>
22063 PR middle-end/80362
22064 * fold-const.c (fold_binary_loc): Look at unstripped ops when
22065 looking for NEGATE_EXPR in -A / -B to A / B folding.
22067 2017-04-10 Martin Liska <mliska@suse.cz>
22069 PR gcov-profile/80224
22070 * gcov.c (print_usage): Fix usage string.
22071 (get_gcov_intermediate_filename): Remove.
22072 (output_gcov_file): Use both for normal and intermediate format.
22073 (generate_results): Do not initialize special file for
22074 intermediate format.
22076 2017-04-10 Richard Biener <rguenther@suse.de>
22078 PR tree-optimization/80304
22079 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
22082 2017-04-10 Nathan Sidwell <nathan@acm.org>
22085 * config/rs6000/rs6000.c (rs6000_vector_type): New.
22086 (rs6000_init_builtins): Use it.
22088 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22090 * config/arm/arm.md (<mrc>): Add mode to SET source.
22091 (<mrrc>): Likewise.
22093 2017-04-10 Richard Biener <rguenther@suse.de>
22095 PR middle-end/80344
22096 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
22098 2017-04-10 Jakub Jelinek <jakub@redhat.com>
22101 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
22102 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
22103 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
22104 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
22105 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
22106 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
22107 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
22108 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
22109 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
22110 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
22111 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
22112 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
22113 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
22114 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
22115 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
22116 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
22117 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
22118 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
22119 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
22120 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
22121 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
22122 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
22123 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
22125 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
22127 PR rtl-optimization/70478
22128 * lra-constraints.c: Reverse the last patch.
22130 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
22132 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
22133 Add comment for WCHAR_T.
22135 2017-04-08 Martin Liska <mliska@suse.cz>
22138 2017-04-07 Martin Liska <mliska@suse.cz>
22141 * ipa-split.c (split_function): Add function part to a same comdat
22144 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22147 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
22149 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
22151 * rs6000/rs6000.c (vec_load_pendulum): Rename...
22152 (vec_pairing): ...to this.
22153 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
22154 (rs6000_sched_init): Adjust for name change.
22155 (struct rs6000_sched_context): Likewise.
22156 (rs6000_init_sched_context): Likewise.
22157 (rs6000_set_sched_context): Likewise.
22159 2017-04-07 Jakub Jelinek <jakub@redhat.com>
22165 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
22167 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
22168 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
22169 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
22171 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
22173 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
22175 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22177 PR rtl-optimization/70703
22178 * ira-color.c (update_conflict_hard_regno_costs): Use
22179 int64_t instead of HOST_WIDE_INT.
22181 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22183 PR rtl-optimization/70478
22184 * lra-constraints.c (process_alt_operands): Disfavor alternative
22185 insn memory operands.
22187 2017-04-07 Jeff Law <law@redhat.com>
22189 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
22190 CALL and NOTE_INSN_CALL_ARG_LOCATION.
22192 2017-04-07 Martin Liska <mliska@suse.cz>
22195 * config/aarch64/aarch64.c (aarch64_process_target_attr):
22196 Show error message instead of an ICE.
22198 2017-04-07 Martin Liska <mliska@suse.cz>
22201 * ipa-split.c (split_function): Add function part to a same comdat
22204 2017-04-07 Richard Biener <rguenther@suse.de>
22206 PR middle-end/80341
22207 * tree.c (get_unwidened): Also handle ! for_type case for
22209 * convert.c (do_narrow): Split out from ...
22210 (convert_to_integer_1): ... here. Do not pass final truncation
22211 type to get_unwidened for TRUNC_DIV_EXPR.
22213 2017-04-07 Richard Biener <rguenther@suse.de>
22215 * tree-affine.c (wide_int_ext_for_comb): Take type rather
22217 (aff_combination_const): Adjust.
22218 (aff_combination_scale): Likewise.
22219 (aff_combination_add_elt): Likewise.
22220 (aff_combination_add_cst): Likewise.
22221 (aff_combination_convert): Likewise.
22222 (add_elt_to_tree): Likewise. Remove unused argument.
22223 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
22225 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
22227 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
22229 * config/arm/arm.c (arm_default_short_enums): Use
22230 ARM_DEFAULT_SHORT_ENUMS.
22231 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
22233 2017-04-06 Jakub Jelinek <jakub@redhat.com>
22236 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
22237 members with redundant out-of-class redeclaration.
22239 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
22242 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
22243 * config/i386/i386.md (*zero_extendsidi2):
22244 Add (?*x,*x) and (?*v,*v) alternatives.
22246 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
22249 * config/i386/i386.c (ix86_expand_builtin)
22250 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
22251 mode from insn data. Convert operands to insn operand mode.
22252 Copy operands that don't satisfy insn predicate to a register.
22254 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
22256 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
22259 2017-04-06 Richard Biener <rguenther@suse.de>
22261 PR tree-optimization/80334
22262 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
22263 preserve alignment of accesses.
22265 2017-04-06 Richard Biener <rguenther@suse.de>
22267 PR tree-optimization/80262
22268 * tree-sra.c (build_ref_for_offset): Preserve address-space
22270 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
22271 Drop useless address-space information on MEM_REF offsets.
22273 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
22275 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
22277 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
22279 PR rtl-optimization/70703
22280 * ira-color.c (update_conflict_hard_regno_costs): Use
22281 HOST_WIDE_INT instead of long.
22283 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
22286 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
22287 not defined for x86_64 target. Add -mmmx target option when __SSE2__
22289 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
22290 for x86_64 target. Handle -m3dnowa option.
22292 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
22294 PR rtl-optimization/70703
22295 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
22296 (update_conflict_hard_regno_costs): Use long instead of unsigned
22297 arithmetic for cost calculation.
22299 2017-04-05 Jakub Jelinek <jakub@redhat.com>
22300 Bernd Edlinger <bernd.edlinger@hotmail.de>
22303 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
22306 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
22309 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
22310 ptr_mode with Pmode throughout.
22311 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
22312 into probe_stack_range and use DImode.
22314 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
22317 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
22318 call_eh_return is true.
22320 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22322 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
22323 Initialize last_match_fntype_index.
22325 2017-04-05 Jakub Jelinek <jakub@redhat.com>
22328 * tree-nvr.c: Include internal-fn.h.
22329 (pass_return_slot::execute): Ignore internal calls without
22332 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22333 Richard Biener <rguenther@suse.de>
22336 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
22337 captures used multiple times, except for the last use.
22338 * generic-match-head.c: Include gimplify.h.
22340 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22342 PR tree-optimization/79390
22343 * target.h (struct noce_if_info): Declare.
22344 * targhooks.h (default_noce_conversion_profitable_p): Declare.
22345 * target.def (noce_conversion_profitable_p): New target hook.
22346 * ifcvt.h (struct noce_if_info): New type, moved from ...
22347 * ifcvt.c (struct noce_if_info): ... here.
22348 (noce_conversion_profitable_p): Renamed to ...
22349 (default_noce_conversion_profitable_p): ... this. No longer
22351 (noce_try_store_flag_constants, noce_try_addcc,
22352 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
22353 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
22354 instead of noce_conversion_profitable_p.
22355 * config/i386/i386.c: Include ifcvt.h.
22356 (ix86_option_override_internal): Don't override
22357 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
22358 (ix86_noce_conversion_profitable_p): New function.
22359 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
22360 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
22361 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
22362 * doc/tm.texi: Regenerated.
22364 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22366 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
22369 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22372 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
22373 instructions for small multiply cores.
22375 2017-04-04 Jeff Law <law@redhat.com>
22377 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
22379 (mips_expand_vec_perm_const): Initialize elements in orig_perm
22380 that are not set by the loop over the elements.
22382 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22385 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
22386 int mode, convert_modes it to mode as unsigned, otherwise use
22387 lowpart_subreg to mode rather than SImode.
22388 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
22389 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
22390 Use DImode instead of SImode for the shift count operand.
22391 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
22394 2017-04-04 Richard Biener <rguenther@suse.de>
22396 PR middle-end/80281
22397 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
22398 arithmetic done for the negate or the plus. Simplify.
22399 (A - (-B) -> A + B): Likewise.
22400 * fold-const.c (split_tree): Make sure to not negate pointers.
22402 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
22404 PR rtl-optimization/60818
22405 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
22406 a compare of comparisons with the thing compared if this results
22407 in a different machine mode.
22409 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
22411 * alias.c (base_alias_check): Fix typo in comment.
22412 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
22413 * cgraphunit.c (symbol_table::compile): Likewise.
22414 * collect2.c (maybe_run_lto_and_relink): Likewise.
22415 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
22416 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
22417 * config/avr/avr.c (avr_map_op_t): Likewise.
22418 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
22419 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
22420 * config/epiphany/epiphany.md (movcc): Likewise.
22421 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
22422 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
22424 * config/mips/mips.c (mips_save_restore_reg): Likewise.
22425 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
22426 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
22427 * config/sh/sh.c (sh_rtx_costs): Likewise.
22428 * fold-const.c (fold_truth_andor): Likewise.
22429 * genautomata.c (collapse_flag): Likewise.
22430 * gengtype.h (struct type::u::s): Likewise.
22431 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
22432 * input.c (FORMAT_AMOUNT): Likewise.
22433 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
22434 (known_aggs_to_agg_replacement_list): Likewise.
22435 * ipa-inline-analysis.c: Likewise.
22436 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
22437 * ipa-polymorphic-call.c
22438 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
22439 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
22440 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
22442 * modulo-sched.c (apply_reg_moves): Likewise.
22443 * omp-expand.c (build_omp_regions_1): Likewise.
22444 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
22445 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
22446 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
22447 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
22448 * value-prof.c: Likewise.
22449 * var-tracking.c (val_reset): Likewise.
22451 2017-04-03 Richard Biener <rguenther@suse.de>
22453 PR tree-optimization/80275
22454 * fold-const.c (split_address_to_core_and_offset): Handle
22457 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
22459 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
22460 descriptors is at least equal to that of functions.
22462 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22464 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
22466 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22469 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
22470 (mov<IMOD4:mode>): New expander.
22471 (*mov<IMOD4:mode>_internal): New insn and split pattern.
22473 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
22475 PR rtl-optimization/79405
22476 * fwprop.c (propagations_left): New variable.
22477 (forward_propagate_into): Decrement it.
22478 (fwprop_init): Initialize it.
22479 (fw_prop): If the variable has reached zero, stop propagating.
22480 (fwprop_addr): Ditto.
22482 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22485 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
22486 a FUNCTION_DECL, pass it as decl instead of origin to
22489 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
22491 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
22494 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
22497 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
22498 TARGET_VSX_SMALL_INTEGER.
22500 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22502 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22503 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
22505 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
22507 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
22508 extraction from odd-numbered MSA register.
22510 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22512 PR middle-end/80173
22513 * expmed.c (store_bit_field_1): Don't attempt to create
22514 a word subreg out of hard registers wider than word if they
22515 have HARD_REGNO_NREGS of 1 for their mode.
22517 PR middle-end/80163
22518 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
22519 conversions to integer types wider than word and pointer.
22522 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
22523 (rtx_equal_for_cselib_p): Pass 0 to it.
22524 * cselib.c (cselib_hasher::equal): Likewise.
22525 (rtx_equal_for_cselib_1): Add depth argument. If depth
22526 is 128, don't look up VALUE locs and punt. Increment
22527 depth in recursive calls when walking VALUE locs.
22529 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
22531 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
22532 (make_gcov_file_name): Use the canonical path name for generating
22534 (read_line): Fix handling of files with ascii null bytes.
22536 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
22538 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
22539 to initialise a vector register instead
22540 of using a const_int.
22542 2017-03-30 Jakub Jelinek <jakub@redhat.com>
22544 PR translation/80189
22545 * gimplify.c (omp_default_clause): Use %qs instead of %s in
22546 diagnostic messages.
22548 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
22551 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
22552 (dfp_diex_<mode>): Update mode of operand 1.
22553 * doc/extend.texi (dxex, dxexq): Document change to return type.
22554 (diex, diexq): Document change to argument type.
22556 2017-03-30 Martin Jambor <mjambor@suse.cz>
22559 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
22560 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
22561 it reflects the signature changes performed at the callee side.
22562 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
22563 to cgraph_build_function_type_skip_args.
22564 (build_function_decl_skip_args): Adjust call to the above function.
22566 2017-03-30 Jakub Jelinek <jakub@redhat.com>
22569 * config/i386/sse.md
22570 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
22571 register as dest whenever it is a MEM not rtx_equal_p to the
22572 corresponding dup operand, and when forcing into reg move the
22573 reg into the memory afterwards.
22574 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
22575 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
22576 for the force_reg mode.
22577 (avx512vl_vextractf128<mode>): Use register as dest either
22578 always when a MEM, or when it is a MEM not rtx_equal_p to the
22579 corresponding dup operand, or even not when it is a CONST_VECTOR
22580 depending on the mode and lo vs. hi.
22581 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
22583 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
22584 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
22585 Likewise. Require that operands[2] is even.
22586 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
22587 Remove extraneous parens. Require that operands[2] is a multiple
22589 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
22590 operands[0] is a MEM if <mask_applied>, the predicates/constraints
22591 disallow memory then.
22593 2017-03-30 Richard Biener <rguenther@suse.de>
22595 PR tree-optimization/77498
22596 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
22597 to non-constants over backedges.
22599 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
22601 PR rtl-optimization/80233
22602 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
22603 as last_combined_insn. Do not test for BARRIER_P separately.
22605 2017-03-29 Andreas Schwab <schwab@suse.de>
22608 * calls.c (prepare_call_address): Convert funexp to Pmode before
22609 copying to temp reg.
22611 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22613 PR tree-optimization/80158
22614 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
22615 Handle possible future case of more than one alternate
22617 (replace_rhs_if_not_dup): Likewise.
22618 (replace_one_candidate): Likewise.
22620 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
22622 PR rtl-optimization/80193
22623 * ira.c (ira): Do not check allocation for LRA.
22625 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
22627 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
22628 (nvptx_output_simt_exit): Declare.
22629 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
22630 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
22631 (init_softstack_frame): Move initialization of crtl->is_leaf to...
22632 (nvptx_declare_function_name): ...here. Emit declaration of local
22633 memory space buffer for omp_simt_enter insn.
22634 (nvptx_output_unisimt_switch): New.
22635 (nvptx_output_softstack_switch): New.
22636 (nvptx_output_simt_enter): New.
22637 (nvptx_output_simt_exit): New.
22638 * config/nvptx/nvptx.h (struct machine_function): New fields
22639 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
22640 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
22641 (UNSPECV_SIMT_EXIT): Ditto.
22642 (omp_simt_enter_insn): New insn.
22643 (omp_simt_enter): New expansion.
22644 (omp_simt_exit): New insn.
22645 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
22647 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
22648 (expand_GOMP_SIMT_ENTER_ALLOC): New.
22649 (expand_GOMP_SIMT_EXIT): New.
22650 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
22651 (GOMP_SIMT_ENTER_ALLOC): Ditto.
22652 (GOMP_SIMT_EXIT): Ditto.
22653 * target-insns.def (omp_simt_enter): New insn.
22654 (omp_simt_exit): Ditto.
22655 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
22657 (lower_rec_simd_input_clauses): Implement SIMT privatization.
22658 (lower_rec_input_clauses): Likewise.
22659 (lower_lastprivate_clauses): Handle SIMT privatization.
22661 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
22662 (ompdevlow_adjust_simt_enter): New.
22663 (find_simtpriv_var_op): New.
22664 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
22665 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
22667 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
22668 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
22669 (copy_decl_for_dup_finish): Ditto.
22671 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
22673 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
22676 * config/i386/i386.c (ix86_option_override_internal): Always
22677 allow -mpreferred-stack-boundary=3 for 64-bit targets.
22679 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22681 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
22683 2017-03-28 Bin Cheng <bin.cheng@arm.com>
22685 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
22686 mark new edge's irreducible flag accordign to it.
22687 (vect_do_peeling): Check loop preheader edge's irreducible flag
22688 and pass it to function slpeel_add_loop_guard.
22690 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
22692 PR tree-optimization/80218
22693 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
22694 Update block frequencies and counts.
22696 2017-03-28 Richard Biener <rguenther@suse.de>
22698 PR tree-optimization/78644
22699 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
22700 of a simplification result we may not use it at all.
22702 2017-03-28 Richard Biener <rguenther@suse.de>
22705 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
22706 without arguments, generate default definition of a SSA name.
22708 2017-03-28 Richard Biener <rguenther@suse.de>
22710 PR middle-end/80222
22711 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
22712 TYPE_REF_CAN_ALIAS_ALL references.
22713 * fold-const.c (fold_indirect_ref_1): Likewise.
22715 2017-03-28 Martin Liska <mliska@suse.cz>
22718 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
22719 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
22721 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22722 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
22724 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
22725 (EXTRA_SPECS): Define.
22726 (SUBTARGET_EXTRA_SPECS): Likewise.
22727 (SUBTARGET_CPP_SPEC): Likewise.
22728 * config/arc/elf.h (EXTRA_SPECS): Renamed to
22729 SUBTARGET_EXTRA_SPECS.
22730 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
22732 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
22734 * config/arc/simdext.md (vst64_insn): Update pattern.
22735 (vld32wh_insn): Likewise.
22736 (vld32wl_insn): Likewise.
22737 (vld64_insn): Likewise.
22738 (vld32_insn): Likewise.
22740 2017-03-28 Marek Polacek <polacek@redhat.com>
22743 * fold-const.c (fold_comparison): Use protected_set_expr_location
22744 instead of SET_EXPR_LOCATION.
22746 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
22748 * tree.c (add_expr): Avoid name lookup warning.
22750 2017-03-27 Jeff Law <law@redhat.com>
22752 PR tree-optimization/80216
22753 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
22754 function name. Limit recursion depth.
22755 (record_temporary_equivalences): Corresponding changes.
22757 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
22759 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
22762 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22765 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
22767 * cfgcleanup.c (reg_note_cfa_p): New array.
22768 (insns_have_identical_cfa_notes): New function.
22769 (old_insns_match_p): Don't cross-jump in between /f
22770 and non-/f instructions. If both i1 and i2 are frame related,
22771 verify all CFA notes, their order and content.
22773 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
22776 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
22777 HImode and SImode with zero extend to DImode to one insn.
22778 (bswap<mode>2_extenddi): Likewise.
22779 (bswapsi2_extenddi): Likewise.
22780 (bswaphi2_extendsi): Likewise.
22781 (bswaphi2): Combine bswap HImode and SImode into one insn.
22782 Separate memory insns from swapping register.
22783 (bswapsi2): Likewise.
22784 (bswap<mode>2): Likewise.
22785 (bswaphi2_internal): Delete, no longer used.
22786 (bswapsi2_internal): Likewise.
22787 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
22788 store, and gpr<-gpr swap insns.
22789 (bswap<mode>2_store): Likewise.
22790 (bswaphi2_reg): Register only splitter, combine with the splitter.
22791 (bswaphi2 splitter): Likewise.
22792 (bswapsi2_reg): Likewise.
22793 (bswapsi2 splitter): Likewise.
22794 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
22795 the insns into load, store, and register/register insns.
22796 (bswapdi2_ldbrx): Likewise.
22797 (bswapdi2_load): Likewise.
22798 (bswapdi2_store): Likewise.
22799 (bswapdi2_reg): Likewise.
22801 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
22803 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
22804 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
22806 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
22809 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
22811 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
22812 special handling for target option conflicts between dform
22813 options (-mpower9-dform, -mpower9-dform-vector,
22814 -mpower9-dform-scalar) and -mno-direct-move.
22816 2017-03-27 Richard Biener <rguenther@suse.de>
22818 PR tree-optimization/80181
22819 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
22821 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22823 * config/arc/predicates.md (move_double_src_operand): Replace the
22824 call to move_double_src_operand with a call to address_operand.
22826 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22828 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
22829 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
22830 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
22832 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22834 * config/arc/predicates.md (long_immediate_loadstore_operand):
22835 Consider scaled addresses cases.
22837 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
22839 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
22840 restored when in interrupt.
22841 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
22842 doesn't have delay slot.
22844 2017-03-27 Richard Biener <rguenther@suse.de>
22847 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
22848 inlined thunk clones.
22850 2017-03-27 Jakub Jelinek <jakub@redhat.com>
22853 * asan.c (instrument_derefs): Copy over last operand from
22854 original COMPONENT_REF to the new COMPONENT_REF with
22855 DECL_BIT_FIELD_REPRESENTATIVE.
22856 * ubsan.c (instrument_object_size): Likewise.
22858 2017-03-27 Richard Biener <rguenther@suse.de>
22860 PR tree-optimization/80170
22861 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
22862 sure DR/SCEV didnt fold in constants we do not see when looking
22863 at the reference base alignment.
22865 2017-03-27 Richard Biener <rguenther@suse.de>
22867 PR middle-end/80171
22868 * gimple-fold.c (fold_ctor_reference): Properly guard against
22869 NULL return value from canonicalize_constructor_val.
22871 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
22874 * config/i386/i386.c (ix86_expand_builtin)
22875 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
22876 flags reg setting and flags reg using instructions.
22877 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
22878 clobbering instructions to zero extend op2.
22880 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
22882 * doc/install.texi (Configuration) <--with-aix-soname>:
22883 Update link to AIX ld.
22885 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
22887 PR rtl-optimization/80160
22888 PR rtl-optimization/80159
22889 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
22890 reg_alternate_class into account.
22892 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
22895 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
22896 to consider in curr_insn_transform.
22898 2017-03-24 Jakub Jelinek <jakub@redhat.com>
22900 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
22901 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
22902 and emit_mode_inner.
22904 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22906 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
22907 argument to the overloaded builtin variants. Use the new flag to
22908 deprecate certain builtin variants.
22909 * config/s390/s390-builtin-types.def: Add new builtin types.
22910 * config/s390/s390-builtins.h: Support new flags field for
22911 overloaded builtins.
22912 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
22913 (s390_macro_to_expand): Enable vector float data type.
22914 (s390_cpu_cpp_builtins_internal): Indicate support of the new
22915 builtins by incrementing the __VEC__ version number.
22916 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
22918 (s390_resolve_overloaded_builtin): Emit error messages depending
22919 on the builtin flags.
22920 * config/s390/s390.c (s390_expand_builtin): Support additional
22921 flags argument. Change error message to match the messages
22922 emitted in s390-c.c.
22923 * config/s390/s390.md: New UNSPEC_* constants.
22924 (op_type): Add new instruction types.
22925 * config/s390/vecintrin.h: Add new builtins and test data class
22927 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
22928 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
22929 (VEC_INEXACT, VEC_NOINEXACT): New constants.
22930 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
22931 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
22932 ("vec_mergel<mode>"): V_HW -> VEC_HW.
22934 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
22935 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
22936 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
22937 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
22939 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
22940 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
22941 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
22942 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
22944 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
22945 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
22946 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
22947 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
22948 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
22949 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
22950 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
22952 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
22953 ("vec_scatter_element<V_HW_4:mode>_DI")
22954 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
22955 ("vec_fpint<mode>", "vflls")
22956 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
22957 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
22958 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
22959 ("*vec_cmphe<mode>_cc"): ... these.
22961 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
22962 mode constant instead of magic value.
22964 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22966 * config/s390/s390.c (s390_expand_vec_compare): Support other
22967 vector floating point modes than just V2DF.
22968 (s390_expand_vcond): Likewise.
22969 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
22970 (s390_cannot_change_mode_class): Prevent mode changes between TF
22971 and V1TF in vector registers.
22972 * config/s390/s390.md (DF, SF): New mode attributes.
22973 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
22974 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
22975 SFmode support for VRs.
22976 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
22978 (VFT, VF_HW): New mode iterators.
22979 (vw, sdx): New mode attributes.
22980 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
22981 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
22982 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
22983 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
22984 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
22985 also the new vector floating point modes. Renaming to ...
22987 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
22988 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
22989 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
22990 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
22991 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
22992 ("vec_unordered<mode>"): ... these.
22994 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
22995 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
22996 ("*vec_extendv2df"): New insn definitions.
22998 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23000 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
23001 ("mulditi3_2", "*muldi3_sign"): New patterns.
23002 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
23003 rename the pattern definition.
23005 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23007 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
23009 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
23011 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23013 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
23014 instruction if possible.
23015 * config/s390/vector.md (vec_halfnumelts): New mode
23017 ("*vec_vllezlf<mode>"): New pattern.
23019 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23021 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
23022 ("popcountv4si2", "popcountv2di2"): Rename to ...
23023 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
23024 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
23026 ("popcount<mode>2_vxe"): New pattern.
23028 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23030 * common/config/s390/s390-common.c (processor_flags_table): Add
23032 * config.gcc: Add arch12.
23033 * config/s390/driver-native.c (s390_host_detect_local_cpu):
23034 Default to arch12 for unknown CPU model numbers.
23035 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
23036 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
23037 PROCESSOR_max sanity check.
23038 * config/s390/s390-opts.h (enum processor_type): Add
23040 * config/s390/s390.c (processor_table): Add arch12.
23041 (s390_expand_builtin): Add check for B_VXE flag.
23042 (s390_issue_rate): Add PROCESSOR_ARCH12.
23043 (s390_get_sched_attrmask): Likewise.
23044 (s390_get_unit_mask): Likewise.
23045 (s390_sched_score): Enable z13 scheduling for arch12.
23046 (s390_sched_reorder): Likewise.
23047 (s390_sched_variable_issue): Likewise.
23048 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
23050 (s390_tune_attr): Use z13 scheduling also for arch12.
23051 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
23052 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
23053 (TARGET_VXE_P): New macros.
23054 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
23055 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
23056 * config/s390/s390.opt: Add arch12 as processor_type.
23058 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23060 * config/s390/s390.md
23061 ("fixuns_truncdddi2", "fixuns_trunctddi2")
23062 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
23063 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
23065 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
23066 Rename expanders to ...
23068 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
23069 ("fixuns_truncdddi2_emu"): ... these.
23071 ("fixuns_trunc<mode>si2_emu"): New expander.
23073 ("*fixuns_truncdfdi2_z13"): Rename to ...
23074 ("*fixuns_truncdfdi2_vx"): ... this.
23076 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23078 * config/s390/2964.md: Remove the single element vector compare
23079 instructions which are no longer used.
23080 * config/s390/s390.c (s390_select_ccmode): Remove handling of
23082 (s390_canonicalize_comparison): Remove handling of DFmode
23084 (s390_expand_vec_compare_scalar): Remove function.
23085 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
23086 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
23088 ("*cmp<mode>_ccs"): Add wfcdb instruction.
23090 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23092 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
23094 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
23095 will anyway by matched by mov<mode>_64dfp.
23097 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23099 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
23100 vlef/vstef. Add missing operand to vleif.
23102 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23104 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
23105 pair for all vector types with 64 bit elements.
23106 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
23107 * config/s390/vector.md (V_HW_64): ... here.
23108 (V_128_NOSINGLE): New mode iterator.
23109 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
23110 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
23111 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
23112 ("*vec_load_pairv2di"): Change to ...
23113 ("*vec_load_pair<mode>"): ... this one.
23115 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23117 * config/s390/constraints.md: Add comments.
23118 (jKK): Reject element sizes > 8 bytes.
23119 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
23121 * config/s390/s390.md: Add the s_operand checks formerly in
23122 s390_split_ok_p to various splitters where they are still
23124 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
23125 for 128 bit vectors. Plus two splitters.
23127 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23129 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
23132 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23135 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
23136 error if the boundary argument is not constant.
23138 2017-03-24 Jakub Jelinek <jakub@redhat.com>
23140 PR rtl-optimization/80112
23141 * loop-doloop.c (doloop_condition_get): Don't check condition
23142 if cmp isn't SET with IF_THEN_ELSE src.
23144 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23146 PR tree-optimization/80158
23147 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
23148 replacing a candidate statement, also replace it for the
23149 candidate's alternate interpretation.
23150 (replace_rhs_if_not_dup): Likewise.
23151 (replace_one_candidate): Likewise.
23153 2017-03-24 Richard Biener <rguenther@suse.de>
23155 PR tree-optimization/80167
23156 * graphite-isl-ast-to-gimple.c
23157 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
23159 (translate_isl_ast_to_gimple::get_rename): Likewise.
23161 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23163 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
23164 handling of certain combinations of target options, including the
23165 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
23166 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
23168 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23171 * config/arm/arm.md (*load_multiple): Add reload_completed to
23172 matching condition.
23174 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23175 Richard Biener <rguenth@suse.de>
23177 PR tree-optimization/79908
23178 PR tree-optimization/80136
23179 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
23180 been cast away, gimplify_and_add suffices.
23182 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
23184 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
23186 2017-03-23 Richard Biener <rguenther@suse.de>
23188 PR tree-optimization/80032
23189 * gimplify.c (gimple_push_cleanup): Forced unconditional
23190 cleanups still have to go to the conditional_cleanups
23193 2017-03-22 Jakub Jelinek <jakub@redhat.com>
23195 PR tree-optimization/80072
23196 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
23198 (next_operand_entry_id): Change type to unsigned int.
23199 (sort_by_operand_rank): Make sure to return the right return value
23200 even if unsigned fields are bigger than INT_MAX.
23201 (struct oecount): Change cnt and id type to unsigned int.
23202 (oecount_hasher::equal): Formatting fix.
23203 (oecount_cmp): Make sure to return the right return value
23204 even if unsigned fields are bigger than INT_MAX.
23205 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
23208 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
23209 TREE_READONLY on result if writing it more than once.
23212 * doc/invoke.texi (-fsanitize=thread): Document that with
23213 -fnon-call-exceptions atomics are not able to throw
23217 * tsan.c: Include tree-eh.h.
23218 (instrument_builtin_call): Call maybe_clean_eh_stmt or
23219 maybe_clean_or_replace_eh_stmt where needed.
23220 (instrument_memory_accesses): Add cfg_changed argument.
23221 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
23222 if it returned true.
23223 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
23225 PR rtl-optimization/63191
23226 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
23227 wrapper function, moved the whole old content into ...
23228 (ix86_delegitimize_address_1): ... this. New inline function.
23229 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
23230 true as last argument instead of ix86_delegitimize_address.
23232 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
23234 * config/aarch64/aarch64.c (generic_branch_cost): Copy
23235 cortexa57_branch_cost.
23237 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
23239 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
23241 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23244 * doc/md.texi (Constraints): Document wA constraint.
23245 * config/rs6000/constraints.md (wA): New.
23246 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
23247 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
23248 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
23249 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
23251 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
23254 * gimplify.c (is_oacc_declared): New function.
23255 (oacc_default_clause): Use it to set default flags for acc declared
23256 variables inside parallel regions.
23257 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
23258 declared variables.
23259 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
23260 declare attribute to any decl as necessary.
23262 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
23265 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
23266 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
23267 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
23268 (arm_arch_lpae): This.
23269 * config/arm/arm.c (arm_arch7ve): Rename into ...
23270 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
23271 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
23274 2017-03-22 Martin Liska <mliska@suse.cz>
23277 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
23278 error message instead of an ICE.
23280 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23282 * doc/extend.texi (6.11 Additional Floating Types): Revise.
23284 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
23286 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23288 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23291 2017-03-21 Martin Sebor <msebor@redhat.com>
23293 * doc/extend.texi: Use "cannot" instead of "can't."
23294 * doc/hostconfig.texi: Same.
23295 * doc/install.texi: Same.
23296 * doc/invoke.texi: Same.
23297 * doc/loop.texi: Same.
23298 * doc/md.texi: Same.
23299 * doc/objc.texi: Same.
23300 * doc/rtl.texi: Same.
23301 * doc/tm.texi: Same.
23302 * doc/tm.texi.in: Same.
23303 * doc/trouble.texi: Same.
23305 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
23308 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
23309 (collect_checksum_attributes): Set it.
23310 (die_checksum_ordered): Use it.
23312 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23314 PR tree-optimization/79908
23315 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
23316 change: For a VA_ARG whose LHS has been cast away, use
23317 force_gimple_operand to construct the side effects.
23319 2017-03-21 David Malcolm <dmalcolm@redhat.com>
23321 PR translation/80001
23322 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
23323 more amenable to translation.
23324 (oacc_loop_auto_partitions): Likewise.
23326 2017-03-21 Marek Polacek <polacek@redhat.com>
23327 Martin Sebor <msebor@redhat.com>
23329 PR tree-optimization/80109
23330 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
23331 on INTEGRAL_TYPE_P.
23333 2017-03-21 Jakub Jelinek <jakub@redhat.com>
23334 Segher Boessenkool <segher@kernel.crashing.org>
23337 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
23338 check reg_used_between_p between insn and one of succ or succ2
23339 depending on if succ is artificial insn not inserted into insn
23342 2017-03-21 Martin Liska <mliska@suse.cz>
23344 PR gcov-profile/80081
23345 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
23346 * doc/gcc.texi: Include gcov-dump stuff.
23347 * doc/gcov-dump.texi: New file.
23349 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
23351 PR rtl-optimization/79150
23352 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
23353 conditional jump, if the jump is the last insn of the loop.
23355 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23356 Richard Biener <rguenth@suse.de>
23358 PR tree-optimization/79908
23359 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
23360 been cast away, use force_gimple_operand to construct the side
23363 2017-03-21 Martin Liska <mliska@suse.cz>
23365 PR libfortran/79956
23366 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
23369 2017-03-21 Brad Spengler <spender@grsecurity.net>
23372 * plugin.c (htab_hash_plugin): New function.
23373 (add_new_plugin): Use it and adjust.
23374 (parse_plugin_arg_opt): Adjust.
23375 (init_one_plugin): Likewise.
23377 2017-03-21 Richard Biener <rguenther@suse.de>
23379 PR tree-optimization/80032
23380 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
23381 if set force the cleanup to happen unconditionally.
23382 (gimplify_target_expr): Push inserted clobbers with force_uncond
23383 to avoid them being removed by control-dependent DCE.
23385 2017-03-21 Richard Biener <rguenther@suse.de>
23387 PR tree-optimization/80122
23388 * tree-inline.c (copy_bb): Do not expans va-arg packs or
23389 va_arg_pack_len when the inlined call stmt requires pack
23391 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
23393 2017-03-21 Jakub Jelinek <jakub@redhat.com>
23396 * tsan.c (instrument_builtin_call): If the memory model argument
23397 is not a constant, assume it is valid.
23400 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
23403 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
23405 PR rtl-optimization/79910
23406 * combine.c (can_combine_p): Do not allow combining an I0 or I1
23407 if its dest is used by an insn before I2 (other than the combined
23408 insns themselves, which are properly handled already).
23410 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
23413 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
23415 * combine.c (record_used_regs): New static function.
23416 (try_combine): Handle situations where there is an additional
23417 instruction between I2 and I3 which needs to have a LOG_LINK
23421 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
23423 * combine.c (try_combine): Delete redundant i1 test. Call
23424 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
23426 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23429 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
23430 alternatives 13/14.
23432 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23434 PR tree-optimization/80054
23435 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
23436 the optimization if a PHI or any of its arguments is not dominated
23437 by the candidate's basis. Use gphi* rather than gimple* as
23439 (replace_profitable_candidates): Clean up a gimple* variable that
23440 should be a gphi* variable.
23442 2017-03-20 Martin Sebor <msebor@redhat.com>
23445 * doc/extend.texi (attribute constructor): Document present limitation.
23447 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
23450 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
23451 __POWER9_VECTOR__ #ifdef control, change template definition to
23452 use Power9-specific built-in function.
23453 (vec_any_eq): Likewise.
23454 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
23455 to control outcomes from this test.
23456 (vector_ae_<mode>p): For VEC_F modes, likewise.
23458 2017-03-20 Ian Lance Taylor <iant@google.com>
23460 * config/i386/i386.c (ix86_function_regparm): Save an extra
23461 register for -fsplit-stack with DECL_STATIC_CHAIN.
23463 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23466 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
23467 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
23469 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23471 * config/riscv/riscv.c (riscv_print_operand): Use "fence
23473 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
23476 2017-03-20 Marek Polacek <polacek@redhat.com>
23479 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
23481 2017-03-20 Richard Biener <rguenther@suse.de>
23483 PR tree-optimization/80113
23484 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
23485 allocate extra SSA name for PHI def.
23486 (add_close_phis_to_outer_loops): Likewise.
23487 (add_close_phis_to_merge_points): Likewise.
23488 (copy_loop_close_phi_args): Likewise.
23489 (copy_cond_phi_nodes): Likewise.
23491 2017-03-20 Martin Liska <mliska@suse.cz>
23493 PR middle-end/79753
23494 * tree-chkp.c (chkp_build_returned_bound): Do not build
23495 returned bounds for a LHS that's not a BOUNDED_P type.
23497 2017-03-20 Martin Liska <mliska@suse.cz>
23501 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
23502 COMPLEX_CST and VECTOR_CST.
23504 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23507 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
23508 target operand. A new splitter adds the clobber statement in case
23509 the target operand is dead anyway.
23511 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
23513 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
23514 to age-old versions of binutils and glibc.
23516 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
23518 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
23520 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
23522 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
23524 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
23526 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
23527 requirement for binutils 2.13.
23529 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
23531 * combine.c (try_combine): Delete redundant i1 test. Call
23532 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
23534 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
23536 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
23537 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
23539 <riscv64-*-elf>: Re-arrange section
23540 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
23541 <riscv32-*-linux>: Likewise.
23542 <riscv64-*-elf>: Likewise
23543 <riscv64-*-linux>: Likewise.
23545 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
23548 * aarch64.opt(verbose-cost-dump): Fix typo.
23550 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
23553 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
23554 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
23556 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
23558 * reload.c (find_reloads): When reloading a nonoffsettable address,
23559 use RELOAD_OTHER for it and its address reloads.
23561 PR rtl-optimization/79910
23562 * combine.c (record_used_regs): New static function.
23563 (try_combine): Handle situations where there is an additional
23564 instruction between I2 and I3 which needs to have a LOG_LINK
23567 2017-03-17 Jeff Law <law@redhat.com>
23569 PR tree-optimization/71437
23570 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
23571 conditional in the hash table first.
23572 (vrp_dom_walker::before_dom_children): Extract condition from
23573 ASSERT_EXPR. Record condition, its inverion and any implied
23574 conditions as well.
23576 2017-03-17 Marek Polacek <polacek@redhat.com>
23577 Markus Trippelsdorf <markus@trippelsdorf.de>
23579 PR tree-optimization/80079
23580 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
23583 2017-03-17 Richard Biener <rguenther@suse.de>
23585 PR middle-end/80075
23586 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
23587 Properly verify the LHS before the RHS possibly claims to be
23589 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
23592 2017-03-17 Martin Jambor <mjambor@suse.cz>
23594 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
23595 (List of -O2 options): Likewise.
23596 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
23599 2017-03-17 Tom de Vries <tom@codesourcery.com>
23601 * gcov-dump.c (print_usage): Print bug_report_url.
23603 2017-03-17 Richard Biener <rguenther@suse.de>
23605 PR middle-end/80050
23606 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
23607 (parser::peek): Likewise.
23609 2017-03-17 Richard Biener <rguenther@suse.de>
23611 PR tree-optimization/80048
23612 * sese.c (free_sese_info): Properly release rename_map and
23613 copied_bb_map elements.
23615 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
23617 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
23618 Add linked-list forward and backlinks. Insert on
23619 construction, remove on destruction.
23620 (class pass_store_merging): Add m_stores_head field.
23621 (pass_store_merging::terminate_and_process_all_chains):
23622 Iterate over m_stores_head list.
23623 (pass_store_merging::terminate_all_aliasing_chains):
23625 (pass_store_merging::execute): Check for debug stmts first.
23626 Push new chains onto the m_stores_head stack.
23628 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
23631 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
23632 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
23633 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
23635 2017-03-16 Jeff Law <law@redhat.com>
23637 PR tree-optimization/71437
23638 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
23639 member function. Implementation moved into after_dom_children
23640 member function and into the threader's thread_outgoing_edges
23642 (dom_opt_dom_walker::after_dom_children): Simplify by moving
23643 some code into new thread_outgoing_edges.
23644 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
23645 definition. Simplify marker handling (do it here). Assume we always
23646 have the available expression and the const/copies tables.
23647 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
23649 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
23650 * tree-vrp.c (equiv_stack): No longer file scoped.
23651 (vrp_dom_walker): New class.
23652 (vrp_dom_walker::before_dom_children): New member function.
23653 (vrp_dom_walker::after_dom_children): Likewise.
23654 (identify_jump_threads): Setup domwalker. Use it rather than
23655 walking edges in a random order by hand. Simplify setup/finalization.
23656 (finalize_jump_threads): Remove.
23657 (vrp_finalize): Do not call identify_jump_threads here.
23658 (execute_vrp): Do it here instead and call thread_through_all_blocks
23661 PR tree-optimization/71437
23662 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
23664 (simplify_stmt_for_jump_threading): Add basic_block argument. All
23666 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
23667 (dom_opt_dom_walker::thread_across_edge): Remove
23668 handle_dominating_asserts argument. All callers changed.
23669 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
23670 changes. Remove calls to lhs_of_dominating_assert. Other
23671 uses of handle_dominating_asserts turn into unconditional code
23672 (simplify_control_stmt_condition_1): Likewise.
23673 (simplify_control_stmt_condition): Likewise.
23674 (thread_through_normal_block, thread_across_edge): Likewise.
23675 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
23676 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
23677 object if it is not an SSA_NAME.
23678 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
23679 before calling into the VRP specific simplifiers.
23680 (identify_jump_threads): Remove handle_dominating_asserts
23683 2017-03-16 Jakub Jelinek <jakub@redhat.com>
23686 * tree-diagnostic.c (default_tree_printer): No longer static.
23687 * tree-diagnostic.h (default_tree_printer): New prototype.
23689 2017-03-16 Tamar Christina <tamar.christina@arm.com>
23691 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
23692 Change ins into fmov.
23694 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23696 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
23697 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
23698 Use h_con constraint for operand 1.
23699 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
23700 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
23702 2017-03-15 Jeff Law <law@redhat.com>
23704 PR tree-optimization/71437
23705 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
23706 (record_temporary_equivalences): Use it.
23708 PR tree-optimization/71437
23709 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
23710 tree-ssa-scopedtables.
23711 (lookup_avail_expr, build_and_record_new_cond): Likewise.
23712 (record_conditions, record_cond, vuse_eq): Likewise.
23713 (record_edge_info): Adjust to API tweak of record_conditions.
23714 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
23715 (record_temporary_equivalences, optimize_stmt): Likewise.
23716 (eliminate_redundant_computations): Likewise.
23717 (record_equivalences_from_stmt): Likewise.
23718 * tree-ssa-scopedtables.c: Include options.h and params.h.
23719 (vuse_eq): New function, moved from tree-ssa-dom.c
23720 (build_and_record_new_cond): Likewise.
23721 (record_conditions): Likewise. Accept vector of conditions rather
23722 than edge_equivalence structure for first argument.
23723 for the first argument.
23724 (avail_exprs_stack::lookup_avail_expr): New member function, moved
23725 from tree-ssa-dom.c.
23726 (avail_exprs_stack::record_cond): Likewise.
23727 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
23728 from tree-ssa-dom.c.
23729 (avail_exprs_stack): Add new member functions lookup_avail_expr
23731 (record_conditions): Declare.
23733 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
23736 * lra-constraints.c (process_alt_operands): Increase reject for
23737 reloading an input/output operand.
23739 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
23742 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
23743 insns to convert from signed/unsigned char/short to IEEE 128-bit
23745 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
23747 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
23750 * config/i386/i386.c (ix86_vector_duplicate_value): Create
23751 subreg of inner mode for values already in registers.
23753 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
23755 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
23756 iteration reg is used after the loop.
23758 2017-03-14 Martin Sebor <msebor@redhat.com>
23760 PR tree-optimization/79800
23761 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
23762 precision in negative-positive range.
23763 (format_floating): Call non-const overload with adjusted precision.
23765 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
23768 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
23771 2017-03-14 Martin Sebor <msebor@redhat.com>
23773 PR middle-end/80020
23774 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
23775 * builtins.def (aligned_alloc): Use it.
23778 * Makefile.in (GTFILES): Add calls.c.
23779 * calls.c: Include "gt-calls.h".
23781 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
23783 PR rtl-optimization/79728
23784 * regs.h (struct target_regs): New field
23785 x_contains_allocatable_regs_of_mode.
23786 (contains_allocatable_regs_of_mode): New macro.
23787 * reginfo.c (init_reg_sets_1): Initialize it, and change
23788 contains_reg_of_mode so it includes global regs as well.
23789 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
23790 rather than contains_regs_of_mode.
23792 2017-03-14 Martin Liska <mliska@suse.cz>
23794 * doc/invoke.texi: Document options that can't be combined with
23795 -fcheck-pointer-bounds.
23797 2017-03-14 Martin Liska <mliska@suse.cz>
23799 PR middle-end/79831
23800 * doc/invoke.texi (-Wchkp): Document the option.
23802 2017-03-14 Martin Liska <mliska@suse.cz>
23804 * Makefile.in: Install gcov-dump.
23806 2017-03-14 Martin Liska <mliska@suse.cz>
23808 * multiple_target.c (expand_target_clones): Bail out for
23809 an invalid attribute.
23811 2017-03-14 Richard Biener <rguenther@suse.de>
23813 * alias.c (struct alias_set_entry): Pack properly.
23814 * cfgloop.h (struct loop): Likewise.
23815 * cse.c (struct set): Likewise.
23816 * ipa-utils.c (struct searchc_env): Likewise.
23817 * loop-invariant.c (struct invariant): Likewise.
23818 * lra-remat.c (struct cand): Likewise.
23819 * recog.c (struct change_t): Likewise.
23820 * rtl.h (struct address_info): Likewise.
23821 * symbol-summary.h (function_summary): Likewise.
23822 * tree-loop-distribution.c (struct partition): Likewise.
23823 * tree-object-size.c (struct object_size_info): Likewise.
23824 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
23825 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
23826 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
23827 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
23828 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
23829 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
23830 (struct _stmt_vec_info): Likewise.
23832 2017-03-14 Martin Liska <mliska@suse.cz>
23835 * multiple_target.c (create_dispatcher_calls): Check that
23836 a target can create a function dispatcher.
23838 2017-03-14 Martin Liska <mliska@suse.cz>
23841 * multiple_target.c (expand_target_clones): Drop local.local
23842 flag for default implementation.
23844 2017-03-14 Richard Biener <rguenther@suse.de>
23846 PR tree-optimization/80030
23847 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
23849 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
23851 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
23852 gcc_fallthrough() instead of __attribute__((fallthrough));
23854 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23856 * doc/gcc.texi: Remove "up" link to (DIR).
23857 * doc/gccint.texi: Ditto.
23859 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
23861 * doc/install.texi (Specific) <avr>: Remove reference to
23864 2017-03-13 Jeff Law <law@redhat.com>
23866 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
23867 attribute rather than comments.
23869 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
23870 match_scratch operand is highest.
23872 2017-03-13 Martin Liska <mliska@suse.cz>
23874 PR middle-end/78339
23875 * ipa-pure-const.c (warn_function_noreturn): If the declarations
23876 is a CHKP clone, use original declaration.
23878 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23880 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
23881 (arc_conditional_register_usage): Use a different allocation order
23882 when optimizing for size.
23883 * common/config/arc/arc-common.c (arc_option_optimization_table):
23884 Section anchors default on when optimizing for size.
23886 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23888 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
23890 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23892 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
23893 * config/arc/arc.md (cpu_facility): Add cd variant.
23894 (*movqi_insn): Add code density variant.
23895 (*movhi_insn): Likewise.
23896 (*movqi_insn): Likewise.
23897 (*addsi3_mixed): Likewise.
23898 (subsi3_insn): Likewise.
23900 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23902 * config/arc/arc.md (movsi_cond_exec): Update constraint.
23904 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
23906 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
23907 expressions with MINUS and UNARY ops.
23909 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23912 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
23914 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
23915 between vec_select and vector argument.
23916 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
23917 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
23918 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
23919 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
23920 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
23921 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
23923 2017-03-13 Richard Biener <rguenther@suse.de>
23926 * params.def (vect-max-peeling-for-alignment): Fix typo.
23928 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23930 * doc/install.texi (Specific) <mips-*-*>: Remove description of
23931 issue that only occurred with binutils below 2.18.
23933 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23935 * doc/install.texi (Specific) <cris-axis-elf>: No longer
23936 refer to binutils 2.11/2.12 minimum.
23938 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
23940 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
23941 ftp.kernel.org and simplify binutils requirement.
23943 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
23945 * doc/invoke.texi (Warning Options): Fix spelling of link-time
23947 (Optimize Options): Ditto. Also remove redundancy.
23949 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23951 PR translation/79848
23952 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
23954 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
23955 to G_ to avoid double translation.
23957 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23959 PR translation/79923
23960 * auto-profile.c (get_combined_location): Convert leading
23961 character of diagnostics to lower case and remove trailing period.
23962 (read_profile): Likewise for various diagnostics.
23963 * config/arm/arm.c (arm_option_override): Remove trailing period
23964 from various diagnostics.
23965 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
23966 (msp430_expand_delay_cycles): Likewise.
23968 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23971 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
23972 full command-line argument, rather than just "str".
23973 (aarch64_validate_march): Likewise.
23974 (aarch64_validate_mtune): Likewise.
23976 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
23978 PR rtl-optimization/78911
23979 * lra-assigns.c (must_not_spill_p): New function.
23980 (spill_for): Use it.
23982 2017-03-10 Jakub Jelinek <jakub@redhat.com>
23984 PR tree-optimization/79981
23985 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
23986 ATOMIC_COMPARE_EXCHANGE ifn result.
23987 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
23988 IFN_ATOMIC_COMPARE_EXCHANGE.
23990 2017-03-10 David Malcolm <dmalcolm@redhat.com>
23993 * opts.c (parse_sanitizer_options): Add missing question mark to
23994 "did you mean" message.
23996 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23998 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
24000 (VMULEUH_UNS): Likewise.
24001 (VMULOUB_UNS): Likewise.
24002 (VMULOUH_UNS): Likewise.
24003 * config/rs6000/rs6000.c (builtin_function_type): Remove
24004 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
24006 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24009 * read-rtl-function.c (function_reader::read_rtx_operand): Update
24010 x with result of extra_parsing_for_operand_code_0.
24011 (function_reader::extra_parsing_for_operand_code_0): Convert
24012 return type from void to rtx, returning x. When reading
24013 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
24014 larger size containing struct block_symbol.
24016 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
24018 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
24019 -mfloat128-hardware without -m64.
24021 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
24024 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
24025 entries to the case statement that marks unsigned arguments to
24026 overloaded functions.
24028 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24030 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
24031 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
24033 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
24036 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
24037 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
24039 2017-03-10 Martin Liska <mliska@suse.cz>
24043 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
24044 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
24047 2017-03-10 Olivier Hainque <hainque@adacore.com>
24049 * tree-switch-conversion (array_value_type): Start by resetting
24050 candidate type to it's main variant.
24052 2017-03-10 Jakub Jelinek <jakub@redhat.com>
24054 PR rtl-optimization/79909
24055 * combine.c (try_combine): Use simplify_replace_rtx on individual
24056 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
24057 of the whole CALL_INSN_FUNCTION_USAGE.
24059 PR tree-optimization/79972
24060 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
24061 get_range_info on SSA_NAMEs. Formatting fixes.
24063 2017-03-10 Richard Biener <rguenther@suse.de>
24064 Jakub Jelinek <jakub@redhat.com>
24066 PR tree-optimization/77975
24067 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
24068 edge to be constant.
24069 (get_val_for): For constant x return it. Formatting fix.
24070 (loop_niter_by_eval): Avoid pointless looping if the next iteration
24071 would use the same bases as the current one.
24073 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24075 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
24076 instead of vec_select for V1TImode.
24077 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
24079 (VSX_LE_128): Add V1TI to this mode iterator.
24080 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
24081 (*vsx_le_perm_store_<mode>): Likewise.
24082 (pre-reload splitter for VSX stores): Likewise.
24083 (post-reload splitter for VSX stores): Likewise.
24084 (*vsx_xxpermdi2_le_<mode>): Likewise.
24085 (*vsx_lxvd2x2_le_<mode>): Likewise.
24086 (*vsx_stxvd2x2_le_<mode>): Likewise.
24088 2017-03-09 Michael Eager <eager@eagercon.com>
24090 Correct failures with --enable-checking=yes,rtl.
24092 * config/microblaze/microblaze.c (microblaze_expand_shift):
24093 Replace GET_CODE test with CONST_INT_P and INTVAL test with
24094 test for const0_rtx.
24095 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
24096 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
24098 2017-03-09 Richard Biener <rguenther@suse.de>
24100 PR tree-optimization/79977
24101 * graphite-scop-detection.c (scop_detection::merge_sese):
24102 Handle the case of extra exits to blocks dominating the entry.
24104 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
24106 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24109 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
24111 PR rtl-optimization/79949
24112 * lra-constraints.c (process_alt_operands): Check memory when
24113 trying to predict a cycle. Print about the overall increase.
24115 2017-03-09 Richard Biener <rguenther@suse.de>
24117 PR middle-end/79971
24118 * gimple-expr.c (useless_type_conversion_p): Preserve
24119 TYPE_SATURATING for fixed-point types.
24121 2017-03-09 Richard Biener <rguenther@suse.de>
24124 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
24125 alignment of BLKmode params.
24127 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24130 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
24131 (VALL_NO_V2Q): Likewise.
24133 * config/aarch64/aarch64-simd.md
24134 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
24136 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
24137 VALL_NO_V2Q mode iterator.
24138 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
24140 2017-03-09 Martin Liska <mliska@suse.cz>
24142 PR tree-optimization/79631
24143 * tree-chkp-opt.c (chkp_is_constant_addr): Call
24144 tree_int_cst_sign_bit just for INTEGER constants.
24146 2017-03-09 Martin Liska <mliska@suse.cz>
24150 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
24153 2017-03-09 Marek Polacek <polacek@redhat.com>
24156 * tree.c (inchash::add_expr): Handle TREE_VEC.
24158 2017-03-09 Martin Liska <mliska@suse.cz>
24161 (chkp_narrow_size_and_offset): New function.
24162 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
24163 (void chkp_parse_bit_field_ref): New function.
24164 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
24165 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
24167 2017-03-09 Martin Liska <mliska@suse.cz>
24170 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
24171 (chkp_find_bounds_1): Remove gcc_unreachable.
24173 2017-03-09 Jakub Jelinek <jakub@redhat.com>
24176 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
24177 BUILT_IN_SYNC*, determine the access type from the size suffix and
24178 always build a MEM_REF with that type. Handle forgotten
24179 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
24182 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
24183 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
24184 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
24185 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
24186 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
24187 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
24188 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
24189 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
24190 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
24191 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
24192 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
24193 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
24194 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
24195 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
24196 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
24197 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
24198 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
24199 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
24200 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
24201 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
24202 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
24203 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
24204 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
24205 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
24206 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
24207 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
24208 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
24209 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
24210 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
24211 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
24212 definitions outside of __OPTIMIZE__ guarded section.
24215 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
24216 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
24217 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
24218 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
24221 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24223 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
24224 ("vfenez<mode>"): Add missing constraints.
24226 2017-03-08 Martin Sebor <msebor@redhat.com>
24229 * config/nds32/nds32.c (nds32_option_override):
24230 Fix misspelled diagnostic.
24232 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24235 * gimplify.c (gimplify_omp_for): Replace index var in outer
24236 taskloop statement with an artificial variable and add
24237 OMP_CLAUSE_PRIVATE clause for it.
24239 2017-03-08 Richard Biener <rguenther@suse.de>
24241 PR tree-optimization/79955
24242 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
24243 for accesses that are completely outside of the variable.
24245 2017-03-08 Andrew Haley <aph@redhat.com>
24247 PR tree-optimization/79943
24248 * tree-ssa-loop-split.c (compute_new_first_bound): When
24249 calculating the new upper bound, (END-BEG) should be added, not
24252 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24254 * config/avr/avr.md (setmemhi): Make sure match_dup
24255 operand number comes before match_scratch.
24257 2017-03-08 Richard Biener <rguenther@suse.de>
24259 PR tree-optimization/79920
24260 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
24261 with ncopies == 1 to ...
24262 (vect_transform_slp_perm_load): ... here. Properly compute
24263 all element loads by iterating VF times over the group. Do
24264 not handle ncopies (computed in a broken way) in
24265 vect_create_mask_and_perm.
24267 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24270 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
24271 is a uniform vector, use uniform_vector_p return value instead of
24272 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
24274 2017-03-07 Marek Polacek <polacek@redhat.com>
24276 PR middle-end/79809
24277 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
24278 (alloca_call_type): Likewise.
24280 2017-03-07 Martin Liska <mliska@suse.cz>
24282 * gcov.c (process_args): Put comment to correct location.
24284 2017-03-07 Martin Liska <mliska@suse.cz>
24286 PR middle-end/68270
24287 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
24288 Use array_at_struct_end_p instead of DECL_CHAIN (field).
24289 (chkp_narrow_bounds_for_field): Likewise.
24290 (chkp_parse_array_and_component_ref): Pass one more argument to
24293 2017-03-07 Richard Biener <rguenther@suse.de>
24295 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
24298 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
24300 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
24301 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
24303 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24306 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
24307 to end of description.
24308 (PARAM_MAX_STORES_TO_MERGE): Likewise.
24310 2017-03-07 Jakub Jelinek <jakub@redhat.com>
24312 PR rtl-optimization/79901
24313 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
24315 (*avx512f_<code><mode>3<mask_name>): ... this.
24316 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
24317 iterator instead of VI8_AVX2_AVX512BW.
24319 PR rtl-optimization/79901
24320 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
24321 min/max expander, expand it using expand_vec_cond_expr.
24324 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
24327 2017-03-06 Jakub Jelinek <jakub@redhat.com>
24330 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
24331 to void * for PCH reasons.
24332 * dwarf2out.c (output_loc_operands, output_die): Cast
24333 v.val_vec.array to unsigned char *.
24335 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
24338 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
24341 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
24343 PR rtl-optimization/79571
24344 * lra-constraints.c (process_alt_operands): Calculate static
24345 reject and subtract it from overall when only addresses will be
24348 2017-03-06 Julia Koval <julia.koval@intel.com>
24351 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
24352 incoming stack boundary to 128 for 64-bit targets.
24354 2017-03-06 Richard Biener <rguenther@suse.de>
24356 PR tree-optimization/79894
24357 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
24358 to NULL after folding it.
24360 2017-03-06 Richard Biener <rguenther@suse.de>
24362 PR tree-optimization/79824
24363 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
24364 check disabling peeling for gaps.
24366 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
24368 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24369 attributes): Document gettimeofday.
24371 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
24373 * config/s390/s390.c (s390_option_override_internal): Set
24374 PARAM_MIN_VECT_LOOP_BOUND
24376 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
24378 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
24379 * config/s390/s390.md: Likewise.
24381 2017-03-06 Jakub Jelinek <jakub@redhat.com>
24384 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
24385 (<avx2_avx512>_perm<mode>): Rename to ...
24386 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
24388 (<avx512>_perm<mode>_mask): Rename to ...
24389 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
24391 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
24392 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
24393 instead of VI8F_256_512.
24394 (avx512f_perm<mode>): New define_expand.
24395 (avx512f_perm<mode>_mask): Likewise.
24396 (avx512f_perm<mode>_1<mask_name>): New define_insn.
24397 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
24399 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24401 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
24402 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
24404 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
24406 2017-03-06 Martin Liska <mliska@suse.cz>
24409 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
24410 when having a SSA NAME w/o VAR_DECL assigned to it.
24412 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24414 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
24415 msa_dpsub_<su>_d): Fix MODE for vec_select.
24417 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24419 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
24421 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
24423 2017-03-06 Richard Biener <rguenther@suse.de>
24425 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
24426 * plugin.c (register_plugin_info): Likewise.
24427 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
24429 2017-03-05 Jakub Jelinek <jakub@redhat.com>
24431 * config/i386/sse.md (sse_storehps, sse_storelps,
24432 avx_<castmode><avxsizesuffix>_<castmode>,
24433 avx512f_<castmode><avxsizesuffix>_<castmode>,
24434 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
24435 in condition that at least one operand is not a MEM.
24437 2017-03-03 Jakub Jelinek <jakub@redhat.com>
24439 PR middle-end/79805
24440 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
24441 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
24443 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
24444 gimple_call_nothrow_p flag based on whether original builtin can throw.
24445 If it can, emit following stmts on the fallthrough edge.
24446 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
24447 don't create new bb if inserting just debug stmts on the edge, try to
24448 insert them on the fallthru bb or just reset debug stmts.
24450 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
24453 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
24454 restore recog_data (including the operand rtxes inside it) around
24455 the call to get_insn_template.
24457 2017-03-03 Martin Sebor <msebor@redhat.com>
24459 PR tree-optimization/79699
24460 * context.c (context::~context): Free MPFR caches to avoid
24461 a memory leak on program exit.
24463 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24465 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
24466 Use wide_int::ulow () instead of .elt (0).
24468 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24470 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
24471 (*pushxf): Limit oF constraint to 32bit targets and add oC
24472 constraint for 64bit targets.
24473 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
24474 (*pushdf): Change rmF constraint to rmC.
24476 2017-03-03 Martin Liska <mliska@suse.cz>
24478 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
24479 Remove unused variable.
24481 2017-03-03 Jakub Jelinek <jakub@redhat.com>
24484 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
24485 is a memory operand, increase num_memory.
24486 (ix86_expand_args_builtin): Likewise.
24488 2017-03-03 Jan Hubicka <jh@suse.cz>
24491 * ipa-devirt.c (maybe_record_node): Properly handle
24492 __cxa_pure_virtual visibility.
24494 2017-03-03 Martin Liska <mliska@suse.cz>
24496 PR tree-optimization/79803
24497 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
24499 (pass_loop_prefetch::execute): Disabled optimization if an
24500 assumption about L1 cache size is not met.
24502 2017-03-03 Martin Liska <mliska@suse.cz>
24504 PR rtl-optimization/79574
24505 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
24506 (hash_scan_set): Likewise.
24507 (dump_hash_table): Likewise.
24508 (hoist_code): Likewise.
24510 2017-03-03 Richard Biener <rguenther@suse.de>
24512 * fixed-value.c (fixed_from_string): Restore use of elt (1)
24513 in place of uhigh ().
24514 (fixed_convert_from_real): Likewise.
24516 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24519 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
24521 2017-03-03 Richard Biener <rguenther@suse.de>
24523 PR middle-end/79818
24524 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
24525 TYPE_OVERFLOW_UNDEFINED check.
24527 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24529 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
24531 (vector_ae_<mode>_p): Likewise.
24532 (vector_nez_<mode>_p): Likewise.
24533 (vector_ne_v2di_p): Likewise.
24534 (vector_ae_v2di_p): Likewise.
24535 (vector_ne_<mode>_p): Likewise.
24536 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
24538 (vsx_tsqrt<mode>2_fe): Likewise.
24540 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
24543 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
24545 2017-03-02 Jakub Jelinek <jakub@redhat.com>
24547 PR rtl-optimization/79780
24548 * cprop.c (one_cprop_pass): When second and further conditional trap
24549 in a single basic block is turned into an unconditional trap, turn it
24550 into a deleted note to avoid RTL verification failures.
24552 2017-03-02 Richard Biener <rguenther@suse.de>
24554 * fold-const.c (const_binop): Use ulow () instead of elt (0).
24556 2017-03-02 Richard Biener <rguenther@suse.de>
24558 PR tree-optimization/79345
24560 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
24561 param and abort the walk, returning -1 if it is hit.
24562 (walk_aliased_vdefs): Take a limit param and pass it on.
24563 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
24564 defaulting to 0 and return a signed int.
24565 * tree-ssa-uninit.c (struct check_defs_data): New struct.
24566 (check_defs): New helper.
24567 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
24568 about uninitialized memory.
24569 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
24570 bogus uninitialized warning.
24571 (fixed_convert_from_real): Likewise.
24573 2017-03-02 Bin Cheng <bin.cheng@arm.com>
24575 PR tree-optimization/66768
24576 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
24577 iv_use if base object can't be determined.
24579 2017-03-02 Jakub Jelinek <jakub@redhat.com>
24581 PR tree-optimization/79345
24582 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
24583 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
24584 (get_pattern_stats): Initialize it.
24585 * genemit.c (gen_expand): Verify match_scratch numbers come after
24586 match_operand/match_dup numbers.
24587 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
24588 match_scratch numbers.
24589 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
24591 * config/s390/s390.md (trunctdsd2): Likewise.
24593 2017-03-02 Richard Biener <rguenther@suse.de>
24595 * wide-int.h (wide_int_storage::operator=): Implement in terms
24598 2017-03-02 Richard Biener <rguenther@suse.de>
24600 PR tree-optimization/79777
24601 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
24602 the to insert expression to sth existing.
24604 2017-03-01 Martin Sebor <msebor@redhat.com>
24606 PR middle-end/79692
24607 * gimple-ssa-sprintf.c
24608 (directive::known_width_and_precision): New function.
24609 (format_integer): Use it.
24610 (get_mpfr_format_length): Consider the full range of precision
24611 when computing %g output with the # flag. Set the likely byte
24612 count to 3 rather than 1 when precision is indeterminate.
24613 (format_floating): Correct the lower bound of precision.
24615 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24617 * doc/invoke.texi: Document default code model for 64-bit Linux.
24619 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24622 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
24623 udiv rather than div since input pattern is unsigned.
24625 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
24627 * config/i386/i386.c (print_reg): Warn for values of
24628 unsupported size in integer register.
24630 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
24633 * config/rs6000/predicates.md (current_file_function_operand): Do
24634 not allow self calls to be local if the function is replaceable.
24636 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
24639 * config/rs6000/altivec.h (vec_ctz and others): Change the
24640 preprocessor macro that controls conditional compilation from
24641 _ARCH_PWR9 to __POWER9_VECTOR__.
24642 (vec_all_ne): Change parameterization of __altivec_scalar_pred
24643 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
24644 control (instead of _ARCH_PWR9 control) so that template
24645 definition uses power9-specific function.
24646 (vec_any_eq): Likewise.
24647 (vec_all_ne): Change macro definition to use a power9-specific
24648 expansion under #ifdef __POWER9_VECTOR__ control (instead of
24649 _ARCH_PWR9 control).
24650 (vec_any_eq) Likewise.
24651 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
24652 expansion for CMPNEF to remove support for xvcmpnesp instruction.
24653 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
24654 support for xvcmpnedp instruction.
24655 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
24656 macro expansion so that Power9 implementation of vec_all_ne does
24657 not use the AltiVec predicate framework.
24658 (VCMPNEH_P): Likewise.
24659 (VCMPNEW_P): Likewise.
24660 (VCMPNED_P): Likewise.
24661 (VCMPNEFP_P): Likewise.
24662 (VCMPNEDP_P): Likewise.
24663 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
24664 implementation of vec_any_eq to not use AltiVec predicate
24666 (VCMPAEH_P): Likewise.
24667 (VCMPAEW_P): Likewise.
24668 (VCMPAED_P): Likewise.
24669 (VCMPAEFP_P): Likewise.
24670 (VCMPAEDP_P): Likewise.
24671 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
24672 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
24673 not use the AltiVec predicate framework.
24674 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
24675 of vec_any_eq to not use AltiVec predicate framework.
24676 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
24677 support for predefined __POWER9_VECTOR__ macro to indicate that
24678 Power9 instruction selection is enabled.
24679 (altivec_overloaded_builtins): Remove extraneous
24680 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
24681 function argument types RS6000_BTI_bool_V16QI and
24682 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
24683 entry for overloaded function argument types RS6000_BTI_bool_V4SI
24684 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
24685 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
24686 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
24687 Power9 for implementations of vec_cmpne. Change the signature for
24688 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
24689 (representing vec_all_ne) to remove the previously described first
24690 argument of type RS6000_BTI_INTSI, as this was an artifact of
24691 reliance on the AltiVec predicate framework, which is no longer
24692 used in the implementation of these functions. Add
24693 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
24694 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
24695 since, unlike the AltiVec predicate framework implementation, we
24696 do not share function descriptors between vec_alle and vec_anyeq.
24697 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
24698 set of modes that receive special treatment even when
24699 TARGET_P9_VECTOR is true. The special treatment emits code that
24700 does not depend on Power9 instructions.
24701 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
24702 define_expand to not rely on AltiVec predicate framework.
24703 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24705 (vector_ne_v2di_p): Change this define_expand to not rely on
24706 AltiVec predicate framework.
24707 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
24709 (vector_ne_<mode>_p): Change this define_expand to not rely on
24710 AltiVec predicate framework.
24711 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
24713 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
24714 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
24715 define_insn pattern.
24716 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
24717 define_insn pattern because the xvcmpne<VSs>. instruction is not
24719 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
24720 instruction is not supported.
24722 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24724 * config/nvptx/nvptx.c: Include intl.h.
24726 2017-03-01 Martin Jambor <mjambor@suse.cz>
24729 * ipa-prop.h (ipa_bits): Removed field known.
24730 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
24731 to pointers. Adjusted their comments to warn about their sharing.
24732 (ipcp_transformation_summary): Change bits to a vector of pointers.
24733 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
24734 (ipa_get_ipa_bits_for_value): Declare.
24735 * tree-vrp.h (value_range): Mark as GTY((for_user)).
24736 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
24737 (ipa_bits_hash_table): Likewise.
24738 (ipa_vr_ggc_hash_traits): Likewise.
24739 (ipa_vr_hash_table): Likewise.
24740 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
24741 being pointers and vr_known being removed.
24742 (ipa_set_jf_unknown): Likewise.
24743 (ipa_get_ipa_bits_for_value): New function.
24744 (ipa_set_jfunc_bits): Likewise.
24745 (ipa_get_value_range): New overloaded functions.
24746 (ipa_set_jfunc_vr): Likewise.
24747 (ipa_compute_jump_functions_for_edge): Use the above functions to
24748 construct bits and vr parts of jump functions.
24749 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
24750 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24752 (ipcp_grow_transformations_if_necessary): Also allocate
24753 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
24755 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
24756 them. Fix too long lines.
24757 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
24758 vr_known being removed.
24759 (ipa_read_jump_function): Use new setter functions to construct bits
24760 and vr parts of jump functions or set them to NULL.
24761 (write_ipcp_transformation_info): Adjust for bits being pointers.
24762 (read_ipcp_transformation_info): Likewise.
24763 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
24765 Include gt-ipa-prop.h.
24766 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
24768 (ipcp_store_bits_results): Likewise.
24769 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
24770 Do not write to existing jump functions but use a temporary instead.
24772 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24775 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
24776 attempt to use its first operand as BIT_FIELD_REF base.
24778 2017-03-01 Richard Biener <rguenther@suse.de>
24780 PR middle-end/79721
24781 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
24782 interpolating formula in wrapping arithmetic.
24783 (chrec_apply): Convert chrec_evaluate return value to wanted type.
24785 2017-03-01 Jakub Jelinek <jakub@redhat.com>
24787 PR tree-optimization/79734
24788 * tree-vect-generic.c (expand_vector_condition): Optimize
24789 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
24790 Handle VEC_COND_EXPR where comparison has different inner width from
24791 type's inner width.
24793 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
24795 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
24796 markup, and similar issues. Remove @opindex entries for things
24797 that aren't options. Add missing -mmpy-option entries.
24799 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24801 PR tree-optimization/79737
24802 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
24803 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
24804 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
24805 instead of byte_size. Formatting fix.
24806 (shift_bytes_in_array_right): Formatting fix.
24808 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
24811 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
24812 condition on optimize for the leaf function test.
24814 2017-02-28 Martin Liska <mliska@suse.cz>
24817 * read-rtl-function.c (function_reader::handle_unknown_directive):
24818 Bail out when one uses -flto.
24820 2017-02-28 Martin Liska <mliska@suse.cz>
24822 * common.opt: Replace space with tabular for options of <number>
24824 * config/i386/i386.opt: Show <number> value for
24825 -mlarge-data-threshold.
24826 * opts.c (print_filtered_help): Do not display number in hexadecimal
24829 2017-02-28 Martin Liska <mliska@suse.cz>
24831 * common.opt: Fix --help=option -Q for options which are of
24834 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
24836 * config/i386/i386.c (print_reg): Error out for values
24837 of 8-bit size in invalid integer register.
24839 2017-02-28 Martin Sebor <msebor@redhat.com>
24841 PR tree-optimization/79691
24842 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
24844 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24847 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
24848 gcc_unreachable with output_operand_lossage.
24850 2017-02-28 Richard Biener <rguenther@suse.de>
24852 PR tree-optimization/79740
24853 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
24855 (visit_nary_op): Insert the nary into the hashtable if we
24856 pattern-matched sth.
24857 * tree-ssa-pre.c (eliminate_insert): Robustify.
24859 2017-02-28 Richard Biener <rguenther@suse.de>
24861 PR middle-end/79731
24862 * fold-const.c (decode_field_reference): Reject out-of-bound
24865 2017-02-28 Jakub Jelinek <jakub@redhat.com>
24867 * config/i386/i386.c: Include intl.h.
24868 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
24869 instead of just cond ? "..." : "...".
24870 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
24871 * coverage.c (read_counts_file): Likewise.
24872 * omp-offload.c: Include intl.h.
24873 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
24874 of just cond ? "..." : "...".
24875 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
24876 of just cond ? "..." : "...".
24878 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
24881 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
24883 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
24884 'tune for' CPU name.
24885 * config/arm/arm-cpu-data.h: Regenerated.
24887 2017-02-28 Richard Biener <rguenther@suse.de>
24889 PR tree-optimization/79732
24890 * tree-inline.c (expand_call_inline): Do not shadow var.
24892 2017-02-28 Richard Biener <rguenther@suse.de>
24894 PR tree-optimization/79723
24895 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
24896 address-space properly.
24898 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
24900 * doc/optinfo.texi (Optimization groups): Fix option used for
24902 * doc/invoke.texi (-fopt-info): Document "omp".
24903 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
24904 (OPTGROUP_ALL): Add OPTGROUP_OMP.
24905 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
24906 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
24907 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
24909 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
24911 * dumpfile.c (optgroup_options): Instead of "openmp", associate
24912 OPTGROUP_OMP with "omp".
24914 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
24917 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
24918 for arithmetic shift of unsigned V2DI.
24920 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
24922 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
24923 arc/linux.h headers.
24924 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
24925 (LINK_SPEC): Likewise.
24926 (ARC_TLS_EXTRA_START_SPEC): Likewise.
24927 (EXTRA_SPECS): Likewise.
24928 (STARTFILE_SPEC): Likewise.
24929 (ENDFILE_SPEC): Likewise.
24930 (LIB_SPEC): Likewise.
24931 (TARGET_SDATA_DEFAULT): Likewise.
24932 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
24933 (MULTILIB_DEFAULTS): Likewise.
24934 (DWARF2_UNWIND_INFO): Likewise.
24935 * config/arc/big.h: New file.
24936 * config/arc/elf.h: Likewise.
24937 * config/arc/linux.h: Likewise.
24938 * config/arc/t-uClibc: Remove.
24940 2017-02-27 Bin Cheng <bin.cheng@arm.com>
24942 PR tree-optimization/77536
24943 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
24944 (tree_transform_and_unroll_loop): Use above function to compute the
24945 estimated niter of unrolled loop and use it when scaling profile.
24946 Also use count info rather than frequency if it's non-zero.
24947 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
24948 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
24949 (vect_transform_loop): Call above function.
24951 2017-02-27 Richard Biener <rguenther@suse.de>
24953 PR tree-optimization/45397
24954 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
24955 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
24956 (visit_nary_op): Add pattern matching for CSEing sign-changed
24957 or truncated operations with wider ones.
24959 2017-02-27 Richard Biener <rguenther@suse.de>
24961 PR tree-optimization/79690
24962 * tree-vect-stmts.c (vectorizable_store): Use vector type
24963 built from the DR with address-space.
24965 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
24967 * doc/invoke.texi (Optimize Options): Refine the description
24968 of asan-use-after-return.
24970 2017-02-25 Alan Modra <amodra@gmail.com>
24972 PR rtl-optimization/79584
24973 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
24974 base, not ad->base_term, the reg within base. Remove assertion
24975 that ad->base == ad->base_term. Replace gen_int_mode using
24976 bogus mode with const0_rtx.
24978 2017-02-25 Jakub Jelinek <jakub@redhat.com>
24980 PR middle-end/79396
24981 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
24982 FMA_EXPR like tcc_binary or tcc_unary.
24984 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
24987 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
24989 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
24990 (output_loc_operands): Handle DW_OP_call_ref and
24991 DW_OP_GNU_variable_value.
24992 (struct variable_value_struct): New type.
24993 (struct variable_value_hasher): Likewise.
24994 (variable_value_hash): New variable.
24995 (string_types): Remove.
24996 (copy_loc_descr): New function.
24997 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
24998 (prepend_loc_descr_to_each): New function.
24999 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
25000 instead of add_loc_descr_to_each if the first argument is single
25001 location list and the second has multiple.
25002 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
25003 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
25004 when looking for variable value which doesn't have other location info.
25005 (loc_list_from_tree): Formatting fix.
25006 (gen_array_type_die): Simplify DW_AT_string_length handling.
25007 (adjust_string_types): Remove.
25008 (gen_subprogram_die): Don't call adjust_string_types nor test/set
25009 string_types. Call resolve_variable_values.
25010 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
25011 (resolve_addr_in_expr): Likewise. Add A argument.
25012 (copy_deref_exprloc): Remove deref argument. Adjust for the
25013 original expression being DW_OP_GNU_variable_value with optionally
25014 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
25015 optionally after it.
25016 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
25017 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
25018 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
25019 (variable_value_hasher::hash, variable_value_hasher::equal): New
25021 (resolve_variable_value_in_expr, resolve_variable_value,
25022 resolve_variable_values, note_variable_value_in_expr,
25023 note_variable_value): New functions.
25024 (dwarf2out_early_finish): Call note_variable_value on all toplevel
25027 2017-02-24 Jakub Jelinek <jakub@redhat.com>
25030 * opts.h (handle_generated_option): Add GENERATED_P argument.
25031 * opts-common.c (handle_option): Adjust function comment.
25032 (handle_generated_option): Add GENERATED_P argument, pass it to
25034 (control_warning_option): Pass false to handle_generated_option
25036 * opts.c (maybe_default_option): Pass true to handle_generated_option
25038 * optc-gen.awk: Likewise.
25040 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25042 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
25043 a REG, look at the REG it is a SUBREG of.
25044 (splitter for cmpeqsi_t): Ditto.
25046 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25048 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
25049 the special USEs with the pattern of the insn, not the insn itself.
25051 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
25054 * doc/invoke.texi: Document -mload-store-pairs.
25056 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25057 Sandra Loosemore <sandra@codesourcery.com>
25059 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
25060 argument isn't a CONST_INT.
25061 (nios2_alternate_compare_const): Assert op is a CONST_INT.
25062 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
25063 (nios2_validate_compare): Bypass alternate compare logic if *op2
25064 is not a CONST_INT.
25065 (ldstwm_operation_p): Return false if first_base is not a REG or
25066 if first_offset is not a CONST_INT.
25068 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25070 * config/cris/cris.md: Use correct operand in a define_peephole2.
25072 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25074 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
25076 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25078 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
25079 this_insn if it is an INSN or JUMP_INSN.
25080 (force_offsettable): Look at base, not at addr.
25081 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
25082 on things that aren't necessarily CONST_INTs.
25084 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
25086 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
25087 -mfpmath=sse is the default also for x86-32 targets with SSE2
25088 instruction set when @option{-ffast-math} is enabled
25090 2017-02-24 Jeff Law <law@redhat.com>
25092 PR rtl-optimizatoin/79286
25093 * ira.c (update_equiv_regs): Drop may_trap_p exception to
25096 2017-02-24 Richard Biener <rguenther@suse.de>
25098 PR tree-optimization/79389
25099 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
25102 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
25104 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
25105 function comment to reflect reality.
25106 (loop_exits_before_overflow): Fix typo in function description.
25108 2017-02-24 Richard Biener <rguenther@suse.de>
25110 PR tree-optimization/79389
25111 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
25112 properly that a threading opportunity exists. Detect conditional
25113 copy/constant propagation opportunities.
25115 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
25117 * config/visium/visium.md (type): Add trap.
25118 (b): New mode attribute.
25119 (*btst): Rename into...
25120 (*btst<mode>): ...this and adjust.
25121 (*cbranchsi4_btst_insn): Rename into...
25122 (*cbranch<mode>4_btst_insn): ...this and adjust.
25123 (trap): New define_insn.
25125 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25127 PR tree-optimization/79389
25128 * ifcvt.c (struct noce_if_info): Add rev_cond field.
25129 (noce_reversed_cond_code): New function.
25130 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
25131 reversed_comparison_code. Formatting fix.
25132 (noce_try_store_flag): Test rev_cond != NULL in addition to
25133 reversed_comparison_code.
25134 (noce_try_store_flag_constants): Likewise.
25135 (noce_try_store_flag_mask): Likewise.
25136 (noce_try_addcc): Use rev_cond if non-NULL instead of
25137 reversed_comparison_code.
25138 (noce_try_cmove_arith): Likewise. Formatting fixes.
25139 (noce_try_minmax, noce_try_abs): Clear rev_cond.
25140 (noce_find_if_block): Initialize rev_cond.
25141 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
25142 instead of false as last argument never attempt to reverse it
25145 2017-02-23 Bin Cheng <bin.cheng@arm.com>
25147 PR tree-optimization/79663
25148 * tree-predcom.c (combine_chains): Process refs in reverse order
25149 only for ZERO length chains, and add explaining comment.
25151 2017-02-23 Jeff Law <law@redhat.com>
25153 PR tree-optimization/79578
25154 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
25155 in call to operand_equal_p.
25157 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
25160 * config/i386/cpuid.h: Fix another undefined behavior.
25162 2017-02-23 Richard Biener <rguenther@suse.de>
25164 PR tree-optimization/79683
25165 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
25166 vector types for data-refs.
25168 2017-02-23 Martin Liska <mliska@suse.cz>
25170 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
25172 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25174 PR middle-end/79665
25175 * internal-fn.c (get_range_pos_neg): Moved to ...
25176 * tree.c (get_range_pos_neg): ... here. No longer static.
25177 * tree.h (get_range_pos_neg): New prototype.
25178 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
25179 are known to be in between 0 and signed maximum inclusive, try to
25180 expand both unsigned and signed divmod and use the cheaper one from
25183 2017-02-22 Jeff Law <law@redhat.com>
25185 PR tree-optimization/79578
25186 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
25187 to compare base operands.
25189 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
25192 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
25193 gpc_reg_operand instead of fpr_reg_operand.
25195 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
25197 * config/mips/mips.c (mips_return_in_memory): Force FP
25198 vector types to be returned in memory for o32 ABI.
25200 2017-02-22 Jakub Jelinek <jakub@redhat.com>
25202 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
25203 instead of DW_TAG_member for static data member declarations and don't
25204 set no_linkage_name for static inline data members.
25205 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
25208 2017-02-22 Martin Liska <mliska@suse.cz>
25210 * doc/invoke.texi: Replace inequality signs with square brackets
25213 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25215 PR tree-optimization/68644
25216 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
25218 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
25221 * lra-constraints.c (simplify_operand_subreg): Handle
25222 WORD_REGISTER_OPERATIONS targets.
25224 2017-02-22 Jakub Jelinek <jakub@redhat.com>
25227 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
25228 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
25229 elimination by swapping fld*.
25231 2017-02-22 Richard Biener <rguenther@suse.de>
25233 PR tree-optimization/79673
25234 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
25235 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
25236 irrelevant address-space qualifiers and avoiding a
25237 ADDR_SPACE_CONVERT_EXPR from fold_convert.
25239 2017-02-22 Richard Biener <rguenther@suse.de>
25241 PR tree-optimization/79666
25242 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
25243 to not symbolically negate if that may introduce undefined
25246 2017-02-22 Martin Liska <mliska@suse.cz>
25249 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
25250 * data-streamer-out.c (streamer_write_gcov_count_stream):
25252 * value-prof.c (stream_out_histogram_value): Make assert more
25253 precise based on type of counter.
25255 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
25258 * config/i386/i386.md (standard_x87sse_constant_load splitter):
25259 Use nonimmediate_operand instead of memory_operand for operand 1.
25260 (float-extend standard_x87sse_constant_load splitter): Ditto.
25262 2017-02-21 Jeff Law <law@redhat.com>
25264 PR tree-optimization/79621
25265 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
25266 blocks with edges to themselves.
25268 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25271 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
25272 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
25273 Use gimple_call_builtin_p.
25276 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
25277 on temporarily removed DEBUG_INSNs.
25279 PR tree-optimization/79649
25280 * tree-loop-distribution.c (classify_partition): Give up on
25281 non-generic address space loads/stores.
25283 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
25285 * doc/loop.texi (Loop manipulation): Remove nonexistent
25286 tree_ssa_loop_version from the documentation.
25287 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
25289 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25292 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
25293 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
25294 * config/rs6000/rs6000.c: Include except.h.
25295 (rs6000_expand_split_stack_prologue): Call
25296 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
25298 2017-02-21 Martin Jambor <mjambor@suse.cz>
25301 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
25302 have been analyzed.
25304 2017-02-21 Martin Jambor <mjambor@suse.cz>
25306 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
25307 for backward compatibility only.
25308 * doc/invoke.texi (Option Summary): Remove all references to
25309 -fipa-cp-alignment.
25311 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
25315 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25317 * lra-constraints.c (curr_insn_transform): Handle
25318 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
25320 2017-02-21 Martin Liska <mliska@suse.cz>
25322 * config/i386/i386.opt: Replace -masm-dialect with -masm.
25324 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
25326 PR translation/79638
25327 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
25329 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
25332 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
25333 (arm_function_ok_for_sibcall): Return false for an indirect call by
25334 descriptor if all the argument registers are used.
25335 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
25336 alignment of the function.
25338 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25340 PR tree-optimization/61441
25341 * simplify-rtx.c (simplify_const_unary_operation): For
25342 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
25343 the sNaN unmodified.
25345 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25347 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25348 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25349 instead of SYSTEM_HEADER_DIR.
25351 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
25352 Martin Liška <mliska@suse.cz>
25354 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
25355 Fix typos and grammar, use active voice, and clarify.
25357 2017-02-20 Marek Polacek <polacek@redhat.com>
25359 PR middle-end/79537
25360 * gimplify.c (gimplify_expr): Handle unused *&&L;.
25363 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
25365 2017-02-20 Jakub Jelinek <jakub@redhat.com>
25368 * config/i386/i386.c (ix86_expand_builtin): Handle
25369 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
25370 ix86_builtins_isa[fcode].isa as a requirement of those
25371 flags and any other flag in the bitmask.
25372 (ix86_init_mmx_sse_builtins): Use 0 instead of
25373 ~OPTION_MASK_ISA_64BIT as mask.
25374 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
25375 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
25376 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
25377 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
25379 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25382 * lra-constraints.c (split_reg): Check requested split mode
25383 is supported by the register.
25385 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25387 * lra-constraints.c (simplify_operand_subreg): Remove early
25390 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25393 * lra-constraints.c (curr_insn_transform): Tighten condition
25394 for converting SUBREG reloads from OP_OUT to OP_INOUT.
25396 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25399 * lra-constraints.c (curr_insn_transform): Handle
25400 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
25402 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
25405 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
25407 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
25409 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
25412 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
25415 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25417 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
25420 2017-02-18 Jakub Jelinek <jakub@redhat.com>
25422 * final.c (last_columnnum, override_columnnum): New variables.
25423 (final_start_function): Set last_columnnum, pass it to begin_prologue
25424 hook and pass 0 to dwarf2out_begin_prologue.
25425 (final_scan_insn): Update override_columnnum. Pass last_columnnum
25426 to source_line debug hook.
25427 (notice_source_line): Compute last_columnnum and for debug_column_info
25428 return true on column changes.
25429 * debug.h (struct gcc_debug_hooks): Add column argument to
25430 source_line and begin_prologue hooks.
25431 (debug_nothing_int_charstar_int_bool): Remove prototype.
25432 (debug_nothing_int_int_charstar,
25433 debug_nothing_int_int_charstar_int_bool): New prototypes.
25434 (dwarf2out_begin_prologue): Add column argument.
25435 * debug.c (do_nothing_debug_hooks): Adjust source_line and
25436 begin_prologue hooks.
25437 (debug_nothing_int_charstar_int_bool): Remove.
25438 (debug_nothing_int_int_charstar,
25439 debug_nothing_int_int_charstar_int_bool): New functions.
25440 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
25441 through to dwarf2out_source_line.
25442 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
25443 (dwarf2out_source_line): Add column argument, emit it if requested.
25444 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
25446 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
25447 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
25448 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
25449 through to dwarf2out_begin_prologue.
25450 (vmsdbgout_source_line): Add column argument, pass it through to
25451 dwarf2out_source_line.
25452 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
25453 dbxout_source_line caller.
25454 (dbxout_source_line): Add column argument.
25456 * common.opt (gno-column-info, gcolumn-info): New options.
25457 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
25458 (check_die): Also test for multiple DW_AT_decl_column attributes.
25459 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
25460 DW_AT_decl_column if requested.
25461 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
25463 (gen_variable_die): Likewise.
25464 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
25465 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
25468 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
25469 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
25470 (ix86_handle_option): Handle OPT_m3dnowa.
25471 * doc/invoke.texi (-m3dnowa): Document.
25472 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
25473 -m3dnowa instead of -m3dnow -march=athlon.
25476 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
25477 instead of gcc_assert for K, r and R code checks. Formatting fixes.
25479 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25482 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
25483 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
25484 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
25485 generator for vsx_xxpermdi_<mode>_be.
25486 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
25487 force big-endian semantics.
25488 (vsx_xxpermdi_<mode>_be): New define_expand with same
25489 implementation as previous version of vsx_xxpermdi_<mode>.
25491 2017-02-17 Jakub Jelinek <jakub@redhat.com>
25493 PR tree-optimization/79327
25494 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
25495 variable, its initialization and use.
25497 2017-02-17 Julia Koval <julia.koval@intel.com>
25499 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
25500 (OPTION_MASK_ISA_PKU_UNSET): New.
25501 (ix86_handle_option): Handle -mrdpid.
25502 * config/i386/cpuid.h (bit_RDPID): New.
25503 * config/i386/driver-i386.c (host_detect_local_cpu):
25504 Detect RDPID feature.
25505 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
25506 * config/i386/i386-c.c (ix86_target_macros_internal):
25508 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
25509 (ix86_valid_target_attribute_inner_p): Add "rdpid".
25510 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
25511 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
25512 * config/i386/i386.md (define_insn "rdpid"): New.
25513 * config/i386/i386.opt Add -mrdpid.
25514 * config/i386/immintrin.h (_rdpid_u32): New.
25516 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
25518 PR rtl-optimization/79541
25519 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
25520 instead of transforming it into USE.
25522 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
25524 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
25525 If HONOR_SNANS (SFmode) force the input to a register.
25526 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
25527 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
25528 an frsp or similar insn.
25530 2017-02-17 Martin Liska <mliska@suse.cz>
25532 PR rtl-optimization/79577
25533 * params.def (selsched-max-sched-times): Increase minimum to 1.
25535 2017-02-17 Martin Liska <mliska@suse.cz>
25537 PR rtl-optimization/79574
25538 * gcse.c (want_to_gcse_p): Prevent integer overflow.
25540 2017-02-17 Martin Liska <mliska@suse.cz>
25542 PR tree-optimization/79529
25543 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
25544 ssa_defined_default_def_p to handle cases which are implicitly
25546 * tree-ssa.c (ssa_defined_default_def_p): New function.
25547 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
25548 which are implicitly defined.
25549 * tree-ssa.h (ssa_defined_default_def_p): Declare.
25551 2017-02-17 Richard Biener <rguenther@suse.de>
25553 PR middle-end/79576
25554 * params.def (max-ssa-name-query-depth): Limit to 10.
25556 2017-02-17 Richard Biener <rguenther@suse.de>
25558 PR tree-optimization/79552
25559 * tree-ssa-structalias.c (visit_loadstore): Properly verify
25562 2017-02-17 Richard Biener <rguenther@suse.de>
25565 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
25567 2017-02-17 Marek Polacek <polacek@redhat.com>
25569 PR middle-end/79536
25570 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
25571 (fold_negate_expr): New wrapper.
25573 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
25575 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
25576 Correct terminology and de-emphasize pre-standard behavior.
25578 2017-02-16 Alan Modra <amodra@gmail.com>
25580 PR rtl-optimization/79286
25581 * ira.c (def_dominates_uses): New function.
25582 (update_equiv_regs): Don't create an equivalence for insns that
25583 may trap where the register def does not dominate the use.
25585 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
25587 PR rtl-optimization/78127
25588 * lra.c (lra): Call lra_eliminate before finish the loop after
25591 2017-02-16 Richard Biener <rguenther@suse.de>
25593 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
25595 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
25596 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
25597 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
25598 (isl_val_int_from_wi): New function.
25599 (extract_affine_gmp): Rename to ...
25600 (extract_affine_wi): ... this, take a widest_int.
25601 (extract_affine_int): Just wrap extract_affine_wi.
25602 (add_param_constraints): Use isl_val_int_from_wi.
25603 (add_loop_constraints): Likewise, and extract_affine_wi.
25605 2017-02-15 Jeff Law <law@redhat.com>
25607 PR middle-end/79521
25608 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
25609 ira_init_register_move_cost_if_necessary.
25611 2017-02-15 Martin Sebor <msebor@redhat.com>
25613 PR middle-end/32003
25614 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
25615 removed in a prior commit.
25617 2017-02-15 Bin Cheng <bin.cheng@arm.com>
25619 PR tree-optimization/79347
25620 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
25621 counters during peeling.
25623 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
25625 * Makefile.in (site.exp): Remove "set ISLVER".
25627 2017-02-15 Jakub Jelinek <jakub@redhat.com>
25630 * real.c (real_from_integer): Call real_convert even for decimal.
25632 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
25635 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
25637 2017-02-14 Andrew Pinski <apinski@cavium.com>
25639 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
25640 cores and change the partno/implementer to be correct.
25641 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
25642 the 'B" as the implementer.
25643 * config/aarch64/aarch64-tune.md: Regenerate.
25645 2017-02-14 Carl Love <cel@us.ibm.com>
25647 * config/rs6000/rs6000.c: Add case statement entry to make the
25648 xvcvuxdsp built-in argument unsigned.
25649 * config/rs6000/vsx.md: Fix the source and return operand types so they
25650 match the instruction definitions from the ISA document. Fix typo
25651 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
25654 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
25657 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
25658 member early_clobber_alts.
25659 * lra-lives.c (reg_early_clobber_p): New.
25660 (process_bb_lives): Use it.
25661 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
25662 (debug_operand_data): Initialize early_clobber_alts.
25663 (setup_operand_alternative): Set up early_clobber_alts.
25664 (collect_non_operand_hard_regs): Ditto. Pass early clobber
25665 alternatives to new_insn_reg.
25666 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
25668 (lra_update_insn_regno_info): Pass the new arg.
25670 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25672 PR middle-end/79505
25673 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
25674 (new_oacc_loop_raw): Don't clear already cleared fields.
25677 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
25678 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
25679 _mm512_prefetch_i64gather_ps): New inline functions and macros.
25681 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
25684 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
25686 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
25689 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
25690 the extra instruction to the right place to store 128-bit constant
25693 2017-02-14 Martin Sebor <msebor@redhat.com>
25695 PR middle-end/79448
25696 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
25697 warning for strings of unknown length.
25699 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
25701 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
25703 2017-02-14 Jeff Law <law@redhat.com>
25706 * ira-costs.c (scan_one_insn): Initialize register move costs
25707 for pseudos seen in USE/CLOBBER insns.
25709 PR tree-optimization/79095
25710 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
25711 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
25712 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
25713 if the operands are known to be not equal, then the resulting range
25715 (intersect_ranges): If the new range is ~[0,0] and the old range is
25716 wide, then prefer ~[0,0].
25717 * tree-vrp.c (overflow_comparison_p_1): New function.
25718 (overflow_comparison_p): New function.
25719 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
25720 if NAME is used in an overflow test.
25721 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
25722 overflow check that can be expressed as an equality test, then adjust
25723 ops to be that equality test.
25725 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25727 * config/s390/s390-builtin-types.def: Remove flags argument.
25728 * config/s390/s390.c (s390_init_builtins): Likewise.
25730 2017-02-14 Martin Liska <mliska@suse.cz>
25732 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
25733 vector. Fix trailing white spaces.
25735 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
25737 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
25740 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25742 PR rtl-optimization/68664
25743 * config/arm/arm.c (arm_sched_can_speculate_insn):
25744 New function. Declare prototype.
25745 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25747 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25749 PR rtl-optimization/68664
25750 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
25752 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
25754 2017-02-14 Amit Pawar <amit.pawar@amd.com>
25756 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
25757 max skip bytes for function, loop and jump.
25759 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25761 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
25762 ABS_EXPR for gimple dump.
25764 2017-02-14 Jakub Jelinek <jakub@redhat.com>
25767 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
25769 PR tree-optimization/79408
25770 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
25771 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
25772 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
25773 also if rhs1 is INTEGER_CST.
25775 2017-02-14 Richard Biener <rguenther@suse.de>
25777 PR middle-end/79432
25778 * tree-into-ssa.c (insert_phi_nodes): When the function can
25779 have abnormal edges rewrite SSA names with broken use-def
25780 dominance out of SSA and register them for PHI insertion.
25782 2017-02-13 Martin Sebor <msebor@redhat.com>
25784 PR middle-end/79496
25785 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
25786 clearing info.nowrite flag when snprintf size argument is a range.
25788 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25790 * cprop.c (cprop_jump): Add missing space in string literal.
25791 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
25792 (get_constraint_for_component_ref): Likewise.
25793 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
25794 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
25795 * lra-constraints.c (process_alt_operands): Likewise.
25796 * ipa-inline.c (inline_small_functions): Likewise.
25797 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
25798 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
25799 * trans-mem.c (diagnose_tm_1_op): Likewise.
25800 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
25801 (grid_parallel_clauses_gridifiable): Likewise.
25803 * config/nvptx/mkoffload.c (process): Add space in between
25806 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
25807 "MOD4_SSE_REGS" and "ALL_REGS".
25809 * spellcheck.c (test_data): Add , in between "foo" and "food".
25811 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25814 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
25815 boundary crossing check and subsequent code generation agree.
25817 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25819 * config/aarch64/aarch64.c (has_memory_op): Delete.
25820 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
25823 2017-02-13 Jakub Jelinek <jakub@redhat.com>
25825 PR rtl-optimization/79388
25826 PR rtl-optimization/79450
25827 * combine.c (distribute_notes): When removing TEM_INSN for which
25828 corresponding dest has last value recorded, invalidate that last
25831 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25833 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
25834 of explicit '@'. Add missing assembly comment marker on branch costs
25837 2017-02-13 Nathan Sidwell <nathan@acm.org>
25839 * gengtype-lex.l (<in_struct>): Add '/'.
25841 2017-02-13 Martin Liska <mliska@suse.cz>
25844 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
25846 2017-02-13 Richard Biener <rguenther@suse.de>
25848 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
25850 * configure: Re-generate.
25851 * config.in: Likewise.
25852 * graphite-dependences.c: Simplify as if
25853 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
25854 * graphite-isl-ast-to-gimple.c: Likewise.
25855 * graphite-optimize-isl.c: Likewise.
25856 * graphite-poly.c: Likewise.
25857 * graphite-sese-to-poly.c: Likewise.
25858 * graphite.h: Likewise.
25859 * toplev.c: Include isl/version.h and use isl_version () for
25860 printing the ISL version.
25861 * doc/install.texi: Update ISL requirement.
25863 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25865 * doc/standards.texi (Standards): Update reference to
25868 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
25870 * doc/extend.texi (Named Address Spaces): sourceware.org now
25872 * doc/install.texi (Binaries): Ditto.
25875 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25877 * doc/cpp.texi: Replace "stringify"/"stringification" with C
25878 standard terminology "stringize"/"stringizing" throughout.
25879 * doc/cppinternals.texi: Likewise.
25881 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
25883 * doc/extend.texi: Fix some spelling mistakes and typos.
25884 * doc/invoke.texi: Likewise.
25886 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25889 * params.def (inline-min-speedup) Change from 10 to 8.
25891 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25893 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
25896 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
25899 * ipa-inline-analysis.c (get_minimal_bb): New function.
25900 (record_modified): Use it.
25901 (remap_edge_change_prob): Handle also ancestor functions.
25903 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
25905 * doc/contrib.texi (Contributors): Remove broken link into
25906 the Mauve CVS repository.
25908 2017-02-11 Jakub Jelinek <jakub@redhat.com>
25910 PR middle-end/79454
25911 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
25912 result computation whenever lhs doesn't have vector mode, not
25913 just when it has BLKmode.
25915 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25917 * doc/makefile.texi (profiledbootstrap): Refer to the
25918 installation instructions only in textual form.
25920 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25923 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
25925 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
25927 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
25928 (Specific): Update mingw-w64 reference.
25930 (Specific): Remove broken link to Renesas RX processor.
25932 2017-02-10 Richard Biener <rguenther@suse.de>
25934 * toplev.c (process_options): Do not mention obsolete graphite
25935 options when printing sorry message about missing graphite support.
25936 Mention -floop-nest-optimize.
25938 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
25940 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
25941 (vtst_p16): Likewise.
25942 (vtstq_p8): Likewise.
25943 (vtstq_p16): Likewise.
25945 (vtstq_p64): Likewise.
25946 * config/arm/arm_neon.h (vgetq_lane_p64): New.
25947 (vset_lane_p64): New.
25948 (vsetq_lane_p64): New.
25950 2017-02-10 Jakub Jelinek <jakub@redhat.com>
25952 PR tree-optimization/79411
25953 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
25954 stmt operands are SSA_NAMEs used in abnormal phis.
25955 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
25958 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25961 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
25964 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
25966 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
25968 2017-02-09 Jakub Jelinek <jakub@redhat.com>
25970 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
25974 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
25975 not arbitrary TREE_CONSTANT.
25978 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
25979 "omp declare target link" attribute unless is_global_var.
25980 * omp-offload.c (find_link_var_op): Likewise.
25982 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
25983 Chung-Lin Tang <cltang@codesourcery.com>
25985 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
25987 (gimplify_adjust_omp_clauses): Don't delete TILE.
25988 (gimplify_omp_for): Deal with TILE.
25989 * internal-fn.c (expand_GOACC_TILE): New function.
25990 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
25992 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
25993 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
25995 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
25996 avoid DIV for outermost collapse var.
25997 (expand_oacc_for): Insert tile element loop as needed. Adjust.
25998 Remove out of date comments, fix whitespace.
25999 * omp-general.c (omp_extract_for_data): Deal with tiling.
26000 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
26001 adjust OLF_DIM_BASE value.
26002 (struct omp_for_data): Add tiling field.
26003 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
26004 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
26005 for auto loops. Remove default auto determining, moved to
26006 oacc_loop_fixed_partitions.
26007 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
26008 stmts, add e_mask field.
26009 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
26010 (oacc_thread_numbers): Use oacc_dim_call.
26011 (oacc_xform_tile): New.
26012 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
26013 (finish_oacc_loop): Adjust for ifns vector.
26014 (oacc_loop_discover_walk): Append loop abstraction sites to list,
26015 add case for GOACC_TILE fns.
26016 (oacc_loop_xform_loop): Delete.
26017 (oacc_loop_process): Iterate over call list directly, and add
26018 handling for GOACC_TILE fns.
26019 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
26021 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
26022 vector partitioning to outer loops. Assign 2 partitions to loops
26023 when available. Add TILE handling.
26024 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
26025 (execite_oacc_device_lower): Process GOACC_TILE fns,
26026 ignore unknown specs.
26027 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
26028 * tree.c (omp_clause_num_ops): Adjust TILE ops.
26029 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
26031 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
26033 * configure.ac (ACX_BUGURL): Update.
26034 * configure: Regenerate.
26036 2017-02-09 Richard Biener <rguenther@suse.de>
26038 PR tree-optimization/69823
26039 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
26040 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
26042 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
26044 * config/arc/arc-c.def: Add __NPS400__ definition.
26045 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
26046 (TARGET_NPS400): Define.
26048 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
26050 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
26052 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
26053 pointer, arch_info.
26054 (arc_cpu_types): Fill the arch_info field with a pointer into the
26055 arc_arch_types table.
26056 (arc_selected_cpu): Declare.
26057 * config/arc/arc.c (arc_selected_cpu): Make global.
26058 (arc_selected_arch): Delete.
26059 (arc_base_cpu): Delete.
26060 (arc_override_options): Remove references to deleted variables,
26061 update access to arch information.
26062 (ARC_OPT): Update access to arch information.
26063 (ARC_OPTX): Likewise.
26064 * config/arc/arc.h (arc_base_cpu): Remove declaration.
26065 (TARGET_ARC600): Update access to arch information.
26066 (TARGET_ARC601): Likewise.
26067 (TARGET_ARC700): Likewise.
26068 (TARGET_EM): Likewise.
26069 (TARGET_HS): Likewise.
26070 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
26073 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
26076 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
26077 condition/operands for integer GE/LE/GEU/LEU operations.
26079 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
26081 PR translation/79397
26082 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
26085 2017-02-08 Martin Jambor <mjambor@suse.cz>
26088 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
26089 whether allocation happened.
26090 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
26091 nothing was allocated.
26093 2017-02-08 Jakub Jelinek <jakub@redhat.com>
26095 PR tree-optimization/79408
26096 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
26097 constant, but SSA_NAME with a known integer range, use the minimum
26098 of that range instead of op1 to determine if modulo can be replaced
26099 with its first operand.
26101 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26103 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
26105 2017-02-08 Richard Biener <rguenther@suse.de>
26107 PR tree-optimization/71824
26108 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26109 Check all loops contained in the merged region.
26111 2017-02-07 Andrew Pinski <apinski@cavium.com>
26113 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
26115 2017-02-07 Andrew Pinski <apinski@cavium.com>
26117 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
26118 (thunderxt88): Likewise.
26119 (thunderxt81): Disable LSE and change v8.1 to v8.
26120 (thunderxt83): Likewise.
26122 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26123 Richard Biener <rguenther@suse.de>
26125 PR middle-end/79399
26126 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
26127 type from int to size_t.
26128 * ira-costs.c (struct_costs_size): Change type from int to size_t.
26130 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26132 PR rtl-optimization/79386
26133 * cprop.c (bypass_conditional_jumps): Initialize
26134 bypass_last_basic_block already before splitting bbs after
26135 unconditional traps...
26136 (bypass_conditional_jumps): ... rather than here.
26139 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
26140 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
26141 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
26142 fix -masm=intel patterns.
26144 2017-02-07 Richard Biener <rguenther@suse.de>
26146 PR tree-optimization/79256
26147 PR middle-end/79278
26148 * builtins.c (get_object_alignment_2): Use min_align_of_type
26149 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
26150 and ADJUST_FIELD_ALIGN.
26152 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
26154 * doc/tm.texi: Regenerate.
26155 * stor-layout.c (layout_decl): Adjust.
26156 (update_alignment_for_field): Likewise.
26157 (place_field): Likewise.
26158 (min_align_of_type): Likewise.
26159 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
26160 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
26161 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
26162 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
26163 * config/frv/frv.c (frv_adjust_field_align): Likewise.
26164 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
26165 * config/i386/i386.c (x86_field_alignment): Likewise.
26166 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
26167 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
26168 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
26169 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
26170 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
26171 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
26175 2017-01-30 Richard Biener <rguenther@suse.de>
26177 PR tree-optimization/79256
26178 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26179 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26182 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
26184 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
26185 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
26186 builtins to SImode and emit a zero-extend, if necessary.
26188 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26190 * docs/invoke.texi (RISC-V Options): Alphabetize.
26192 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26194 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
26197 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26199 * config/riscv/riscv.c: New file.
26200 * common/config/riscv/riscv-common.c: Likewise.
26201 * config.gcc: Likewise.
26202 * config/riscv/constraints.md: Likewise.
26203 * config/riscv/elf.h: Likewise.
26204 * config/riscv/generic.md: Likewise.
26205 * config/riscv/linux.h: Likewise.
26206 * config/riscv/multilib-generator: Likewise.
26207 * config/riscv/peephole.md: Likewise.
26208 * config/riscv/pic.md: Likewise.
26209 * config/riscv/predicates.md: Likewise.
26210 * config/riscv/riscv-builtins.c: Likewise.
26211 * config/riscv/riscv-c.c: Likewise.
26212 * config/riscv/riscv-ftypes.def: Likewise.
26213 * config/riscv/riscv-modes.def: Likewise.
26214 * config/riscv/riscv-opts.h: Likewise.
26215 * config/riscv/riscv-protos.h: Likewise.
26216 * config/riscv/riscv.h: Likewise.
26217 * config/riscv/riscv.md: Likewise.
26218 * config/riscv/riscv.opt: Likewise.
26219 * config/riscv/sync.md: Likewise.
26220 * config/riscv/t-elf-multilib: Likewise.
26221 * config/riscv/t-linux: Likewise.
26222 * config/riscv/t-linux-multilib: Likewise.
26223 * config/riscv/t-riscv: Likewise.
26224 * configure.ac: Likewise.
26225 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
26226 Waterman as RISC-V maintainers.
26227 * doc/install.texi: Add RISC-V entries.
26228 * doc/invoke.texi: Add RISC-V options section.
26229 * doc/md.texi: Add RISC-V constraints section.
26230 * configure: Regenerated.
26232 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
26235 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
26236 false values to be constant vectors with all 0 or all 1 bits set.
26237 (vcondu<mode><mode>): Likewise.
26238 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
26240 (fpmask_comparison_operator): Update comment.
26241 (vecint_comparison_operator): New predicate.
26242 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
26243 vector conditionals when the true and false values are constant
26244 vectors with all 0 bits or all 1 bits set.
26246 2017-02-06 Martin Sebor <msebor@redhat.com>
26248 PR tree-optimization/79376
26249 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
26251 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
26253 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
26254 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
26255 to simplify split condition.
26257 2017-02-06 Jakub Jelinek <jakub@redhat.com>
26259 * omp-expand.c (oxpand_omp_atomic_fetch_op,
26260 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
26263 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
26265 PR rtl-optimization/68664
26266 * target.def (can_speculate_insn): New hook.
26267 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
26268 * doc/tm.texi: Regenerate.
26269 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
26270 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
26271 (rs6000_sched_can_speculate_insn): New function.
26273 2017-02-06 Jakub Jelinek <jakub@redhat.com>
26275 PR tree-optimization/79284
26276 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
26277 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
26278 vectorizable_mask_load_store, vectorizable_operation,
26279 vect_is_simple_cond, get_same_sized_vectype): Use it instead
26280 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
26281 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
26282 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
26283 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
26284 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
26285 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
26286 is_gimple_assign (stmt). Replace another such test with
26287 is_gimple_assign (stmt).
26289 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
26292 * config/avr/avr.c (rtl-iter.h): Include it.
26293 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
26294 (avr_legitimate_combined_insn): ...and implementation.
26296 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
26298 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
26299 * config/s390/s390.c (s390_const_operand_ok)
26300 (s390_canonicalize_comparison, s390_extract_part)
26301 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
26302 (s390_contiguous_bitmask_p, s390_rtx_costs)
26303 (legitimize_pic_address): Likewise.
26304 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
26305 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
26306 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
26307 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
26308 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
26310 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
26312 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
26313 REGNO($0) == REGNO($1).
26315 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26317 * config/s390/linux.h(SIZE_TYPE): Add comment.
26319 2017-02-06 Julian Brown <julian@codesourcery.com>
26320 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26321 Virendra Pathak <virendra.pathak@broadcom.com>
26323 * config/aarch64/aarch64-cores.def: Change the scheduler
26325 * config/aarch64/aarch64.md: Include thunderx2t99.md.
26326 * config/aarch64/thunderx2t99.md: New file.
26328 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26330 * doc/standards.texi (Go Language): Update link to language
26333 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
26335 * tree-eh.c (lower_resx): Sanitize profile.
26336 (cleanup_empty_eh_move_lp): Likewise.
26338 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
26341 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
26343 * cfgloopmanip.h (loop_version): Update prototype.
26344 * modulo-sched.c (sms_schedule): Update call of loop_version.
26345 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
26346 * tree-parloops.c (gen_parallel_loop): Likewise.
26347 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
26348 * tree-ssa-loop-split.c (split_loop): Likewise.
26349 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
26350 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
26352 2017-02-05 Martin Liska <mliska@suse.cz>
26355 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
26357 (print_operand_address): Initialize a struct to zero.
26359 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26361 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
26362 garbage collector only in textual form.
26364 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26366 * doc/extend.texi (x86 specific memory model extensions for
26367 transactional memory): Simplify a phrase.
26369 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
26372 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
26373 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
26374 (atomic_storedi_1): Likewise.
26376 2017-02-04 Jakub Jelinek <jakub@redhat.com>
26378 PR tree-optimization/79338
26379 * tree-parloops.c (gather_scalar_reductions): Don't call
26380 vect_analyze_loop_form for loop->inner before destroying loop's
26383 2017-02-03 Martin Sebor <msebor@redhat.com>
26385 PR tree-optimization/79327
26386 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
26387 when precision has resulted in leading zeros.
26388 (format_integer): Adjust the likely counter to assume an unknown
26389 argument that may be zero is non-zero.
26391 2017-02-03 Jason Merrill <jason@redhat.com>
26394 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
26395 avoid copying non-taken branch.
26397 2017-02-03 Jakub Jelinek <jakub@redhat.com>
26399 PR tree-optimization/79340
26400 * tree-vect-loop.c (vectorizable_reduction): Release
26401 vec_defs elements after safe_splicing them into other vectors.
26404 PR tree-optimization/79327
26405 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
26406 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
26408 (format_integer): Use wide_int_to_tree instead of build_int_cst
26409 + to_?hwi. If argmin is NULL, just set argmin and argmax to
26410 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
26411 of shortest and longest sequence.
26413 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
26415 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
26416 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
26418 2017-02-03 Walter Lee <walt@tilera.com>
26421 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
26422 after initial stackframe link reg save.
26423 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
26425 2017-02-03 Jakub Jelinek <jakub@redhat.com>
26428 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
26429 wu for stxssp alternative.
26431 2017-02-03 Martin Sebor <msebor@redhat.com>
26433 PR tree-optimization/79352
26434 * gimple-fold.c (get_range_strlen): Add argument.
26435 (get_range_strlen): Change return type to bool.
26436 (get_maxval_strlen): Pass in a dummy argument.
26437 * gimple-fold.h (get_range_strlen): Change return type to bool.
26438 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
26439 * tree.h (array_at_struct_end_p): Add argument.
26440 * tree.c (array_at_struct_end_p): Handle it.
26442 2017-02-03 Martin Liska <mliska@suse.cz>
26445 * multiple_target.c (create_dispatcher_calls): Redirect edge
26446 from a caller of a dispatcher.
26447 (expand_target_clones): Make the clones local.
26448 (ipa_target_clone): Do both target clones and resolvers.
26449 (ipa_dispatcher_calls): Remove the pass.
26450 (pass_dispatcher_calls::gate): Likewise.
26451 (make_pass_dispatcher_calls): Likewise.
26452 * passes.def (pass_target_clone): Put as very first IPA early
26455 2017-02-03 Martin Liska <mliska@suse.cz>
26457 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
26458 in case of a function with ifunc attribute.
26460 2017-02-03 Martin Liska <mliska@suse.cz>
26462 * cgraph.c (cgraph_node::dump): Dump function version info.
26463 * symtab.c (symtab_node::dump_base): Add missing new line.
26465 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26467 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
26468 (ifcombine_ifandif): Use it.
26470 2017-02-03 Martin Liska <mliska@suse.cz>
26472 * doc/invoke.texi: Document default value for
26473 use-after-scope-direct-emission-threshold.
26475 2017-02-03 Martin Liska <mliska@suse.cz>
26477 PR tree-optimization/79339
26478 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
26479 (format_floating): Likewise.
26481 2017-02-03 Martin Liska <mliska@suse.cz>
26484 * ipa-prop.c (ipa_node_params_t::insert): Remove current
26486 (ipa_node_params_t::remove): Likewise.
26487 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
26488 initialization from removed ipa_node_params_t::insert.
26489 (ipa_node_params::~ipa_node_params): Move from removed
26490 ipa_node_params_t::release.
26491 * symbol-summary.h (symbol_summary::m_released): New member.
26492 Do not release a summary twice. Do not allow to call finalizer
26493 for types of a summary that live in GGC memory.
26495 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26497 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
26500 2017-02-02 Martin Sebor <msebor@redhat.com>
26502 PR middle-end/79275
26503 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
26504 (format_string): Tighten up the range of output for non-constant
26505 strings and correct the expected range for wide non-constant strings.
26507 2017-02-02 Martin Sebor <msebor@redhat.com>
26509 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
26511 PR middle-end/32003
26512 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
26514 (-fdump-tree-@var): Add to index and document how to come up
26515 with pass-specific option and dump file names.
26516 (-fdump-passes): Clarify where to look for output.
26518 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26520 PR middle-end/77445
26521 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
26522 statistics of the analyzed path; allow threading for speed when
26523 any of BBs along the path are optimized for speed.
26525 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
26527 PR middle-end/78468
26528 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
26529 settings of the virtual registers.
26532 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
26534 * explow.c (get_dynamic_stack_size): Take known alignment of stack
26535 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
26538 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26540 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
26541 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
26543 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26545 * config/s390/s390.md: Add missing comments with the expanded
26547 * config/s390/vector.md: Likewise.
26548 * config/s390/vx-builtins.md: Likewise.
26550 2017-02-02 Jakub Jelinek <jakub@redhat.com>
26553 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
26554 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
26555 conditions on a single line.
26557 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26559 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
26560 __S390_VX__ to __VX__.
26562 2017-02-01 Andrew Pinski <apinski@cavium.com>
26564 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
26565 stmt_info to record_stmt_cost.
26566 (vect_get_known_peeling_cost): Pass stmt_info if known to
26568 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
26569 cpu_vector_cost field into
26570 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
26571 field into vec_int_stmt_cost and vec_fp_stmt_cost.
26572 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
26573 splitting of scalar_stmt_cost and vec_stmt_cost.
26574 (thunderx_vector_cost): Likewise.
26575 (cortexa57_vector_cost): LIkewise.
26576 (exynosm1_vector_cost): Likewise.
26577 (xgene1_vector_cost): Likewise.
26578 (thunderx2t99_vector_cost): Improve after the splitting of the two
26580 (aarch64_builtin_vectorization_cost): Update for the splitting of
26581 scalar_stmt_cost and vec_stmt_cost.
26583 2017-02-01 Torvald Riegel <triegel@redhat.com>
26584 Richard Henderson <rth@redhat.com>
26586 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
26587 conditional on existance of a fast atomic load.
26588 * optabs-query.c (can_atomic_load_p): New function.
26589 * optabs-query.h (can_atomic_load_p): Declare it.
26590 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
26591 no fast atomic load is available for the particular size of access.
26592 (expand_atomic_compare_and_swap): Likewise.
26593 (expand_atomic_load): Likewise.
26594 (expand_atomic_store): Likewise.
26595 (expand_atomic_fetch_op): Likewise.
26596 * testsuite/lib/target-supports.exp
26597 (check_effective_target_sync_int_128): Remove x86 because it provides
26598 no fast atomic load.
26599 (check_effective_target_sync_int_128_runtime): Likewise.
26601 2017-02-01 Richard Biener <rguenther@suse.de>
26603 * graphite.c: Include tree-vectorizer.h for find_loop_location.
26604 (graphite_transform_loops): Provide opt-info for optimized nests.
26605 * tree-parloop.c (parallelize_loops): Provide opt-info for
26606 parallelized loops.
26608 2017-02-01 Richard Biener <rguenther@suse.de>
26610 PR middle-end/79315
26611 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
26612 was not set before.
26614 2017-02-01 Richard Biener <rguenther@suse.de>
26616 PR tree-optimization/71824
26617 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26618 Verify the loops are valid in the merged SESE region.
26619 (scop_detection::can_represent_loop_1): Check analyzing the
26620 evolution of the number of iterations in the region succeeds.
26622 2017-01-31 Ian Lance Taylor <iant@golang.org>
26624 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
26625 REG_ARGS_SIZE note to 32-bit push insns and call insn.
26627 2017-01-31 David Malcolm <dmalcolm@redhat.com>
26629 PR preprocessor/79210
26630 * input.c (get_substring_ranges_for_loc): Replace line_width
26631 assertion with error-handling.
26633 2017-01-31 Richard Biener <rguenther@suse.de>
26635 PR tree-optimization/77318
26636 * graphite-sese-to-poly.c (extract_affine): Fix assert.
26637 (create_pw_aff_from_tree): Take loop parameter.
26638 (add_condition_to_pbb): Pass loop of the condition to
26639 create_pw_aff_from_tree.
26641 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26643 * config/s390/s390.c (s390_asan_shadow_offset): New function.
26644 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
26646 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
26650 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
26652 (convert_int_to_float128): Likewise.
26653 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
26654 (convert_int_to_float128): Likewise.
26655 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
26656 (UNSPEC_IEEE128_CONVERT): Likewise.
26657 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
26658 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
26659 Use local variables for IBM extended format.
26660 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
26661 (fix_trunc<mode>si2_fprs): Likewise.
26662 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
26663 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
26664 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
26665 to know that we can now have integers of all sizes in vector
26667 (fix<uns>_<mode>di2_hw): Likewise.
26668 (float<uns>_<mode>si2_hw): Likewise.
26669 (fix_<mode>si2_hw): Likewise.
26670 (fixuns_<mode>si2_hw): Likewise.
26671 (float<uns>_<mode>di2_hw): Likewise.
26672 (float_<mode>di2_hw): Likewise.
26673 (float_<mode>si2_hw): Likewise.
26674 (floatuns_<mode>di2_hw): Likewise.
26675 (floatuns_<mode>si2_hw): Likewise.
26676 (xscvqp<su>wz_<mode>): Delete, no longer used.
26677 (xscvqp<su>dz_<mode>): Likewise.
26678 (xscv<su>dqp_<mode>): Likewise.
26679 (ieee128_mfvsrd_64bit): Likewise.
26680 (ieee128_mfvsrd_32bit): Likewise.
26681 (ieee128_mfvsrwz): Likewise.
26682 (ieee128_mtvsrw): Likewise.
26683 (ieee128_mtvsrd_64bit): Likewise.
26684 (ieee128_mtvsrd_32bit): Likewise.
26686 2017-01-31 Martin Liska <mliska@suse.cz>
26689 * ipa-prop.c (ipa_free_all_node_params): Call release method
26690 instead of ~sumbol_summary to not to trigger double times
26693 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
26695 PR tree-optimization/71691
26696 * bitmap.h (class auto_bitmap): New.
26697 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
26698 is_maybe_undefined instead of ssa_undefined_value_p.
26700 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26702 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
26703 __S390_ARCH_LEVEL__ to __ARCH__.
26705 2017-01-31 Jakub Jelinek <jakub@redhat.com>
26707 PR tree-optimization/79267
26708 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
26709 if should_remove_lhs_p is true.
26711 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
26714 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
26715 (add_alignment_attribute): New.
26716 (base_type_die): Add alignment attribute.
26717 (subrange_type_die): Likewise.
26718 (modified_type_die): Likewise.
26719 (gen_array_type_die): Likewise.
26720 (gen_descr_array_type_die: Likewise.
26721 (gen_enumeration_type_die): Likewise.
26722 (gen_subprogram_die): Likewise.
26723 (gen_variable_die): Likewise.
26724 (gen_field_die): Likewise.
26725 (gen_ptr_to_mbr_type_die): Likewise.
26726 (gen_struct_or_union_type_die): Likewise.
26727 (gen_subroutine_type_die): Likewise.
26728 (gen_typedef_die): Likewise.
26729 (base_type_cmp): Compare alignment attribute.
26731 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26734 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
26735 (setb_unsigned) New pattern for setb with CCUNS.
26736 * config/rs6000/rs6000.c (expand_block_compare): Use a different
26737 subfc./subfe sequence to avoid overflow problems. Generate a
26738 shorter sequence with cmpld/setb for power9.
26739 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
26740 for generating subfc. instruction.
26741 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
26742 now uses this instruction.
26744 2017-01-30 Ian Lance Taylor <iant@google.com>
26747 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
26748 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
26750 2017-01-30 Martin Sebor <msebor@redhat.com>
26752 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
26753 Move constant to the right of a relational operator.
26754 (get_mpfr_format_length, format_character, format_string): Ditto.
26755 (should_warn_p, maybe_warn): Same.
26757 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
26759 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
26762 * asan.c (get_translation_unit_decl): Remove function.
26763 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
26765 2017-01-30 Martin Liska <mliska@suse.cz>
26767 PR gcov-profile/79259
26768 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
26769 -fprofile-generate.
26771 2017-01-30 Martin Liska <mliska@suse.cz>
26774 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
26775 Initialize variables with NULL value.
26777 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
26780 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
26782 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
26784 2017-01-30 Richard Biener <rguenther@suse.de>
26786 PR tree-optimization/79276
26787 * tree-vrp.c (process_assert_insertions): Properly adjust common
26788 when removing a duplicate.
26790 * gcc.dg/torture/pr79276.c: New testcase.
26792 2017-01-30 Richard Biener <rguenther@suse.de>
26794 PR tree-optimization/79256
26795 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26796 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26798 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
26800 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
26803 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
26804 ("*r<noxa>sbg_<mode>_sll_bitmask")
26805 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
26806 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
26807 Use contiguous_bitmask_nowrap_operand.
26809 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26812 * config/rs6000/altivec.h (vec_xl): Revise #define.
26813 (vec_xst): Likewise.
26815 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
26817 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
26819 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
26821 PR rtl-optimization/79194
26822 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
26823 traps before call to bypass_conditional_jumps.
26825 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26827 PR tree-optimization/71374
26828 * lra-constraints.c (check_conflict_input_operands): New.
26829 (match_reload): Use it.
26831 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
26834 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
26835 account to calculate conflict_set.
26837 2017-01-27 Bin Cheng <bin.cheng@arm.com>
26839 PR rtl-optimization/78559
26840 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
26841 other_insn in combine.
26843 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
26845 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
26846 uint16_type_node for BT_UINT16.
26848 2017-01-27 David Malcolm <dmalcolm@redhat.com>
26850 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
26851 "RTL Tests" to menu.
26852 (GIMPLE Tests): New node.
26853 (RTL Tests): New node.
26855 2017-01-27 Richard Biener <rguenther@suse.de>
26857 PR tree-optimization/79245
26858 * tree-loop-distribution.c (distribute_loop): Apply cost
26859 modeling also to detected patterns.
26861 2017-01-27 Richard Biener <rguenther@suse.de>
26863 PR tree-optimization/71433
26864 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
26865 (compare_assert_loc): New function.
26866 (process_assert_insertions): Sort and optimize assert locations
26867 to remove duplicates and push down identical assertions on
26868 edges to their destination block.
26870 2017-01-27 Richard Biener <rguenther@suse.de>
26872 PR tree-optimization/79244
26873 * tree-vrp.c (remove_range_assertions): Forcefully propagate
26874 out SSA names even if abnormal.
26876 2017-01-27 Jakub Jelinek <jakub@redhat.com>
26878 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
26879 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
26880 instead of MPFR_RNDN.
26882 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
26885 * arm.c (arm_option_override): Don't call build_target_option_node
26886 until after doing all option overrides.
26887 (arm_valid_target_attribute_tree): Likewise.
26889 2017-01-27 Martin Liska <mliska@suse.cz>
26891 * doc/invoke.texi (-fprofile-arcs): Document profiling support
26892 for {cd}tors and C++ {cd}tors.
26894 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
26896 * config/s390/s390.md ("*setmem_long_and")
26897 ("*setmem_long_and_31z"): Use zero_extend instead of and.
26899 2017-01-26 Martin Sebor <msebor@redhat.com>
26901 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
26904 2017-01-26 Martin Sebor <msebor@redhat.com>
26906 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
26907 HAVE_DFmode before using XFmode or DFmode.
26908 (parse_directive): Avoid using the z length modifier to avoid
26909 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
26911 PR middle-end/78703
26912 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
26913 to accept adjustment as an array.
26914 (get_int_range): New function.
26915 (struct directive): Make width and prec arrays.
26916 (directive::set_width, directive::set_precision): Call get_int_range.
26917 (format_integer, format_floating): Handle width and precision ranges.
26918 (format_string, parse_directive): Same.
26920 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26923 * dwarf2out.c (generate_skeleton_bottom_up): For children with
26924 comdat_type_p set, just clone them, but keep the children in the
26928 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
26929 which have direct callers with -fvar-tracking-assignments enabled
26931 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
26932 inside of type units.
26934 2017-01-26 Martin Sebor <msebor@redhat.com>
26936 PR middle-end/78703
26937 * gimple-ssa-sprintf.c (struct result_range): Add likely and
26939 (struct format_result): Replace number_chars, number_chars_min,
26940 and number_chars_max with a single member of struct result_range.
26942 (format_result::operator+=): Adjust.
26943 (struct fmtresult): Remove bounded. Handle likely and unlikely
26945 (fmtresult::adjust_for_width_or_precision): New function.
26946 (fmtresult:type_max_digits): New function.
26947 (bytes_remaining): Handle likely and unlikely counters.
26948 (min_bytes_remaining): Remove.
26949 (format_percent): Simplify.
26950 (format_integer, format_floating): Set likely and unlikely counters.
26951 (get_string_length, format_character, format_string): Same.
26952 (format_plain, should_warn_p): New function.
26953 (maybe_warn): Call should_warn_p. Update diagnostic messages
26954 and handle those for all directives, including plain strings.
26955 (format_directive): Handle likely and unlikely counters.
26956 Remove unnecessary quoting from diagnostics. Add an informational
26958 (add_bytes): Remove.
26959 (pass_sprintf_length::compute_format_length): Simplify.
26960 (try_substitute_return_value): Handle likely and unlikely counters.
26962 2017-01-26 Carl Love <cel@us.ibm.com>
26964 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
26965 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
26967 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
26970 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
26971 endianess for subregs into account.
26972 * lra-constraints.c (lra_constraints): Do risky transformations
26973 always on the first iteration.
26974 * lra-lives.c (check_pseudos_live_through_calls): Add arg
26975 last_call_used_reg_set.
26976 (process_bb_lives): Define and use last_call_used_reg_set.
26977 * lra.c (lra): Always continue after lra_constraints on the first
26980 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
26982 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
26984 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
26986 2017-01-26 Jakub Jelinek <jakub@redhat.com>
26988 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
26989 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
26990 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
26991 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
26992 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
26993 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
26994 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
26995 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
26996 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
26998 2017-01-26 Marek Polacek <polacek@redhat.com>
27001 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
27002 for the third operand.
27004 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27006 PR middle-end/79236
27007 * omp-low.c (struct omp_context): Add simt_stmt field.
27008 (scan_omp_for): Return omp_context *.
27009 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
27010 context to the _simt_ SIMD stmt.
27011 (lower_omp_for): For combined SIMD with sibling _simt_
27012 SIMD, make sure to use the same decls in _looptemp_
27013 clauses as in the sibling.
27015 2017-01-26 David Sherwood <david.sherwood@arm.com>
27017 PR middle-end/79212
27018 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
27021 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27024 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
27025 emit fld b; fld a; if possible.
27027 * brig-builtins.def: Update copyright years.
27028 * config/arm/arm_acle_builtins.def: Update copyright years.
27030 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
27033 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
27034 constraint instead of o for the stxsd instruction.
27036 2017-01-25 Carl Love <cel@us.ibm.com>
27038 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
27039 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
27041 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
27043 * doc/invoke.texi (C++ Dialect Options): Fix typo.
27045 2017-01-25 Richard Biener <rguenther@suse.de>
27047 PR tree-optimization/69264
27048 * target.def (vector_alignment_reachable): Improve documentation.
27049 * doc/tm.texi: Regenerate.
27050 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
27052 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
27053 earlier changes with respect to TYPE_USER_ALIGN.
27054 (vector_alignment_reachable_p): Likewise. Improve dumping.
27056 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27059 * config/arm/arm.md (xordi3): Force constant operand into a register
27062 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27064 * doc/invoke.texi (-fstore-merging): Correct default optimization
27065 levels at which it is enabled.
27066 (-O): Move -fstore-merging from list to...
27069 2017-01-25 Richard Biener <rguenther@suse.de>
27072 * omp-expand.c: Include debug.h.
27073 (expand_omp_taskreg): Make sure to generate early debug before
27074 outlining anything from a function.
27075 (expand_omp_target): Likewise.
27076 (grid_expand_target_grid_body): Likewise.
27078 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
27081 * asan.c (get_translation_unit_decl): New function.
27082 (asan_add_global): Extract modules file name from globals
27083 TRANSLATION_UNIT_DECL name.
27085 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
27088 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
27089 for long calls with APCS frame and VFP.
27091 2017-01-24 David Malcolm <dmalcolm@redhat.com>
27093 * cfg.c (original_copy_tables_initialized_p): New function.
27094 * cfg.h (original_copy_tables_initialized_p): New decl.
27095 * cfgrtl.c (relink_block_chain): Guard the call to
27096 free_original_copy_tables with a call to
27097 original_copy_tables_initialized_p.
27098 * cgraph.h (symtab_node::native_rtl_p): New decl.
27099 * cgraphunit.c (symtab_node::native_rtl_p): New function.
27100 (symtab_node::needed_p): Don't assert for early assembly output
27101 for __RTL functions.
27102 (cgraph_node::finalize_function): Set "force_output" for __RTL
27104 (cgraph_node::analyze): Bail out early for __RTL functions.
27105 (analyze_functions): Update assertion to support __RTL functions.
27106 (cgraph_node::expand): Bail out early for __RTL functions.
27107 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
27109 * function.h (struct function): Update comment for field
27111 * gimple-expr.c: Include "tree-pass.h".
27112 (gimple_has_body_p): Return false for __RTL functions.
27113 * Makefile.in (OBJS): Add run-rtl-passes.o.
27114 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
27116 (gcc::pass_manager::get_clean_slate): New accessor.
27117 * passes.c: Include "insn-addr.h".
27118 (should_skip_pass_p): Add logging. Update logic for running
27119 "expand" to be compatible with both __GIMPLE and __RTL. Guard
27120 property-provider override so it is only done for gimple passes.
27122 (skip_pass): New function.
27123 (execute_one_pass): Call skip_pass when skipping passes.
27124 * read-md.c (md_reader::read_char): Support filtering
27125 the input to a subset of line numbers.
27126 (md_reader::md_reader): Initialize fields
27127 m_first_line and m_last_line.
27128 (md_reader::read_file_fragment): New function.
27129 * read-md.h (md_reader::read_file_fragment): New decl.
27130 (md_reader::m_first_line): New field.
27131 (md_reader::m_last_line): New field.
27132 * read-rtl-function.c (function_reader::create_function): Only
27133 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
27134 curr_properties. Set DECL_INITIAL to a dummy block.
27135 (read_rtl_function_body_from_file_range): New function.
27136 * read-rtl-function.h (read_rtl_function_body_from_file_range):
27138 * run-rtl-passes.c: New file.
27139 * run-rtl-passes.h: New file.
27141 2017-01-24 Jeff Law <law@redhat.com>
27143 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
27146 2017-01-24 Bin Cheng <bin.cheng@arm.com>
27148 PR tree-optimization/79159
27149 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
27150 (record_nonwrapping_iv): Improve boundary using above function if no
27151 value range information.
27153 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
27154 Martin Jambor <mjambor@suse.cz>
27156 * brig-builtins.def: New file.
27157 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
27158 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
27159 (DEF_HSAIL_SAT_BUILTIN): Likewise.
27160 (DEF_HSAIL_INTR_BUILTIN): Likewise.
27161 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
27162 * builtin-types.def (BT_INT8): New.
27163 (BT_INT16): Likewise.
27164 (BT_UINT8): Likewise.
27165 (BT_UINT16): Likewise.
27166 (BT_FN_ULONG): Likewise.
27167 (BT_FN_UINT_INT): Likewise.
27168 (BT_FN_UINT_ULONG): Likewise.
27169 (BT_FN_UINT_LONG): Likewise.
27170 (BT_FN_UINT_PTR): Likewise.
27171 (BT_FN_ULONG_PTR): Likewise.
27172 (BT_FN_INT8_FLOAT): Likewise.
27173 (BT_FN_INT16_FLOAT): Likewise.
27174 (BT_FN_UINT32_FLOAT): Likewise.
27175 (BT_FN_UINT16_FLOAT): Likewise.
27176 (BT_FN_UINT8_FLOAT): Likewise.
27177 (BT_FN_UINT64_FLOAT): Likewise.
27178 (BT_FN_UINT16_UINT32): Likewise.
27179 (BT_FN_UINT32_UINT16): Likewise.
27180 (BT_FN_UINT16_UINT16_UINT16): Likewise.
27181 (BT_FN_INT_PTR_INT): Likewise.
27182 (BT_FN_UINT_PTR_UINT): Likewise.
27183 (BT_FN_LONG_PTR_LONG): Likewise.
27184 (BT_FN_ULONG_PTR_ULONG): Likewise.
27185 (BT_FN_VOID_UINT64_UINT64): Likewise.
27186 (BT_FN_UINT8_UINT8_UINT8): Likewise.
27187 (BT_FN_INT8_INT8_INT8): Likewise.
27188 (BT_FN_INT16_INT16_INT16): Likewise.
27189 (BT_FN_INT_INT_INT): Likewise.
27190 (BT_FN_UINT_FLOAT_UINT): Likewise.
27191 (BT_FN_FLOAT_UINT_UINT): Likewise.
27192 (BT_FN_ULONG_UINT_UINT): Likewise.
27193 (BT_FN_ULONG_UINT_PTR): Likewise.
27194 (BT_FN_ULONG_ULONG_ULONG): Likewise.
27195 (BT_FN_UINT_UINT_UINT): Likewise.
27196 (BT_FN_VOID_UINT_PTR): Likewise.
27197 (BT_FN_UINT_UINT_PTR: Likewise.
27198 (BT_FN_UINT32_UINT64_PTR): Likewise.
27199 (BT_FN_INT_INT_UINT_UINT): Likewise.
27200 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
27201 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
27202 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
27203 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
27204 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
27205 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
27206 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
27207 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
27208 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
27209 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
27210 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
27211 * doc/frontends.texi: List BRIG FE.
27212 * doc/install.texi (Testing): Add BRIG tesring requirements.
27213 * doc/invoke.texi (Overall Options): Mention BRIG.
27214 * doc/standards.texi (Standards): Doucment BRIG HSA version.
27216 2017-01-24 Richard Biener <rguenther@suse.de>
27218 PR translation/79208
27219 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
27221 2017-01-24 Martin Jambor <mjambor@suse.cz>
27224 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
27225 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
27226 and known_contexts.
27228 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
27230 PR middle-end/79123
27231 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
27232 casts from signed to unsigned really don't have a range.
27234 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
27236 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
27237 GMP_RNDx for compatiblity.
27239 2017-01-24 Martin Liska <mliska@suse.cz>
27242 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
27243 that would prevent us to call alloca with -1 as argument.
27245 2017-01-24 Jakub Jelinek <jakub@redhat.com>
27247 * dwarf2out.c (output_compilation_unit_header, output_file_names):
27248 Avoid -Wformat-security warning.
27250 2017-01-23 Andrew Pinski <apinski@cavium.com>
27252 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
27255 2017-01-23 Martin Sebor <msebor@redhat.com>
27257 PR middle-end/78703
27258 * gimple-ssa-sprintf.c (warn_level): New global.
27259 (format_integer): Use it here and throughout the rest of the file.
27260 Use the same switch to compute sign as base.
27261 (maybe_warn): New function.
27262 (format_directive): Factor out warnings into maybe_warn.
27263 Add debugging output. Use warn_level.
27264 (add_bytes): Use warn_level.
27265 (pass_sprintf_length::compute_format_length): Add debugging output.
27266 (try_substitute_return_value): Same.
27267 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
27269 PR middle-end/78703
27270 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
27271 (struct fmtresult, format_integer, format_floating): Adjust.
27272 (fmtresult::fmtresult): Set max correctly in two argument ctor.
27273 (get_string_length, format_string,format_directive): Same.
27274 (pass_sprintf_length::compute_format_length): Same.
27275 (try_substitute_return_value): Simplify slightly.
27277 PR middle-end/78703
27278 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
27279 (fmtresult::operator+=): Outlined.
27280 (struct fmtresult): Add ctors.
27281 (struct conversion_spec): Rename...
27282 (struct directive): ...to this. Add and remove data members.
27283 (directive::set_width, directive::set_precision): New functions.
27284 (format_percent): Use fmtresult ctor.
27285 (get_width_and_precision): Remove.
27286 (format_integer): Make naming changes. Avoid computing width and
27288 (format_floating): Same. Adjust indentation.
27289 (format_character, format_none): New functions.
27290 (format_string): Moved character handling to format_character.
27291 (format_directive): Remove arguments, change return type.
27292 (parse_directive): New function.
27293 (pass_sprintf_length::compute_format_length): Move directive
27294 parsing to parse_directive.
27296 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27298 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
27299 (assign_assembler_name_if_needed): ... this.
27300 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
27301 (assign_assembler_name_if_needed): ... this.
27302 (free_lang_data_in_cgraph): Adjust callers.
27303 * cgraphunit.c (cgraph_node::analyze): Likewise.
27304 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
27307 2017-01-23 Richard Biener <rguenther@suse.de>
27309 PR tree-optimization/79088
27310 PR tree-optimization/79188
27311 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
27312 resetting loop bounds after last path deletion. Reset loop
27313 bounds of the target loop, make code match the comments.
27314 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
27315 Make sure loops need no fixups.
27317 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27319 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
27320 exponent support with double type for first argument.
27321 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
27322 type returned by __builtin_vec_extract_sig,
27323 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
27324 functions from "vector int" to "vector unsigned int" or from
27325 "vector long long int" to "vector unsigned long long int".
27326 Changed type returned by __builtin_vec_extract_exp,
27327 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
27328 functions from "vector int" to "vector unsigned int" or from
27329 "vector long long int" to "vector unsigned long long int".
27330 Changed return type of __builtin_vec_test_data_class,
27331 __builtin_vec_test_data_class_sp, and
27332 __builtin_vec_test_data_class_dp from "vector int" to
27333 "vector bool int" or from "vector long long int" to "vector bool
27334 long long int" and changed second argument type from "unsigned
27335 int" to "int". Added new overloaded function forms "vector float
27336 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
27337 "vector float __builtin_vec_insert_exp_sp (vector float, vector
27338 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
27339 double, vector unsigned long long int)" and "vector double
27340 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
27341 long int)". Changed return type of
27342 __builtin_scalar_test_data_class and
27343 __builtin_scalar_test_data_class_sp and
27344 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
27345 int" and changed second argument from "unsigned int" to "int".
27346 Changed type returned by __builtin_scalar_test_neg,
27347 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
27348 from "int" to "bool int". Added new overloaded function form
27349 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
27350 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
27351 exponent double-precision with floating point first argument.
27352 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
27353 documentation of scalar_test_data_class, scalar_test_neg,
27354 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
27355 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
27356 vec_test_data_class built-in functions to reflect refinements in
27357 their type signatures.
27359 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
27361 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
27363 (aarch64_elf_asm_destructor): Likewise.
27365 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
27367 PR rtl-optimization/78634
27368 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
27369 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
27370 * ifcvt.c (noce_try_cmove): Add missing cost check.
27372 PR rtl-optimization/71724
27373 * combine.c (if_then_else_cond): Look for situations where it is
27374 beneficial to undo the work of one of the recursive calls.
27376 2017-01-23 Bin Cheng <bin.cheng@arm.com>
27378 PR tree-optimization/70754
27379 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
27380 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
27381 combined stmt before it if not NULL.
27382 (combine_chains): Process refs reversely and compute dominance point
27385 2017-01-23 Martin Liska <mliska@suse.cz>
27387 PR tree-optimization/79196
27388 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
27389 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
27391 (strlen_optimize_stmt): Call the renamed function.
27393 2017-01-23 Michael Matz <matz@suse.de>
27395 PR tree-optimization/78384
27396 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
27398 2017-01-23 Richard Biener <rguenther@suse.de>
27400 PR tree-optimization/79186
27401 * tree-vrp.c (register_new_assert_for): Make sure we've seen
27402 both incoming edges before moving an assert.
27404 2017-01-23 Martin Jambor <mjambor@suse.cz>
27406 * ipa-prop.c (load_from_param_1): Removed.
27407 (load_from_unmodified_param): Bits from load_from_param_1 put back
27409 (load_from_param): Removed.
27410 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
27411 with stmt. Reverted back to use of load_from_unmodified_param.
27413 2017-01-23 Martin Jambor <mjambor@suse.cz>
27416 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
27417 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
27418 field a pointer to garbage collected vector, mark lattices and
27419 ipcp_orig_node with GTY((skip)).
27420 (ipa_get_param_count): Adjust to descriptors being a pointer.
27421 (ipa_get_param): Likewise.
27422 (ipa_get_type): Likewise.
27423 (ipa_get_param_move_cost): Likewise.
27424 (ipa_set_param_used): Likewise.
27425 (ipa_get_controlled_uses): Likewise.
27426 (ipa_set_controlled_uses): Likewise.
27427 (ipa_is_param_used): Likewise.
27428 (ipa_node_params_t): Move into garbage collector. New methods insert
27430 (ipa_node_params_sum): Annotate wth GTY(()).
27431 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
27433 (ipa_load_from_parm_agg): Adjust declaration.
27434 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
27435 * ipa-profile.c (ipa_profile): Likewise.
27436 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
27437 (ipa_populate_param_decls): Make descriptors parameter garbage
27439 (ipa_dump_param): Adjust to descriptors being a pointer.
27440 (ipa_alloc_node_params): Likewise.
27441 (ipa_initialize_node_params): Likewise.
27442 (load_from_param_1): Make descriptors parameter garbage collected.
27443 (load_from_unmodified_param): Likewise.
27444 (load_from_param): Likewise.
27445 (ipa_load_from_parm_agg): Likewise.
27446 (ipa_node_params::~ipa_node_params): Removed.
27447 (ipa_free_all_node_params): Remove call to delete operator.
27448 (ipa_node_params_t::insert): New.
27449 (ipa_node_params_t::remove): Likewise.
27450 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
27451 copy known_csts and known_contexts vectors.
27452 (ipa_read_node_info): Adjust to descriptors being a pointer.
27453 (ipcp_modif_dom_walker): Make m_descriptors field garbage
27455 (ipcp_transform_function): Make descriptors variable garbage
27458 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
27460 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
27461 * config/i386/avx512dqintrin.h: Ditto.
27462 * config/i386/avx512fintrin.h: Ditto.
27463 * config/i386/i386.c: Handle new builtins.
27464 * config/i386/i386-builtin.def: Add new builtins.
27465 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
27466 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
27468 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27469 Martin Liska <mliska@suse.cz>
27471 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
27472 * asan.c (asan_expand_poison_ifn): Support stores and use
27473 appropriate ASAN report function.
27474 * internal-fn.c (expand_ASAN_POISON_USE): New function.
27475 * internal-fn.def (ASAN_POISON_USE): Declare.
27476 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
27477 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
27478 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
27479 ASAN_POISON calls w/o LHS.
27480 * tree-ssa.c (execute_update_addresses_taken): Create clobber
27481 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
27482 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
27483 * gimplify.c (asan_poison_variables): Add attribute
27484 use_after_scope_memory to variables that really needs to live
27486 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
27487 having the attribute.
27489 2017-01-23 Martin Liska <mliska@suse.cz>
27491 * asan.c (create_asan_shadow_var): New function.
27492 (asan_expand_poison_ifn): Likewise.
27493 * asan.h (asan_expand_poison_ifn): New declaration.
27494 * internal-fn.c (expand_ASAN_POISON): Likewise.
27495 * internal-fn.def (ASAN_POISON): New builtin.
27496 * sanopt.c (pass_sanopt::execute): Expand
27497 asan_expand_poison_ifn.
27498 * tree-inline.c (copy_decl_for_dup_finish): Make function
27500 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
27501 * tree-ssa.c (is_asan_mark_p): New function.
27502 (execute_update_addresses_taken): Rewrite local variables
27503 (identified just by use-after-scope as addressable) into SSA.
27505 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27507 * doc/install.texi (Specific): opensource.apple.com uses https
27508 now. Remove trailing slash.
27510 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27512 * README.Portability: Remove note on an Irix compatibility issue.
27514 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
27516 * gcov.c (INCLUDE_ALGORITHM): Define.
27517 (INCLUDE_VECTOR): Define.
27518 No longer include <vector> and <algorithm> directly.
27520 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
27522 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
27524 * doc/invoke.texi (Code Gen Options): Ditto.
27526 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
27529 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
27531 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
27533 rtl-optimization/79125
27534 * cprop.c (local_cprop_pass): Handle cases where we make an
27535 unconditional trap.
27537 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
27541 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
27542 read from, for big endian.
27544 2017-01-20 Jiong Wang <jiong.wang@arm.com>
27546 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
27547 register pauth builtins for LP64 only.
27549 2017-01-20 Marek Polacek <polacek@redhat.com>
27552 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
27555 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
27557 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
27559 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
27560 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
27561 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
27563 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27566 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
27567 in big-endian mode when they are not a single duplicated value.
27569 2017-01-20 Richard Biener <rguenther@suse.de>
27571 * BASE-VER: Bump to 7.0.1.
27573 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
27575 * omp-low.c (omplow_simd_context): New struct. Use it...
27576 (lower_rec_simd_input_clauses): ...here and...
27577 (lower_rec_input_clauses): ...here to hold common data. Adjust all
27578 references to idx, lane, max_vf, is_simt.
27580 2017-01-20 Graham Markall <graham.markall@embecosm.com>
27582 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
27585 2017-01-20 Martin Jambor <mjambor@suse.cz>
27587 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
27588 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
27590 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
27591 (GTFILES): Rename hsa.c to hsa-common.c.
27592 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
27593 * hsa-dump.c: Likewise.
27594 * hsa-gen.c: Likewise.
27595 * hsa-regalloc.c: Likewise.
27596 * ipa-hsa.c: Likewise.
27597 * omp-expand.c: Likewise.
27598 * omp-low.c: Likewise.
27599 * toplev.c: Likewise.
27601 2017-01-20 Marek Polacek <polacek@redhat.com>
27604 * doc/invoke.texi: Document -Wduplicated-branches.
27605 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
27606 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
27607 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
27608 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
27609 return 0 only when not OEP_LEXICOGRAPHIC.
27610 (fold_build_cleanup_point_expr): Use the expression
27611 location when building CLEANUP_POINT_EXPR.
27612 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
27613 * tree.c (add_expr): Handle error_mark_node.
27615 2017-01-20 Martin Liska <mliska@suse.cz>
27618 * tree-profile.c (init_ic_make_global_vars): Do not call
27620 (gimple_init_gcov_profiler): Likewise.
27622 2017-01-20 Martin Liska <mliska@suse.cz>
27625 * cgraph.h (maybe_create_reference): Remove argument and
27627 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
27629 * ipa-cp.c (create_specialized_node): Likewise.
27630 * symtab.c (symtab_node::maybe_create_reference): Handle
27631 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
27633 2017-01-20 Martin Liska <mliska@suse.cz>
27635 * read-rtl-function.c (function_reader::create_function): Use
27636 build_decl instread of build_decl_stat.
27638 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
27640 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
27641 * config/i386/avx512dqintrin.h: Ditto.
27642 * config/i386/avx512fintrin.h: Ditto.
27643 * config/i386/i386-builtin-types.def: Add new types.
27644 * config/i386/i386.c: Handle new types.
27645 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
27646 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
27647 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
27648 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
27649 (__builtin_ia32_kshiftridi): New.
27650 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
27652 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
27656 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
27657 define to rs6000_init_stack_protect_guard.
27658 (rs6000_init_stack_protect_guard): New function.
27660 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27661 Yunqiang Su <yunqiang.su@imgtec.com>
27663 * config.gcc (supported_defaults): Add madd4.
27664 (with_madd4): Add validation.
27665 (all_defaults): Add madd4.
27666 * config/mips/mips.opt (mmadd4): New option.
27667 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27669 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27671 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
27672 (ISA_HAS_FUSED_MADD4): Likewise.
27673 * doc/invoke.texi (-mmadd4): Document the new option.
27674 * doc/install.texi (--with-madd4): Document the new option.
27676 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27678 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
27679 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
27680 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
27681 (aarch64_init_pauth_hint_builtins): New.
27682 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
27683 (aarch64_expand_builtin): Expand new builtins.
27685 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27687 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
27688 * combine-stack-adj.c (no_unhandled_cfa): Handle
27689 REG_CFA_TOGGLE_RA_MANGLE.
27690 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
27691 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
27692 info for return address signing.
27693 (aarch64_expand_epilogue): Likewise.
27695 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27697 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
27698 * config/aarch64/aarch64-protos.h
27699 (aarch64_return_address_signing_enabled): New declaration.
27700 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
27702 (aarch64_expand_prologue): Sign return address before it's pushed onto
27704 (aarch64_expand_epilogue): Authenticate return address fetched from
27706 (aarch64_override_options): Sanity check for ILP32 and ISA level.
27707 (aarch64_attributes): New function attributes for "sign-return-address".
27708 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
27709 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
27710 ("*do_return"): Generate combined instructions according to key index.
27711 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
27712 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
27714 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
27715 * config/aarch64/aarch64.opt (msign-return-address=): New.
27716 * doc/extend.texi (AArch64 Function Attributes): Documents
27717 "sign-return-address=".
27718 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
27720 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
27722 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
27723 overall option summary.
27725 2017-01-19 Jiong Wang <jiong.wang@arm.com>
27727 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
27728 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
27729 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
27730 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
27732 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
27734 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
27735 -mpower9-minmax by default for -mcpu=power9.
27736 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
27737 128-bit floating point.
27739 2017-01-20 Alan Modra <amodra@gmail.com>
27741 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
27742 optimizing for size.
27744 2017-01-20 Alan Modra <amodra@gmail.com>
27747 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
27748 for strcmp and strncmp from corresponding builtin decl.
27750 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27752 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
27753 instead of i386/rtems-64.h.
27754 * config/i386/rtems-64.h: Remove.
27756 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
27760 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
27762 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
27764 2017-01-19 Tamar Christina <tamar.christina@arm.com>
27766 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
27767 Change int to HOST_WIDE_INT.
27768 * config/aarch64/aarch64-protos.h
27769 (aarch64_simd_gen_const_vector_dup): Likewise.
27770 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
27772 2017-01-19 David Malcolm <dmalcolm@redhat.com>
27774 * langhooks-def.h (lhd_type_for_size): New decl.
27775 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
27776 * langhooks.c (lhd_type_for_size): New function, taken from
27779 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
27781 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
27782 define_bypass for CR latency.
27783 (power9-cracked-alu): Update bypass latency and remove power9-branch.
27784 (power9-alu2): Add define_bypass for CR latency.
27786 (power9-mul): Update insn latency.
27787 (power9-mul-compare): Update insn latency, bypass latency and remove
27790 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27792 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
27794 * config/aarch64/aarch64.md
27795 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
27796 aarch64_nopcrelative_literal_loads.
27797 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
27799 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
27801 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
27802 TARGET_LOONGSON_3A.
27803 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
27805 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
27808 * config.gcc (supported_defaults): Add lxc1-sxc1.
27809 (with_lxc1_sxc1): Add validation.
27810 (all_defaults): Add lxc1-sxc1.
27811 * config/mips/mips.opt (mlxc1-sxc1): New option.
27812 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
27814 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
27815 __mips_no_lxc1_sxc1.
27816 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
27817 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
27818 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
27820 2017-01-19 Richard Biener <rguenther@suse.de>
27822 PR tree-optimization/72488
27823 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
27824 sure to restore SSA info.
27825 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
27827 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
27829 PR rtl-optimization/79121
27830 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
27831 of the inner type when shifting an extended value.
27833 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
27836 * symtab.c (symtab_node::equal_address_to): Fix comparing of
27837 interposable aliases.
27839 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
27842 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
27843 Use the evmergelohi instruction.
27844 (mov_si<mode>_e500_subreg4_2_le): Likewise.
27845 (mov_sitf_e500_subreg8_2_be): Likewise.
27846 (mov_sitf_e500_subreg12_2_le): Likewise.
27847 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
27848 (mov_si<mode>_e500_subreg4_2_be): Likewise.
27849 (mov_sitf_e500_subreg8_2_le): Likewise.
27850 (mov_sitf_e500_subreg12_2_be): Likewise.
27852 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27854 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
27855 attribute from vecsimple to vecperm.
27856 (altivec_vbpermq2): Likewise.
27858 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27861 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
27863 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27864 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
27865 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
27866 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
27867 case where N arg is SIZE_MAX.
27868 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
27869 (cmpstrsi): Add pattern.
27871 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27873 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27874 __builtin_vec_revb builtins.
27875 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
27876 built-in functions to support generation of the ISA 3.0 XXBR<x>
27877 vector byte reverse instructions.
27878 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
27879 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
27880 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
27881 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
27882 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
27883 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
27884 (P9V_BUILTIN_VEC_REVB): Likewise.
27885 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
27886 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
27887 (p9_xxbrq_v16qi): Likewise.
27888 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
27889 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
27890 (p9_xxbrh_v8hi): Likewise.
27891 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
27892 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
27893 vec_revb built-in functions.
27895 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
27897 PR rtl-optimization/78952
27898 * config/i386/i386.md (any_extract): New code iterator.
27899 (*insvqi_2): Use any_extract for source operand.
27900 (*insvqi_3): Use any_shiftrt for source operand.
27902 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
27904 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
27906 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
27908 2017-01-18 Matthias Klose <doko@ubuntu.com>
27910 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
27912 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27914 * config/rs6000/altivec.h (vec_bperm): Change #define.
27915 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
27916 (altivec_vbpermq2): New define_insn.
27917 (altivec_vbpermd): Likewise.
27918 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
27919 function interface.
27920 (VBPERMD): Likewise.
27921 (VBPERM): New polymorphic function interface.
27922 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
27923 Add entries for P9V_BUILTIN_VEC_VBPERM.
27924 * doc/extend.texi: Add interfaces for vec_bperm.
27926 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27928 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
27929 first letter of error messages.
27930 (s390_resolve_overloaded_builtin): Likewise.
27931 * config/s390/s390.c (s390_expand_builtin): Likewise.
27932 (s390_invalid_arg_for_unprototyped_fn): Likewise.
27933 (s390_valid_target_attribute_inner_p): Likewise.
27934 * config/s390/s390.md ("tabort"): Likewise.
27936 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
27938 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
27939 (ISA_AVOID_DIV_HILO): New macro.
27940 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
27941 (ISA_HAS_DDIV): Likewise.
27943 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27945 * doc/invoke.texi (fabi-version): Correct number of occurrences.
27947 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27949 * doc/invoke.texi (fabi-version): Spelling fix.
27951 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27954 * doc/invoke.texi (fabi-version): Mention mangling fix for
27957 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
27960 * doc/invoke.texi (fabi-version): Document discriminator mangling.
27962 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
27965 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
27966 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
27968 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
27970 (stack_protect_test): Ditto.
27971 * config/rs6000/rs6000.opt (mstack-protector-guard=,
27972 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
27974 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
27975 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
27976 -mstack-protector-guard-offset=.
27977 (RS/6000 and PowerPC Options): Ditto.
27979 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
27981 * config/i386/i386.h (MASK_CLASS_P): New define.
27982 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
27983 there are no registers from different register sets also when
27984 mask registers are used. Update function comment.
27985 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
27986 to (*k/*r) and (*k/*km) alternatives.
27988 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
27990 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
27991 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
27992 (EH_RETURN_HANDLER_RTX): New define.
27993 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
27994 Force frame pointer in EH return functions.
27995 (aarch64_expand_epilogue): Add barrier for eh_return.
27996 (aarch64_final_eh_return_addr): Remove.
27997 (aarch64_eh_return_handler_rtx): New function.
27998 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
28000 (aarch64_eh_return_handler_rtx): New prototype.
28002 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28004 * config/rs6000/altivec.h (vec_rlmi): New #define.
28005 (vec_vrlnm): Likewise.
28006 (vec_rlnm): Likewise.
28007 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
28008 (UNSPEC_VRLNM): Likewise.
28009 (VIlong): New mode iterator.
28010 (altivec_vrl<VI_char>mi): New define_insn.
28011 (altivec_vrl<VI_char>nm): Likewise.
28012 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
28014 (VRLDNM): Likewise.
28015 (RLNM): New polymorphic function entry.
28016 (VRLWMI): New monomorphic function entry.
28017 (VRLDMI): Likewise.
28018 (RLMI): New polymorphic function entry.
28019 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
28020 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
28021 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
28024 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28027 * dwarf2out.c (field_byte_offset): Restore the
28028 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
28029 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
28030 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
28033 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
28036 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
28038 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28041 * dwarf2out.c (add_data_member_location_attribute): For constant
28042 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
28043 instead of DW_AT_data_member_location, DW_AT_bit_offset and
28044 DW_AT_byte_size attributes.
28046 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
28048 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
28049 after forcing to constant memory when the code model is medium.
28051 2017-01-17 Julia Koval <julia.koval@intel.com>
28054 * config/i386/avx512fintrin.h
28055 (_mm512_i32gather_ps): Change __addr type to void const*.
28056 (_mm512_mask_i32gather_ps): Ditto.
28057 (_mm512_i32gather_pd): Ditto.
28058 (_mm512_mask_i32gather_pd): Ditto.
28059 (_mm512_i64gather_ps): Ditto.
28060 (_mm512_mask_i64gather_ps): Ditto.
28061 (_mm512_i64gather_pd): Ditto.
28062 (_mm512_mask_i64gather_pd): Ditto.
28063 (_mm512_i32gather_epi32): Ditto.
28064 (_mm512_mask_i32gather_epi32): Ditto.
28065 (_mm512_i32gather_epi64): Ditto.
28066 (_mm512_mask_i32gather_epi64): Ditto.
28067 (_mm512_i64gather_epi32): Ditto.
28068 (_mm512_mask_i64gather_epi32): Ditto.
28069 (_mm512_i64gather_epi64): Ditto.
28070 (_mm512_mask_i64gather_epi64): Ditto.
28071 (_mm512_i32scatter_ps): Change __addr type to void*.
28072 (_mm512_mask_i32scatter_ps): Ditto.
28073 (_mm512_i32scatter_pd): Ditto.
28074 (_mm512_mask_i32scatter_pd): Ditto.
28075 (_mm512_i64scatter_ps): Ditto.
28076 (_mm512_mask_i64scatter_ps): Ditto.
28077 (_mm512_i64scatter_pd): Ditto.
28078 (_mm512_mask_i64scatter_pd): Ditto.
28079 (_mm512_i32scatter_epi32): Ditto.
28080 (_mm512_mask_i32scatter_epi32): Ditto.
28081 (_mm512_i32scatter_epi64): Ditto.
28082 (_mm512_mask_i32scatter_epi64): Ditto.
28083 (_mm512_i64scatter_epi32): Ditto.
28084 (_mm512_mask_i64scatter_epi32): Ditto.
28085 (_mm512_i64scatter_epi64): Ditto.
28086 (_mm512_mask_i64scatter_epi64): Ditto.
28087 * config/i386/avx512pfintrin.h
28088 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
28089 (_mm512_mask_prefetch_i32gather_ps): Ditto.
28090 (_mm512_mask_prefetch_i64gather_pd): Ditto.
28091 (_mm512_mask_prefetch_i64gather_ps): Ditto.
28092 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
28093 (_mm512_prefetch_i32scatter_ps): Ditto.
28094 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
28095 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
28096 (_mm512_prefetch_i64scatter_pd): Ditto.
28097 (_mm512_prefetch_i64scatter_ps): Ditto.
28098 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
28099 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
28100 * config/i386/avx512vlintrin.h
28101 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
28102 (_mm_mmask_i32gather_ps): Ditto.
28103 (_mm256_mmask_i32gather_pd): Ditto.
28104 (_mm_mmask_i32gather_pd): Ditto.
28105 (_mm256_mmask_i64gather_ps): Ditto.
28106 (_mm_mmask_i64gather_ps): Ditto.
28107 (_mm256_mmask_i64gather_pd): Ditto.
28108 (_mm_mmask_i64gather_pd): Ditto.
28109 (_mm256_mmask_i32gather_epi32): Ditto.
28110 (_mm_mmask_i32gather_epi32): Ditto.
28111 (_mm256_mmask_i32gather_epi64): Ditto.
28112 (_mm_mmask_i32gather_epi64): Ditto.
28113 (_mm256_mmask_i64gather_epi32): Ditto.
28114 (_mm_mmask_i64gather_epi32): Ditto.
28115 (_mm256_mmask_i64gather_epi64): Ditto.
28116 (_mm_mmask_i64gather_epi64): Ditto.
28117 (_mm256_i32scatter_ps): Change __addr type to void*.
28118 (_mm256_mask_i32scatter_ps): Ditto.
28119 (_mm_i32scatter_ps): Ditto.
28120 (_mm_mask_i32scatter_ps): Ditto.
28121 (_mm256_i32scatter_pd): Ditto.
28122 (_mm256_mask_i32scatter_pd): Ditto.
28123 (_mm_i32scatter_pd): Ditto.
28124 (_mm_mask_i32scatter_pd): Ditto.
28125 (_mm256_i64scatter_ps): Ditto.
28126 (_mm256_mask_i64scatter_ps): Ditto.
28127 (_mm_i64scatter_ps): Ditto.
28128 (_mm_mask_i64scatter_ps): Ditto.
28129 (_mm256_i64scatter_pd): Ditto.
28130 (_mm256_mask_i64scatter_pd): Ditto.
28131 (_mm_i64scatter_pd): Ditto.
28132 (_mm_mask_i64scatter_pd): Ditto.
28133 (_mm256_i32scatter_epi32): Ditto.
28134 (_mm256_mask_i32scatter_epi32): Ditto.
28135 (_mm_i32scatter_epi32): Ditto.
28136 (_mm_mask_i32scatter_epi32): Ditto.
28137 (_mm256_i32scatter_epi64): Ditto.
28138 (_mm256_mask_i32scatter_epi64): Ditto.
28139 (_mm_i32scatter_epi64): Ditto.
28140 (_mm_mask_i32scatter_epi64): Ditto.
28141 (_mm256_i64scatter_epi32): Ditto.
28142 (_mm256_mask_i64scatter_epi32): Ditto.
28143 (_mm_i64scatter_epi32): Ditto.
28144 (_mm_mask_i64scatter_epi32): Ditto.
28145 (_mm256_i64scatter_epi64): Ditto.
28146 (_mm256_mask_i64scatter_epi64): Ditto.
28147 (_mm_i64scatter_epi64): Ditto.
28148 (_mm_mask_i64scatter_epi64): Ditto.
28149 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
28150 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
28151 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
28152 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
28153 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
28154 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
28155 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
28156 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
28157 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
28158 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
28159 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
28160 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
28161 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
28162 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
28163 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
28164 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
28165 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
28166 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
28167 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
28168 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
28169 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
28170 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
28171 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
28172 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
28173 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
28174 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
28175 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
28176 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
28177 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
28178 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
28179 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
28180 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
28181 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
28182 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
28183 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
28184 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
28185 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
28186 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
28187 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
28188 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
28189 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
28190 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
28191 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
28192 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
28193 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
28194 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
28195 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
28196 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
28197 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
28198 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
28199 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
28200 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
28201 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
28202 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
28203 definitions accordingly.
28205 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
28206 Kuan-Lin Chen <kuanlinchentw@gmail.com>
28209 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
28212 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
28215 * ira-conflicts.c (ira_build_conflicts): Update total conflict
28216 hard regs for inner regno.
28218 2017-01-17 Martin Liska <mliska@suse.cz>
28221 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
28222 assumption and add comment.
28224 2017-01-17 Nathan Sidwell <nathan@acm.org>
28226 * ipa-visibility.c (localize_node): New function, broken out of ...
28227 (function_and_variable_visibility): ... here. Call it.
28229 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
28231 PR middle-end/77445
28232 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
28233 correctly set frequency of oudgoing edge.
28234 (duplicate_thread_path): Fix profile updating.
28236 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28239 * configure.ac: Add GCC_BASE_VER.
28240 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
28241 version from BASE-VER file.
28242 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
28243 (gcc.o): Depend on $(BASEVER).
28244 * common.opt (dumpfullversion): New option.
28245 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
28246 * doc/invoke.texi: Document -dumpfullversion.
28247 * doc/install.texi: Document --with-gcc-major-version-only.
28248 * configure: Regenerated.
28250 2017-01-17 Richard Biener <rguenther@suse.de>
28252 PR tree-optimization/71433
28253 * tree-vrp.c (register_new_assert_for): Merge same asserts
28254 on all incoming edges.
28255 (process_assert_insertions_for): Handle insertions at the
28258 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
28260 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
28261 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
28263 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
28266 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
28269 2017-01-17 Alan Modra <amodra@gmail.com>
28272 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
28273 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
28274 symbolic stack limit when pic.
28276 2017-01-16 Martin Sebor <msebor@redhat.com>
28278 PR tree-optimization/78608
28279 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
28281 2017-01-16 Jeff Law <law@redhat.com>
28284 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
28285 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
28286 for several include directories that may be relative to sysroot.
28287 * config/i386/x-mingw32 (gplus_includedir): Define.
28288 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
28289 (native_system_includedir): Likewise.
28290 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
28291 override if TARGET_SYSTEM_ROOT is defined.
28292 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
28294 PR tree-optimization/79090
28295 PR tree-optimization/33562
28296 PR tree-optimization/61912
28297 PR tree-optimization/77485
28298 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
28299 and computed trims into the dump file.
28301 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
28303 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
28305 2017-01-16 Jakub Jelinek <jakub@redhat.com>
28308 * gimplify.c (gimplify_init_constructor): If want_value and
28309 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
28313 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
28314 sequence. Formatting fixes.
28315 (doloop_optimize): Formatting fixes.
28318 * gcc.c (debug_level_greater_than_spec_func): New function.
28319 (static_spec_functions): Add debug-level-gt spec function.
28320 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
28322 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
28323 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
28324 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
28325 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
28326 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
28327 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
28329 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
28331 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
28332 QImode fixups to general and mask registers only.
28334 2017-01-16 Carl Love <cel@us.ibm.com>
28336 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
28337 for built-in functions
28338 vector signed char vec_nabs (vector signed char)
28339 vector signed short vec_nabs (vector signed short)
28340 vector signed int vec_nabs (vector signed int)
28341 vector signed long long vec_nabs (vector signed long long)
28342 vector float vec_nabs (vector float)
28343 vector double vec_nabs (vector double)
28344 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
28346 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
28347 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
28348 * doc/extend.texi: Update the documentation file for the new built-in
28351 2017-01-16 Martin Sebor <msebor@redhat.com>
28353 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
28356 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28358 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
28359 UNSPEC_VSX__XXSPLTD to require special splat handling.
28361 2017-01-16 David Malcolm <dmalcolm@redhat.com>
28364 * system.h: Poison strndup.
28366 2017-01-16 Alan Modra <amodra@gmail.com>
28369 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
28372 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
28374 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
28376 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
28378 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
28379 call recog here. Assert that INSN_CODE (insn) is non-negative.
28381 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28384 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
28386 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
28387 in the currently scheduled RTL fragment.
28389 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28391 PR rtl-optimization/78751
28392 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
28395 2017-01-14 Jeff Law <law@redhat.com>
28397 PR tree-optimization/79090
28398 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
28399 variable length stores.
28400 (compute_trims): Delete dead assignment to *trim_tail.
28401 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
28404 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
28406 PR rtl-optimization/78626
28407 PR rtl-optimization/78727
28408 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
28409 of a block, and split such blocks after everything else is finished.
28411 2017-01-14 Alan Modra <amodra@gmail.com>
28414 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
28415 target legitimate_combined_insn.
28416 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
28417 (rs6000_legitimate_combined_insn): New function.
28418 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
28420 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
28421 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
28422 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
28424 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
28426 * doc/frontends.texi (G++ and GCC): Remove references to Java.
28428 2017-01-13 Jeff Law <law@redhat.com>
28430 PR tree-optimization/33562
28431 PR tree-optimization/61912
28432 PR tree-optimization/77485
28433 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
28435 (delete_dead_assignment): Likewise.
28436 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
28437 statement to delete_dead_call and delete_dead_assignment.
28439 2017-01-13 David Malcolm <dmalcolm@redhat.com>
28442 * substring-locations.c (format_warning_va): Strengthen case 1 so
28443 that both endpoints of the substring must be within the format
28444 range for just the substring to be printed.
28446 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
28448 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
28449 * config/i386/i386.c (ix86_target_string): Add missing options
28450 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
28451 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
28452 flags_other and ix86_target_other to flags2_other. Display unknown
28454 (ix86_valid_target_attribute_inner_p): Add missing options and
28455 reorder options by implied ISAs, as in ix86_target_string.
28457 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28459 * hash-table.h (hash_table::too_empty_p): New function.
28460 (hash_table::expand): Use it.
28461 (hash_table::traverse): Likewise.
28462 (hash_table::empty_slot): Use sizeof (value_type) instead of
28463 sizeof (PTR) to convert bytes to elements. Shrink the table
28464 if the current size is excessive for the current number of
28467 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28469 * ira-costs.c (record_reg_classes): Break from the inner loop
28470 early once alt_fail is known to be true. Update outer loop
28471 handling accordingly.
28473 2017-01-13 Jeff Law <law@redhat.com>
28475 * tree-ssa-dse.c (decrement_count): New function.
28476 (increment_start_addr, maybe_trim_memstar_call): Likewise.
28477 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
28478 when we know the partially dead statement is a mem* function.
28480 PR tree-optimization/61912
28481 PR tree-optimization/77485
28482 * tree-ssa-dse.c: Include expr.h.
28483 (maybe_trim_constructor_store): New function.
28484 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
28486 PR tree-optimization/33562
28487 PR tree-optimization/61912
28488 PR tree-optimization/77485
28489 * doc/invoke.texi: Document new dse-max-object-size param.
28490 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
28491 * tree-ssa-dse.c: Include params.h.
28492 (dse_store_status): New enum.
28493 (initialize_ao_ref_for_dse): New, partially extracted from
28495 (valid_ao_ref_for_dse, normalize_ref): New.
28496 (setup_live_bytes_from_ref, compute_trims): Likewise.
28497 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
28498 (maybe_trim_partially_dead_store): Likewise.
28499 (maybe_trim_complex_store): Likewise.
28500 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
28501 Track what bytes live from the original store. Return tri-state
28502 for dead, partially dead or live.
28503 (dse_dom_walker): Add constructor, destructor and new private members.
28504 (delete_dead_call, delete_dead_assignment): New extracted from
28506 (dse_optimize_stmt): Make a member of dse_dom_walker.
28507 Use initialize_ao_ref_for_dse.
28509 PR tree-optimization/33562
28510 PR tree-optimization/61912
28511 PR tree-optimization/77485
28512 * sbitmap.h (bitmap_count_bits): Prototype.
28513 (bitmap_clear_range, bitmap_set_range): Likewise.
28514 * sbitmap.c (bitmap_clear_range): New function.
28515 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
28517 2017-01-13 Martin Liska <mliska@suse.cz>
28520 * function.c (set_cfun): Add new argument force.
28521 * function.h (set_cfun): Likewise.
28522 * ipa-inline-transform.c (inline_call): Use the function when
28523 strict alising from is dropped for function we inline to.
28525 2017-01-13 Richard Biener <rguenther@suse.de>
28527 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
28528 for dumping GIMPLE INTEGER_CSTs.
28530 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28532 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
28533 to 201112L since C++17.
28535 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
28538 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
28539 if -fsanitize=kernel-address is present.
28541 2017-01-13 Richard Biener <rguenther@suse.de>
28543 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
28544 as _Literal ( type ) number in case usual suffixes do not
28545 preserve all information.
28547 2017-01-13 Richard Biener <rguenther@suse.de>
28549 PR tree-optimization/77283
28550 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
28551 and ssa-iterators.h.
28552 (is_feasible_trace): Implement a cost model based on joiner
28555 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
28558 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
28559 char or short to __float128/_Float128 directly.
28561 2017-01-12 Martin Sebor <msebor@redhat.com>
28563 to -Wformat-overflow.
28564 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
28565 (min_bytes_remaining): Same.
28566 (get_string_length): Same.
28567 (format_string): Same.
28568 (format_directive): Same.
28570 (pass_sprintf_length::handle_gimple_call): Same.
28572 2017-01-12 Jakub Jelinek <jakub@redhat.com>
28574 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
28575 info.nowrite calls with no lhs that can't throw. Return bool
28576 whether gsi_remove has been called or not.
28577 (pass_sprintf_length::handle_gimple_call): Return bool whether
28578 try_substitute_return_value called gsi_remove. Formatting fix.
28579 (pass_sprintf_length::execute): Don't use gsi_remove if
28580 handle_gimple_call returned true.
28583 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
28584 be removed due to side-effects, don't remove following barrier nor
28585 turn the successor edge into fallthru edge.
28587 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28590 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
28591 element-reversing loads and stores as not swappable.
28593 2017-01-12 Nathan Sidwell <nathan@acm.org>
28594 Nicolai Stange <nicstange@gmail.com>
28596 * combine.c (try_combine): Don't ignore result of overlap checking
28597 loop. Combine overlap & asm check into single loop.
28599 2017-01-12 Richard Biener <rguenther@suse.de>
28601 * tree-pretty-print.c (dump_generic_node): Provide -gimple
28602 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
28604 2017-01-12 Richard Biener <rguenther@suse.de>
28606 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
28607 and TS_TARGET_OPTION directly derive from TS_BASE.
28608 * tree-core.h (tree_optimization_option): Derive from tree_base.
28609 (tree_target_option): Likewise.
28611 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
28613 * config/i386/i386.c (memory_address_length): Increase len
28614 only when rip_relative_addr_p returns false.
28616 2017-01-11 Julia Koval <julia.koval@intel.com>
28618 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
28619 (OPTION_MASK_ISA_SGX_SET): New.
28620 (ix86_handle_option): Handle OPT_msgx.
28621 * config.gcc: Added sgxintrin.h.
28622 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
28623 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
28624 * config/i386/i386.c (ix86_target_string): Add -msgx.
28626 (ix86_option_override_internal): Handle new options.
28627 (ix86_valid_target_attribute_inner_p): Add sgx.
28628 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
28629 * config/i386/i386.opt: Add msgx.
28630 * config/i386/sgxintrin.h: New file.
28631 * config/i386/x86intrin.h: Add sgxintrin.h.
28633 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28636 * fold-const.c (maybe_nonzero_address): Return 1 for function
28638 (tree_single_nonzero_warnv_p): Don't handle function local objects
28642 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
28645 2017-01-11 David Malcolm <dmalcolm@redhat.com>
28648 * opts.c: Include "spellcheck.h"
28649 (struct string_fragment): New struct.
28650 (struct edit_distance_traits<const string_fragment &>): New
28652 (get_closest_sanitizer_option): New function.
28653 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
28655 2017-01-11 Jakub Jelinek <jakub@redhat.com>
28657 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
28659 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
28660 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
28661 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
28662 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
28663 for initial die_offset if dwarf_split_debug_info.
28664 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
28665 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
28667 (output_skeleton_debug_sections): Formatting fix. Use
28668 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
28669 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
28671 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
28673 * config/arm/cortex-a53.md: Add bypasses for
28674 cortex_a53_r2f_cvt.
28675 (cortex_a53_r2f): Only use for transfers.
28676 (cortex_a53_f2r): Likewise.
28677 (cortex_a53_r2f_cvt): Add reservation for conversions.
28678 (cortex_a53_f2r_cvt): Likewise.
28680 2017-01-11 Tamar Christina <tamar.christina@arm.com>
28682 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
28683 to all inlined functions, change static to extern.
28685 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
28688 * config/arm/arm.c (legitimize_pic_address): Handle reference to
28690 (arm_assemble_integer): Likewise.
28692 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
28694 * config.gcc: Use new awk script to check CPU, FPU and architecture
28695 parameters for --with-... options.
28696 * config/arm/parsecpu.awk: New file
28697 * config/arm/arm-cpus.in: New file.
28698 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
28700 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
28702 * config/arm/t-arm: Update dependency rules.
28703 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
28704 of processing .def files.
28705 * config/arm/genopt.sh: Deleted.
28706 * config/arm/gentune.sh: Deleted.
28707 * config/arm/arm-cores.def: Deleted.
28708 * config/arm/arm-arches.def: Deleted.
28709 * config/arm/arm-fpus.def: Deleted.
28710 * config/arm/arm-tune.md: Regenerated.
28711 * config/arm/arm-tables.opt: Regenerated.
28712 * config/arm/arm-cpu.h: New generated file.
28713 * config/arm/arm-cpu-data.h: New generated file.
28714 * config/arm/arm-cpu-cdata.h: New generated file.
28716 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
28719 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
28721 (input_varpool_node): Unpack dynamically_initialized bit.
28723 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
28725 PR rtl-optimization/79032
28726 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
28727 the alignment of the adjusted memory reference against that of MODE,
28728 instead of the alignment of the original memory reference.
28730 2017-01-11 Martin Jambor <mjambor@suse.cz>
28732 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
28734 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
28735 decorated functions.
28737 2017-01-11 Richard Biener <rguenther@suse.de>
28739 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
28740 set range/nonnull info for PHI results. Do not set it on
28741 stmts marked for removal.
28743 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
28745 * expr.c (store_field): In the bitfield case, fetch the return value
28746 from the registers before applying a single big-endian adjustment.
28747 Always do a final load for a BLKmode value not larger than a word.
28749 2017-01-10 David Malcolm <dmalcolm@redhat.com>
28752 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
28753 that we correctly handle column numbers greater than
28754 LINE_MAP_MAX_COLUMN_NUMBER.
28756 2017-01-10 Martin Sebor <msebor@redhat.com>
28758 PR middle-end/78245
28759 * gimple-ssa-sprintf.c (get_destination_size): Call
28760 {init,fini}object_sizes.
28761 * tree-object-size.c (addr_object_size): Adjust.
28762 (pass_through_call): Adjust.
28763 (pass_object_sizes::execute): Adjust.
28764 * tree-object-size.h (fini_object_sizes): Declare.
28766 2017-01-10 Martin Sebor <msebor@redhat.com>
28768 PR tree-optimization/78775
28769 * builtins.c (get_size_range): Move...
28770 * calls.c: ...to here.
28771 (alloc_max_size): Accept zero argument.
28772 (operand_signed_p): Remove.
28773 (maybe_warn_alloc_args_overflow): Call get_size_range.
28774 * calls.h (get_size_range): Declare.
28776 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
28778 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
28779 from TI's devices.csv file as of September 2016.
28780 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
28782 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
28784 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
28785 * doc/invoke.texi: Likewise.
28786 * doc/md.texi: Likewise.
28787 * doc/objc.texi: Likewise.
28789 2017-01-10 Joshua Conner <joshconner@google.com>
28791 * config/arm/fuchsia-elf.h: New file.
28792 * config/fuchsia.h: New file.
28793 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
28794 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
28796 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
28798 2016-01-10 Richard Biener <rguenther@suse.de>
28800 PR tree-optimization/79034
28801 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28802 Propagate out degenerate PHIs in the joiner.
28804 2017-01-10 Martin Liska <mliska@suse.cz>
28806 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
28807 (sort_congruence_classes_by_decl_uid): Likewise.
28808 (sort_congruence_class_groups_by_decl_uid): Likewise.
28809 (sem_item_optimizer::merge_classes): Sort class, groups in these
28810 classes and members in the groups by DECL_UID of declarations.
28811 This would make merge operations stable.
28813 2017-01-10 Martin Liska <mliska@suse.cz>
28815 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
28816 usage of m_classes_vec.
28817 (sem_item_optimizer::~sem_item_optimizer): Likewise.
28818 (sem_item_optimizer::get_group_by_hash): Likewise.
28819 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
28820 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
28821 (sem_item_optimizer::verify_classes): Likewise.
28822 (sem_item_optimizer::process_cong_reduction): Likewise.
28823 (sem_item_optimizer::dump_cong_classes): Likewise.
28824 (sem_item_optimizer::merge_classes): Likewise.
28825 * ipa-icf.h (congruence_class_hash): Rename from
28826 congruence_class_group_hash. Remove declaration of m_classes_vec.
28828 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
28830 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
28831 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
28832 * config.gcc: Add avx512vpopcntdqintrin.h.
28833 * config/i386/avx512vpopcntdqintrin.h: New.
28834 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
28835 * config/i386/i386-builtin-types.def: Add new types.
28836 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
28837 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
28838 __builtin_ia32_vpopcountq_v8di_mask): New.
28839 * config/i386/i386-c.c (ix86_target_macros_internal): Define
28840 __AVX512VPOPCNTDQ__.
28841 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
28842 (PTA_AVX512VPOPCNTDQ): Define.
28843 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
28844 TARGET_AVX512VPOPCNTDQ_P): Define.
28845 * config/i386/i386.opt: Add mavx512vpopcntdq.
28846 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
28847 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
28849 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
28851 PR middle-end/77484
28852 * predict.def (PRED_CALL): Set to 67.
28854 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
28856 * expr.c (store_field): In the bitfield case, if the value comes from
28857 a function call and is of an aggregate type returned in registers, do
28858 not modify the field mode; extract the value in all cases if the mode
28859 is BLKmode and the size is not larger than a word.
28861 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
28864 * config/i386/cpuid.h: Fix undefined behavior.
28866 2017-01-04 Jeff Law <law@redhat.com>
28868 PR tree-optimization/79007
28869 PR tree-optimization/67955
28870 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
28871 conservative for pt.null when flag_non_call_exceptions is on.
28873 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28875 PR translation/79019
28876 PR translation/79020
28877 * params.def (PARAM_INLINE_MIN_SPEEDUP,
28878 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
28879 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
28881 * config/avr/avr.opt (maccumulate-args): Likewise.
28882 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
28883 * common.opt (freport-bug): Likewise.
28884 * cif-code.def (CIF_FINAL_ERROR): Likewise.
28885 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
28886 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
28887 translatable string.
28888 * config/i386/i386.c (function_value_32): Likewise.
28889 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
28890 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
28892 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
28893 * common/config/msp430/msp430-common.c (msp430_handle_option):
28895 * symtab.c (symtab_node::verify_base): Likewise.
28896 * opts.c (set_debug_level): Likewise.
28897 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
28898 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
28899 missing whitespace to translatable strings.
28900 * config/avr/avr.md (bswapsi2): Fix typo in comment.
28901 * config/sh/superh.h: Likewise.
28902 * config/i386/xopintrin.h: Likewise.
28903 * config/i386/znver1.md: Likewise.
28904 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
28905 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
28906 * double-int.h (struct double_int): Likewise.
28907 * double-int.c (div_and_round_double): Likewise.
28908 * wide-int.cc: Likewise.
28909 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
28910 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
28911 * cfgcleanup.c (crossjumps_occured): Renamed to ...
28912 (crossjumps_occurred): ... this.
28913 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
28916 PR tree-optimization/78899
28917 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
28918 returning bool return struct loop *, NULL for failure and the new
28920 (versionable_outer_loop_p): Don't version outer loop if it has
28921 dont_vectorized bit set.
28922 (tree_if_conversion): When versioning outer loop, ensure
28923 tree_if_conversion is performed also on the inner loop of the
28924 non-vectorizable outer loop copy.
28925 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
28926 LOOP_VECTORIZED in inner loop of the scalar outer loop and
28927 prevent vectorization of it.
28928 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
28929 the outer loop vectorization of the non-scalar version is attempted
28930 before vectorization of the inner loop in scalar version. If
28931 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
28932 vectorization of its inner loop.
28933 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
28934 has 2 inner loops, rename also on edges from bb whose single pred
28935 is outer_loop->header. Fix typo in function comment.
28937 2017-01-09 Martin Sebor <msebor@redhat.com>
28940 * asan.c (asan_emit_stack_protection): Increase local buffer size
28941 to avoid snprintf truncation warning.
28943 2017-01-09 Andrew Pinski <apinski@cavium.com>
28945 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
28946 to reference thunderx2t99 for the tuning structure
28947 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
28949 (thunderx2t99_extra_costs): This.
28950 * config/aarch64/aarch64-tune.md: Regenerate.
28951 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
28952 (vulcan_addrcost_table): This.
28953 (vulcan_regmove_cost): Rename to ...
28954 (thunderx2t99_regmove_cost): This.
28955 (vulcan_vector_cost): Rename to ...
28956 (thunderx2t99_vector_cost): this.
28957 (vulcan_branch_cost): Rename to ...
28958 (thunderx2t99_branch_cost): This.
28959 (vulcan_tunings): Rename to ...
28960 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
28961 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
28963 2017-01-09 Martin Jambor <mjambor@suse.cz>
28967 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
28968 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
28969 (propagate_vr_accross_jump_function): Use the above function for all
28970 value range computations for pass-through jump functions and type
28971 converasion from explicit value range values.
28972 (ipcp_propagate_stage): Do not attempt to deduce types of formal
28973 parameters from TYPE_ARG_TYPES.
28974 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
28975 (ipa_write_node_info): Stream type of the actual argument.
28976 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
28978 2017-01-09 Martin Liska <mliska@suse.cz>
28981 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
28982 (lookup_compiler): Do not show error message with have_E.
28984 2017-01-09 Jakub Jelinek <jakub@redhat.com>
28986 PR tree-optimization/78938
28987 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
28988 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
28989 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
28990 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
28993 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28995 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
28998 2017-01-09 Richard Biener <rguenther@suse.de>
29000 PR tree-optimization/78997
29001 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
29002 name condition properly.
29004 2017-01-09 Richard Biener <rguenther@suse.de>
29007 * dwarf2out.c (is_cxx): New overload with context.
29008 (is_naming_typedef_decl): Use it.
29010 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29012 * invoke.texi (Option Summary): Correct spacing in option lists
29013 and add line breaks to fix over-long lines.
29015 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29017 PR middle-end/17660
29019 * extend.texi (Common Variable Attributes): Add xref to GCC
29020 Internals manual to explain mode attribute keywords.
29022 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29025 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
29026 and Preprocessor Options.
29027 (Options for Linking): Document -pthread here....
29028 (RS/6000 and PowerPC Options): ...not here.
29029 (Solaris 2 Options): ...or here.
29030 * doc/cppopts.texi: Document -pthread.
29032 2017-01-08 Martin Sebor <msebor@redhat.com>
29034 PR middle-end/77708
29035 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
29036 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
29037 New member functions.
29038 (format_directive): Used them.
29040 (pass_sprintf_length::handle_gimple_call): Same.
29041 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
29042 to avoid truncation for any argument.
29043 (extract_affine_mul): Same.
29044 * tree.c (get_file_function_name): Same.
29046 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29048 PR middle-end/77484
29049 * predict.def (PRED_INDIR_CALL): Set to 86.
29051 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29053 PR preprocessor/54124
29054 * doc/cppopts.texi: Reformat -d subtable to list the full name
29055 of the options. Add cross-reference to the docs for the general
29056 compiler -d options.
29057 * doc/invoke.texi (Developer Options): Add cross-reference to the
29058 preprocessor-specific -d option documentation.
29060 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29062 PR preprocessor/13498
29063 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
29064 redudant material, and reflect new command-line options.
29065 (System Headers): Likewise.
29067 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29069 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
29070 -isystem, and -idirafter. Copy-edit.
29071 * doc/cppopts.texi: Copy-edit. Remove contradiction about
29072 default for -ftrack-macro-expansion. Delete obsolete and
29073 badly-formatted implementation details about -fdebug-cpp output.
29074 * doc/cppwarnopts.texi: Copy-edit.
29076 2017-01-07 David Malcolm <dmalcolm@redhat.com>
29079 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
29080 that the transition from a max line width >= 1<<10 to narrower
29081 lines works correctly.
29083 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
29085 * doc/options.texi (PerFunction): New.
29086 * opt-functions.awk (switch_flags): Map both Optimization and
29087 PerFunction to CL_OPTIMIZATION.
29088 * opth-gen.awk: Test for PerFunction flag along with
29090 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
29091 it only when the latter is present. Skip those that don't in
29092 the hash function generator.
29093 * common.opt (fvar-tracking): Mark as PerFunction instead of
29095 (fvar-tracking-assignments): Likewise.
29096 (fvar-tracking-assignments-toggle): Likewise.
29097 (fvar-tracking-uninit): Likewise.
29099 2017-01-07 Jakub Jelinek <jakub@redhat.com>
29101 PR translation/79018
29102 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
29105 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
29108 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
29109 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
29110 TARGET_LONG_JUMP_TABLE_OFFSETS.
29111 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
29112 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
29113 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
29114 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
29115 * config/m68k/m68k.md (tablejump expander): Likewise.
29116 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
29117 TARGET_LONG_JUMP_TABLE_OFFSETS.
29118 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
29119 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
29121 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
29122 David Holsgrove <david.holsgrove@xilinx.com>
29124 * common/config/microblaze/microblaze-common.c
29125 (TARGET_EXCEPT_UNWIND_INFO): Remove.
29126 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
29128 * config/microblaze/microblaze.c (microblaze_must_save_register)
29129 (microblaze_expand_epilogue, microblaze_return_addr): Handle
29131 (microblaze_eh_return): New function.
29132 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
29133 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
29134 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
29135 * config/microblaze/microblaze.md (eh_return): New pattern.
29137 2017-01-06 Jakub Jelinek <jakub@redhat.com>
29139 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
29140 GCC_DIAGNOSTIC_STRINGIFY): Define.
29142 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
29144 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29146 * config/arm/arm.md (<mcrr>): New.
29148 * config/arm/arm.c (arm_arch5te): New.
29149 (arm_option_override): Set arm_arch5te.
29150 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
29152 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
29153 (arm_mcrr_qualifiers): ... this. New.
29154 (MRRC_QUALIFIERS): Define to...
29155 (arm_mrrc_qualifiers): ... this. New.
29156 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
29158 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
29159 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
29160 (MRRCI, mrrc, MRRC): New.
29161 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
29162 VUNSPEC_MRRC2): New.
29164 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29166 * config/arm/arm.md (<mcr>): New.
29168 * config/arm/arm.c (arm_coproc_builtin_available): Add
29169 support for mcr, mrc, mcr2 and mrc2.
29170 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
29171 (arm_mcr_qualifiers): ... this. New.
29172 (MRC_QUALIFIERS): Define to ...
29173 (arm_mrc_qualifiers): ... this. New.
29174 (MCR_QUALIFIERS): Define to ...
29175 (arm_mcr_qualifiers): ... this. New.
29176 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
29178 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
29179 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
29180 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
29181 VUNSPEC_MRC2): New.
29183 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29185 * config/arm/arm.md (*ldc): New.
29189 * config/arm/arm.c (arm_coproc_builtin_available): Add
29190 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
29191 (arm_coproc_ldc_stc_legitimate_address): New.
29192 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
29193 'qualifier_const_pointer'.
29194 (LDC_QUALIFIERS): Define to...
29195 (arm_ldc_qualifiers): ... this. New.
29196 (STC_QUALIFIERS): Define to...
29197 (arm_stc_qualifiers): ... this. New.
29198 * config/arm/arm-protos.h
29199 (arm_coproc_ldc_stc_legitimate_address): New.
29200 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
29201 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
29202 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
29203 stc2, stcl, stc2l): New.
29204 * config/arm/constraints.md (Uz): New.
29205 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
29206 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
29207 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
29208 VUNSPEC_STC2L): New.
29210 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29212 * config/arm/arm.md (<cdp>): New.
29213 * config/arm/arm.c (neon_const_bounds): Rename this ...
29214 (arm_const_bounds): ... this.
29215 (arm_coproc_builtin_available): New.
29216 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
29217 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
29218 (CDP_QUALIFIERS): Define to...
29219 (arm_cdp_qualifiers): ... this. New.
29221 (arm_expand_builtin_args): Add case for 6 arguments.
29222 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
29223 (arm_const_bounds): ... this.
29224 (arm_coproc_builtin_available): New.
29225 * config/arm/arm_acle.h (__arm_cdp): New.
29227 * config/arm/arm_acle_builtins.def (cdp): New.
29229 * config/arm/iterators.md (CDPI,CDP,cdp): New.
29230 * config/arm/neon.md: Rename all 'neon_const_bounds' to
29231 'arm_const_bounds'.
29232 * config/arm/types.md (coproc): New.
29233 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
29234 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
29235 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
29236 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
29238 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29240 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
29241 (UBINOP_QUALIFIERS): New.
29243 (acle_builtin_data): New. Change comment.
29244 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
29245 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
29246 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
29247 arm_acle_builtins.def.
29248 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
29249 (arm_init_acle_builtins): New.
29250 (CRC32_BUILTIN): Remove.
29251 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
29252 crc32cb, crc32ch and crc32cw.
29253 (arm_init_crc32_builtins): Remove.
29254 (arm_init_builtins): Use arm_init_acle_builtins rather
29255 than arm_init_crc32_builtins.
29256 (arm_expand_acle_builtin): New.
29257 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
29258 * config/arm/arm_acle_builtins.def: New.
29260 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29262 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
29263 (arm_builtin_datum): ... this.
29264 (arm_init_neon_builtin): Rename to ...
29265 (arm_init_builtin): ... this. Add a new parameters PREFIX
29266 and USE_SIG_IN_NAME.
29267 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
29268 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
29269 'arm_builtin_datum'.
29270 (arm_init_vfp_builtins): Likewise.
29271 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
29272 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
29273 (arm_expand_neon_args): Rename to ...
29274 (arm_expand_builtin_args): ... this. Rename builtin_arg
29275 enum values and differentiate between ARG_BUILTIN_MEMORY
29276 and ARG_BUILTIN_NEON_MEMORY.
29277 (arm_expand_neon_builtin_1): Rename to ...
29278 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
29279 values, arm_expand_builtin_args and add bool parameter NEON.
29280 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
29281 (arm_expand_vfp_builtin): Likewise.
29282 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
29284 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29286 PR middle-end/77484
29287 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
29288 * predict.c (tree_estimate_probability_bb): Reverse direction of
29289 polymorphic call predictor.
29291 2017-01-06 David Malcolm <dmalcolm@redhat.com>
29293 * passes.c (execute_one_pass): Split out pass-skipping logic into...
29294 (determine_pass_name_match): ...this new function and...
29295 (should_skip_pass_p): ...this new function.
29297 2017-01-06 Nathan Sidwell <nathan@acm.org>
29299 * ipa-visibility.c (function_and_variable_visibility): Reformat
29300 comments and long lines. Remove extrneous if.
29301 * symtab.c (symtab_node::make_decl_local): Fix code format.
29302 (symtab_node::set_section_for_node): Fix comment typo.
29304 2017-01-06 Martin Liska <mliska@suse.cz>
29307 * lra-constraints.c: Rename invariant to lra_invariant.
29308 * predict.c (set_even_probabilities): Initialize e to NULL.
29310 2017-01-05 Martin Sebor <msebor@redhat.com>
29312 PR tree-optimization/78910
29313 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
29314 (format_integer): Correct off-by-one error in the handling
29315 of precision with negative numbers in signed conversions..
29317 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
29319 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
29321 2017-01-05 Jakub Jelinek <jakub@redhat.com>
29323 PR tree-optimization/71016
29324 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
29325 factor_out_conditional_conversion. Formatting fix.
29326 (factor_out_conditional_conversion): Add cond_stmt argument.
29327 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
29328 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
29331 2017-01-05 David Malcolm <dmalcolm@redhat.com>
29333 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
29334 read-rtl-function.o, and selftest-rtl.o.
29335 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
29336 (selftest::aarch64_test_loading_full_dump): New function.
29337 (selftest::aarch64_run_selftests): New function.
29338 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
29339 selftest::aarch64_run_selftests.
29340 * config/i386/i386.c
29341 (selftest::ix86_test_loading_dump_fragment_1): New function.
29342 (selftest::ix86_test_loading_call_insn): New function.
29343 (selftest::ix86_test_loading_full_dump): New function.
29344 (selftest::ix86_test_loading_unspec): New function.
29345 (selftest::ix86_run_selftests): Call the new functions.
29346 * emit-rtl.c (maybe_set_max_label_num): New function.
29347 * emit-rtl.h (maybe_set_max_label_num): New decl.
29348 * function.c (instantiate_decls): Guard call to
29349 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
29350 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
29352 * gensupport.c (gen_reader::gen_reader): Pass "false"
29353 for new "compact" param of rtx_reader.
29354 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
29355 rather than an empty string for NULL strings.
29356 * read-md.c: Potentially include config.h rather than bconfig.h.
29357 Wrap include of errors.h with #ifdef GENERATOR_FILE.
29358 (have_error): New global, copied from errors.c.
29359 (md_reader::read_name): Rename to...
29360 (md_reader::read_name_1): ...this, adding "out_loc" param,
29361 and converting "missing name or number" to returning false, rather
29363 (md_reader::read_name): Reimplement in terms of read_name_1.
29364 (md_reader::read_name_or_nil): New function.
29365 (md_reader::read_string): Handle "(nil)" by returning NULL.
29366 (md_reader::md_reader): Add new param "compact".
29367 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
29368 (md_reader::read_file): New method.
29369 * read-md.h (md_reader::md_reader): Add new param "compact".
29370 (md_reader::read_file): New method.
29371 (md_reader::is_compact): New accessor.
29372 (md_reader::read_name): Convert return type from void to file_location.
29373 (md_reader::read_name_or_nil): New decl.
29374 (md_reader::read_name_1): New decl.
29375 (md_reader::m_compact): New field.
29376 (noop_reader::noop_reader): Pass "false" for new "compact" param
29378 (rtx_reader::rtx_reader): Add new "compact" param.
29379 (rtx_reader::read_rtx_operand): Make virtual and convert return
29380 type from void to rtx.
29381 (rtx_reader::read_until): New decl.
29382 (rtx_reader::handle_any_trailing_information): New virtual function.
29383 (rtx_reader::postprocess): New virtual function.
29384 (rtx_reader::finalize_string): New virtual function.
29385 (rtx_reader::m_in_call_function_usage): New field.
29386 (rtx_reader::m_reuse_rtx_by_id): New field.
29387 * read-rtl-function.c: New file.
29388 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
29389 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
29390 (selftest::verify_three_block_rtl_cfg): New decl.
29391 * read-rtl-function.h: New file.
29392 * read-rtl.c: Potentially include config.h rather than bconfig.h.
29393 For host, include function.h, memmodel.h, and emit-rtl.h.
29394 (one_time_initialization): New function.
29395 (struct compact_insn_name): New struct.
29396 (compact_insn_names): New array.
29397 (find_code): Handle insn codes in compact dumps.
29398 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
29399 (bind_subst_iter_and_attr): Likewise.
29400 (add_condition_to_string): Likewise.
29401 (add_condition_to_rtx): Likewise.
29402 (apply_attribute_uses): Likewise.
29403 (add_current_iterators): Likewise.
29404 (apply_iterators): Likewise.
29405 (initialize_iterators): Guard usage of apply_subst_iterator with
29406 #ifdef GENERATOR_FILE.
29407 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
29408 (md_reader::read_mapping): Likewise.
29409 (add_define_attr_for_define_subst): Likewise.
29410 (add_define_subst_attr): Likewise.
29411 (read_subst_mapping): Likewise.
29412 (check_code_iterator): Likewise.
29413 (rtx_reader::read_rtx): Likewise. Move one-time initialization
29415 (one_time_initialization): New function.
29416 (rtx_reader::read_until): New method.
29417 (read_flags): New function.
29418 (parse_reg_note_name): New function.
29419 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
29420 Handle reuse_rtx ids.
29421 Wrap iterator lookup within #ifdef GENERATOR_FILE.
29422 Add parsing support for RTL dumps, mirroring the special-cases in
29423 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
29424 values, and calling handle_any_trailing_information.
29425 (rtx_reader::read_rtx_operand): Convert return type from void
29426 to rtx, returning return_rtx. Handle case 'e'. Call
29427 finalize_string on XSTR and XTMPL fields.
29428 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
29429 "(nil)" values were omitted. Call the postprocess vfunc on the
29431 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
29432 class ctor. Initialize m_in_call_function_usage. Call
29433 one_time_initialization.
29434 * rtl-tests.c (selftest::test_uncond_jump): Call
29435 set_new_first_and_last_insn.
29436 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
29437 * selftest-rtl.c: New file.
29438 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
29439 (selftest::get_insn_by_uid): New decl.
29440 * selftest-run-tests.c (selftest::run_tests): Call
29441 read_rtl_function_c_tests.
29442 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
29443 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
29446 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
29448 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
29449 operands in a special way. Assert that pos+len <= mode precision.
29451 2017-01-05 Jakub Jelinek <jakub@redhat.com>
29453 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
29454 3 argument Alias with unlimited for the negative form.
29455 (fno-vect-cost-model): Removed.
29457 2017-01-05 Martin Liska <mliska@suse.cz>
29459 * hsa-gen.c (gen_hsa_divmod): New function.
29460 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
29462 2017-01-05 Martin Liska <mliska@suse.cz>
29465 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
29468 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29470 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
29471 small constant length operands.
29473 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29475 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
29476 between loop iterations.
29478 2017-01-05 Martin Liska <mliska@suse.cz>
29481 * gimplify.c (gimplify_decl_expr): Compare to
29482 asan_poisoned_variables instread of checking flags.
29483 (gimplify_target_expr): Likewise.
29484 (gimplify_expr): Likewise.
29485 (gimplify_function_tree): Conditionally initialize
29486 asan_poisoned_variables.
29488 2017-01-04 Jeff Law <law@redhat.com>
29490 PR tree-optimizatin/78812
29491 * rtl.h (contains_mem_rtx_p): Prototype.
29492 * ifcvt.c (containts_mem_rtx_p): Move from here to...
29493 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
29494 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
29495 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
29496 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
29498 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29500 * input.c (assert_char_at_range): Default-initialize actual_range.
29502 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29504 * df-scan.c (df_ref_create_structure): Make regno unsigned,
29505 to match the caller.
29507 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29509 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
29510 insns after final jump in test to emit dummy move.
29512 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29514 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
29515 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
29517 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29519 * multiple_target.c (create_dispatcher_calls): Init e_next.
29520 * tree-ssa-loop-split.c (split_loop): Init border.
29521 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
29524 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
29529 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
29530 (altivec_register_operand): Do not return true if the operand
29531 contains a SUBREG mixing SImode and SFmode.
29532 (vsx_register_operand): Likewise.
29533 (vsx_reg_sfsubreg_ok): New predicate.
29534 (vfloat_operand): Do not return true if the operand contains a
29535 SUBREG mixing SImode and SFmode.
29536 (vint_operand): Likewise.
29537 (vlogical_operand): Likewise.
29538 (gpc_reg_operand): Likewise.
29539 (int_reg_operand): Likewise.
29540 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
29541 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
29542 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
29544 (rs6000_emit_move_si_sf_subreg): New helper function.
29545 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
29546 fixup SUBREGs involving SImode and SFmode.
29547 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
29548 numbers for the new peephole2 optimization.
29549 (peephole2 for SFmode unions): New peephole2 to optimize cases in
29550 the GLIBC math library that do AND/IOR/XOR operations on single
29551 precision floating point.
29552 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
29553 target macros to say whether we need to avoid SUBREGs mixing
29555 (TARGET_ALLOW_SF_SUBREG): Likewise.
29556 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
29557 (UNSPEC_SI_FROM_SF): Likewise.
29558 (iorxor): Change spacing.
29559 (and_ior_xor): New iterator for AND, IOR, and XOR.
29560 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
29561 (movdi_from_sf_zero_ext): Likewise.
29562 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
29563 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
29564 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
29565 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
29566 (fms<mode>4): Likewise.
29567 (fnma<mode>4): Likewise.
29568 (fnms<mode>4): Likewise.
29569 (nfma<mode>4): Likewise.
29570 (nfms<mode>4): Likewise.
29572 2017-01-04 Marek Polacek <polacek@redhat.com>
29575 * doc/invoke.texi: Document -Wpointer-compare.
29577 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29579 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
29582 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
29583 descriptions for -gdwarf-5 and emit them as uleb128 instead of
29586 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
29589 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
29590 documentation of the powerpc_popcntb_ok attribute.
29591 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
29592 code to issue warning messages if a requested CPU configuration is
29593 not supported by the binary (assembler and loader) toolchain.
29594 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
29595 made to define a built-in function that has been disabled.
29596 (paired_init_builtins): Add assertion to prevent ICE if attempt is
29597 made to define a built-in function that has been disabled.
29598 (altivec_init_builtins): Add comment explaining why definition
29599 of the DST built-in functions is not preceded by an assertion
29600 check. Add assertions to prevent ICE if attempts are made to
29601 define an altivec predicate or an abs* built-in function that has
29603 (htm_init_builtins): Add comment explaining why definition of the
29604 htm built-in functions is not preceded by an assertion check.
29606 2017-01-04 Jeff Law <law@redhat.com>
29608 PR tree-optimizatin/67955
29609 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
29610 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
29611 the points-to solution does not include pt_null. Use DECL_PT_UID
29614 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
29616 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
29617 Use gen_int_mode instead of gen_lopwart for const_int operands.
29619 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29621 PR tree-optimization/71563
29622 * match.pd: Simplify X << Y into X if Y is known to be 0 or
29623 out of range value - has low bits known to be zero.
29625 2017-01-04 Alan Modra <amodra@gmail.com>
29627 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
29628 * configure: Regenerate.
29629 * config.in: Regenerate.
29631 2017-01-04 Jakub Jelinek <jakub@redhat.com>
29634 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
29635 a substring of the message, but strcmp with the whole message. Ifdef
29636 ENABLE_NLS, translate the message first using dgettext.
29638 2017-01-03 Jeff Law <law@redhat.com>
29640 PR tree-optimizatin/78856
29641 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
29642 (mark_threaded_blocks): Remove code to truncate thread paths that
29643 cross multiple loop headers. Instead invalidate the cached loop
29644 iteration information and handle case of a thread path walking
29645 into an irreducible region.
29647 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
29650 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
29651 assertions. Add support for doing the signbit if the IEEE 128-bit
29652 floating point value is in a GPR.
29653 * config/rs6000/rs6000.md (Fsignbit): Delete.
29654 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
29655 Update the length attribute if the value is in a GPR.
29656 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
29657 the sign or zero extension instruction, since the value is always 0/1.
29658 (signbit<mode>2_dm2): Delete using <Fsignbit>.
29661 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
29662 extracting SImode to a GPR register so that we can generate a
29663 store, limit the vector to be in a traditional Altivec register
29664 for the vextuwrx instruction.
29666 2017-01-03 Ian Lance Taylor <iant@google.com>
29668 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
29670 2017-01-03 Martin Sebor <msebor@redhat.com>
29672 PR tree-optimization/78696
29673 * gimple-ssa-sprintf.c (format_floating): Correct handling of
29674 precision. Use MPFR for %f for greater fidelity. Correct handling
29676 (pass_sprintf_length::compute_format_length): Set width and precision
29677 specified by asrerisk to void_node for vararg functions.
29678 (try_substitute_return_value): Adjust dump output.
29680 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
29682 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
29684 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
29686 * doc/invoke.texi (SPARC options): Document -mlra as the default.
29687 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
29688 -mlra/-mno-lra was passed to the compiler.
29690 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
29692 PR rtl-optimization/65618
29693 * emit-rtl.c (try_split): Move initialization of "before" and
29694 "after" to just before the call to emit_insn_after_setloc.
29696 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
29698 * doc/md.texi (Standard Names): Remove reference to Java frontend.
29700 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
29702 * dwarf2out.c (gen_enumeration_type_die): When
29703 -gno-strict-dwarf, add a DW_AT_encoding attribute.
29705 2017-01-03 Jakub Jelinek <jakub@redhat.com>
29707 PR tree-optimization/78965
29708 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
29709 Change first argument from const call_info & to call_info &. For %n
29710 set info.nowrite to false.
29712 PR middle-end/78901
29713 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
29714 possibly throwing calls.
29716 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
29717 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
29718 and fns handling, rather than in a separate case SSA_NAME.
29720 2017-01-02 Jeff Law <law@redhat.com>
29722 * config/darwin-driver.c (darwin_driver_init): Const-correctness
29723 fixes for first_period and second_period variables.
29725 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
29728 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
29729 (*insvqi_1): New insn pattern.
29730 (*insvqi_1_mem_rex64): Ditto.
29731 (*insvqi_2): Ditto.
29732 (*insvqi_3): Rename from *insvqi.
29734 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
29736 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
29738 * doc/cfg.texi (Edges): Remove reference to Java.
29739 (Maintaining the CFG): Ditto.
29741 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29743 PR middle-end/77674
29744 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
29745 transparent aliases.
29747 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29749 PR middle-end/77484
29750 * predict.def (PRED_CALL): Update hitrate.
29751 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
29752 * predict.c (tree_estimate_probability_bb): Split CALL predictor
29753 into direct/indirect/polymorphic variants.
29755 2017-01-01 Jakub Jelinek <jakub@redhat.com>
29757 Update copyright years.
29759 * gcc.c (process_command): Update copyright notice dates.
29760 * gcov-dump.c (print_version): Ditto.
29761 * gcov.c (print_version): Ditto.
29762 * gcov-tool.c (print_version): Ditto.
29763 * gengtype.c (create_file): Ditto.
29764 * doc/cpp.texi: Bump @copying's copyright year.
29765 * doc/cppinternals.texi: Ditto.
29766 * doc/gcc.texi: Ditto.
29767 * doc/gccint.texi: Ditto.
29768 * doc/gcov.texi: Ditto.
29769 * doc/install.texi: Ditto.
29770 * doc/invoke.texi: Ditto.
29772 Copyright (C) 2017 Free Software Foundation, Inc.
29774 Copying and distribution of this file, with or without modification,
29775 are permitted in any medium without royalty provided the copyright
29776 notice and this notice are preserved.