1 2017-11-15 Jakub Jelinek <jakub@redhat.com>
3 PR tree-optimization/82977
4 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
5 constructed temporary to strlen_to_stridx.put.
7 2017-11-15 Martin Liska <mliska@suse.cz>
9 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
10 * configure: Regenerate.
12 2017-11-15 Martin Liska <mliska@suse.cz>
15 * config/sh/sh-mem.cc: Use proper probability for
18 2017-11-14 Jeff Law <law@redhat.com>
20 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
21 the red zone for stack_clash_protection_final_dynamic_probe targets
22 when the total dynamic stack size is zero bytes.
24 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
27 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
29 * dumpfile.h (TDF_COMPARE_DEBUG): New.
30 * final.c (rest_of_clean_state): Set it for the
32 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
33 class when TDF_COMPARE_DEBUG is set.
35 * dwarf2out.c (gen_producer_string): Discard
38 2017-11-15 Joseph Myers <joseph@codesourcery.com>
41 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
42 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
43 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
44 (__TGMATH_REAL_2_3): Remove macros.
45 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
46 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
48 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
50 (remquo): Define using __TGMATH_REAL_3.
52 2017-11-14 Jeff Law <law@redhat.com>
54 * vr-values.c: New file with contents extracted from tree-vrp.c.
55 * Makefile.in (OBJS): Add vr-values.o
56 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
57 (set_value_range, set_and_canonicalize_value_range): Likewise.
58 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
59 (value_range_constant_singleton, symbolic_range_p): Likewise.
60 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
61 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
62 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
63 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
64 (find_case_label_range, find_case_label_index): Likewise.
65 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
66 (range_int_cst_singleton_p, value_inside_range): Likewise.
67 (get_single_symbol): Likewise.
68 (switch_update): Move structure definition here.
69 (to_remove_edges, to_update_switch_stmts): Provide externs.
70 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
71 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
72 (vrp_val_is_min, set_value_range): Likewise.
73 (set_and_canonicalize_value_range, copy_value_range): Likewise.
74 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
75 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
76 (range_is_nonnull, range_int_cst_p): Likewwise.
77 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
78 (get_single_symbol, operand_less_p): Likewise
79 (compare_values_warnv, compare_values): Likewise.
80 (value_inside_range, value_range_constant_singleton): Likewise.
81 (zero_nonzero_bitgs_from_vr): Likewise.
82 (extract_range_from_binary_expr_1): Likewise.
83 (overflow_comparison_p): Likewise.
84 (to_remove_edges, to_update_switch_stmts): Likewise.
85 (find_case_label-index, find_case_label_range): Likewise.
86 (switch_update, set_value_range_to_nonnegative): Remove.
87 (set_value_range_to_truthvalue): Likewise.
88 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
89 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
90 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
91 (find_case_label_ranges, test_for_singularity): Likewise.
92 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
93 (x_vr_values): Move to its remaining use site.
95 2017-11-10 Jeff Law <law@redhat.com>
97 * vr-values.h (VR_INITIALIZER): Move #define here.
98 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
99 * Makefile.in (OBJS): Add tree-evrp.o
100 * tree-vrp.h (assert_info): Move structure definition here.
101 (set_value_range_to_varying): Prototype.
102 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
103 (infer_value_range, register_edge_assert_for): Likewise.
104 (stmt_interesting_for_vrp): Likewise.
105 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
106 (set_value_range_to_varying): No longer static.
107 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
108 (infer_value_range, register_edge_assert_for): Likewise.
110 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
112 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
113 generate the XXBRD instruction.
115 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
116 (rs6000_builtin_type_compatible): Treat _Float128 and long double
117 as being compatible if -mabi=ieeelongdouble.
118 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
119 to setup float128 built-ins with hardware support.
120 (BU_FLOAT128_HW_2): Likewise.
121 (BU_FLOAT128_HW_3): Likewise.
122 (BU_FLOAT128_HW_VSX_1): Likewise.
123 (BU_FLOAT128_HW_VSX_2): Likewise.
124 (scalar_extract_expq): Change float128 built-in functions to
125 accommodate having both KFmode and TFmode functions. Use the
126 KFmode variant as the default.
127 (scalar_extract_sigq): Likewise.
128 (scalar_test_neg_qp): Likewise.
129 (scalar_insert_exp_q): Likewise.
130 (scalar_insert_exp_qp): Likewise.
131 (scalar_test_data_class_qp): Likewise.
132 (sqrtf128_round_to_odd): Delete processing the round to odd
133 built-in functions as special built-in functions, and define them
134 as float128 built-ins. Use the KFmode variant as the default.
135 (truncf128_round_to_odd): Likewise.
136 (addf128_round_to_odd): Likewise.
137 (subf128_round_to_odd): Likewise.
138 (mulf128_round_to_odd): Likewise.
139 (divf128_round_to_odd): Likewise.
140 (fmaf128_round_to_odd): Likewise.
141 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
142 support for KFmode and TFmode xststdcqp calls.
143 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
144 point, switch the built-in handlers for the get/set float128
145 exponent, get float128 mantissa, float128 test built-ins, and the
146 float128 round to odd built-in functions. Eliminate creating the
147 float128 round to odd built-in functions as special built-ins.
148 (rs6000_init_builtins): Eliminate special creation of the float128
149 round to odd built-in functions.
150 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
151 function insns to support both TFmode and KFmode variants.
152 (xsxsigqp_<mode>): Likewise.
153 (xsiexpqpf_<mode>): Likewise.
154 (xsiexpqp_<mode>): Likewise.
155 (xststdcqp_<mode>): Likewise.
156 (xststdcnegqp_<mode>): Likewise.
157 (xststdcqp_<mode>): Likewise.
159 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
161 * tree-ssa-threadupdate.c (compute_path_counts): Remove
162 unused path_in_freq_ptr parameter.
163 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
165 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
167 * ipa-inline.c (edge_badness): Dump sreal frequency.
168 (compute_inlined_call_time): Match natural implementaiton ...
169 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
170 forgotten division by CGRAPH_FREQ_BASE.
172 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
174 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
175 Solaris 11. Update comment.
176 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
178 * config/sol2.h (STARTFILE_SPEC): Likewise.
179 * configure: Regenerate.
181 2017-11-14 Carl Love <cel@us.ibm.com>
183 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
184 le_ and be_ prefixes to swap* variables. Remove
185 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
188 2017-11-14 Jason Merrill <jason@redhat.com>
190 Support GTY((cache)) on hash_map.
191 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
192 (ggc_cache_remove): Override it instead of ggc_mx.
193 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
194 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
195 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
196 (simple_cache_map_traits): Override maybe_mx.
197 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
198 (hash_map): Friend gt_cleare_cache.
199 (gt_cleare_cache): New.
200 * tree.h (tree_cache_traits): New hash_map traits class.
201 (tree_cache_map): New typedef.
203 2017-11-14 Richard Biener <rguenther@suse.de>
205 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
206 paramter and handling.
207 (cleanup_control_flow_bb): Likewise.
208 (cleanup_control_flow_pre): New helper performing a DFS walk
209 to call cleanup_control_flow_bb in PRE order.
210 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
211 via cleanup_control_flow_pre.
213 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
215 * config/aarch64/aarch64-simd.md
216 (aarch64_simd_bsl<mode>_internal): Remove DImode.
217 (*aarch64_simd_bsl<mode>_alt): Likewise.
218 (aarch64_simd_bsldi_internal): New.
219 (aarch64_simd_bsldi_alt): Likewise.
221 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
223 * tracer.c (better_p): Do not compare frequencies.
224 * reg-stack.c (better_edge): Likewise.
225 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
228 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
230 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
231 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
232 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
233 * lto-streamer-in.c (input_function): Use update_max_bb_count.
234 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
235 * predict.c (maybe_hot_frequency_p): Inline to ...
236 (maybe_hot_count_p): ... here; rewrite to counts.
237 (counts_to_freqs): Rename to ...
238 (update_max_bb_count): ... this one.
239 (expensive_function_p): Use counts.
240 (estimate_bb_frequencies): Update.
241 (rebuild_frequencies): Update.
242 * predict.h (counts_to_freqs): Rename to ...
243 (update_max_bb_count): ... this one.
244 * profile.c (compute_branch_probabilities): Add debug info
245 * tree-inline.c (expand_call_inline): Update debug info.
246 (optimize_inline_calls): Use update_max_bb_count..
247 (tree_function_versioning): Use update_max_bb_count..
248 * value-prof.c (gimple_value_profile_transformations):
249 Do not use update_max_bb_count.
251 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
253 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
254 always use frequencies.
256 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
258 * bb-reorder.c: Remove frequencies from comments.
259 (better_edge_p): Use profile counts.
260 (find_traces): Dump profile counts.
261 (rotate_loop): Use profile counts.
262 (find_traces_1_round): Likewise.
263 (connect_better_edge_p): Use counts instead of probabilities for
265 (copy_bb_p): Drop early check for non-0 frequency.
266 (sanitize_hot_paths): Update comments.
268 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
270 * ipa-split.c (struct split_point): Add count.
271 (consider_split): Do not compute incoming frequency; compute incoming
272 count and store it to split_point.
273 (split_function): Set count of the call to split part correctly.
275 2017-11-13 Carl Love <cel@us.ibm.com>
277 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
279 2017-11-13 Tom Tromey <tom@tromey.com>
281 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
283 2017-11-13 Carl Love <cel@us.ibm.com>
285 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
286 Add support for builtins:
287 unsigned int vec_first_{,miss}_match_{,or_eos}index,
288 vector {un,}signed {char,int,short},
289 vector {un,}signed {char,int,short}) arguments.
290 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
291 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
292 Add BU_P9V_AV_2 expansions for the builtins.
293 * config/rs6000/altivec.h (vec_first_match_index,
294 vec_first_mismatch_index, vec_first_match_or_eos_index,
295 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
296 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
297 new extern declaration.
298 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
299 * config/rs6000/vsx.md (first_match_index_<mode>,
300 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
301 first_mismatch_or_eos_index_<mode>): Add define expand.
302 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
303 * doc/extend.texi: Update the built-in documenation file for the new
306 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
308 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
309 into the min/max operations for _Float<N> and _Float<N>X types.
311 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
314 * dwarf2out.c (do_eh_frame): New static variable.
315 (dwarf2out_begin_prologue): Set it.
316 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
318 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
320 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
322 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
324 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
327 * debug.h (dwarf2out_do_eh_frame): Declare.
328 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
329 (dwarf2out_do_frame): Use it.
330 (dwarf2out_do_cfi_asm): Likewise.
331 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
332 (dwarf2out_assembly_start): Likewise.
333 (dwarf2out_begin_prologue): Fix comment.
334 * toplev.c (compile_file): Always call dwarf2out_frame_finish
335 if the target needs either debug or unwind DWARF2 info.
336 * lto-opts.c (lto_write_options): Do not save -fexceptions,
337 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
338 -ftrapping-math, -ftrapv and -fwrapv.
340 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
342 * cgraph.c (cgraph_edge::sreal_frequency): New function.
343 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
344 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
345 (estimate_edge_size_and_time): Likewise.
346 (ipa_merge_fn_summary_after_inlining): Likewise.
347 * ipa-inline.c (cgraph_freq_base_rec): Remove.
348 (compute_uninlined_call_time): Use sreal_frequency.
349 (compute_inlined_call_time): Likewise.
350 (ipa_inline): Do not initialize cgraph_freq_base_rec.
351 * profile-count.c: Include sreal.h.
352 (profile_count::to_sreal_scale): New.
353 * profile-count.h: Forward declare sreal.
354 (profile_count::to_sreal_scale): Declare.
356 2017-11-13 Nathan Sidwell <nathan@acm.org>
358 * diagnostic.c (maybe_line_and_column): New.
359 (diagnostic_get_location_text): Use it.
360 (diagnostic_report_current_module): Likewise.
361 (test_diagnostic_get_location_text): Add tests.
363 2017-11-13 Luis Machado <luis.machado@linaro.org>
365 * doc/md.texi (Specifying processor pipeline description): Fix
366 incorrect latency for the div instruction example.
368 2017-11-13 Jakub Jelinek <jakub@redhat.com>
370 PR tree-optimization/78821
371 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
372 that bit_not_p is the same.
373 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
374 (split_group): Count precisely bit_not_p bits in each statement.
375 (invert_op): New function.
376 (imm_store_chain_info::output_merged_store): Use invert_op to
377 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
378 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
380 2017-11-13 Martin Liska <mliska@suse.cz>
382 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
383 (struct source_info): Likewise.
384 (add_branch_counts): Likewise.
385 (add_line_counts): Likewise.
386 (function_summary): Likewise.
387 (output_intermediate_line): Likewise.
388 (generate_results): Likewise.
390 2017-11-13 Martin Liska <mliska@suse.cz>
392 * gcov.c (struct block_info): Remove typedef for block_t.
393 (struct line_info): Likewise.
394 (line_info::has_block): Likewise.
395 (EXIT_BLOCK): Likewise.
398 (get_cycles_count): Likewise.
399 (process_file): Likewise.
400 (read_graph_file): Likewise.
401 (solve_flow_graph): Likewise.
402 (find_exception_blocks): Likewise.
403 (add_line_counts): Likewise.
404 (accumulate_line_info): Likewise.
405 (output_line_details): Likewise.
407 2017-11-13 Martin Liska <mliska@suse.cz>
409 * gcov.c (struct arc_info): Remove typedef for arc_t.
410 (struct line_info): Likewise.
411 (add_branch_counts): Likewise.
412 (output_branch_count): Likewise.
413 (function_info::~function_info): Likewise.
415 (output_intermediate_line): Likewise.
416 (read_graph_file): Likewise.
417 (solve_flow_graph): Likewise.
418 (find_exception_blocks): Likewise.
419 (add_line_counts): Likewise.
420 (accumulate_line_info): Likewise.
421 (output_line_details): Likewise.
422 (output_function_details): Likewise.
424 2017-11-13 Martin Liska <mliska@suse.cz>
426 * gcov.c (struct function_info): Remove typedef for function_t.
427 (struct source_info): Likewise.
428 (source_info::get_functions_at_location): Likewise.
429 (solve_flow_graph): Likewise.
430 (find_exception_blocks): Likewise.
431 (add_line_counts): Likewise.
432 (output_intermediate_file): Likewise.
433 (process_file): Likewise.
434 (generate_results): Likewise.
435 (release_structures): Likewise.
436 (read_graph_file): Likewise.
437 (read_count_file): Likewise.
438 (accumulate_line_counts): Likewise.
439 (output_lines): Likewise.
441 2017-11-13 Martin Liska <mliska@suse.cz>
443 * gcov.c (function_info::function_info): Remove num_counts
444 and add vector<gcov_type>.
445 (function_info::~function_info): Use the vector.
446 (process_file): Likewise.
447 (read_graph_file): Likewise.
448 (read_count_file): Likewise.
449 (solve_flow_graph): Likewise.
451 2017-11-13 Martin Liska <mliska@suse.cz>
453 * gcov.c (function_info::is_artificial): New function.
454 (process_file): Erase all artificial early.
455 (generate_results): Skip as all artificial are already
458 2017-11-13 Martin Liska <mliska@suse.cz>
460 * gcov.c (read_graph_file): Store to global vector of functions.
461 (read_count_file): Iterate the vector.
462 (process_file): Likewise.
463 (generate_results): Likewise.
464 (release_structures): Likewise.
466 2017-11-13 Jakub Jelinek <jakub@redhat.com>
468 PR tree-optimization/82954
469 * gimple-ssa-store-merging.c
470 (imm_store_chain_info::coalesce_immediate_stores): If
471 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
473 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
475 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
476 Upddate call to ENDIAN_LANE_N.
477 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
478 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
479 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
480 and use aarch64_endian_lane_rtx.
481 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
483 2017-11-12 Tom de Vries <tom@codesourcery.com>
485 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
487 2017-11-12 Tom de Vries <tom@codesourcery.com>
489 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
491 2017-11-12 Tom de Vries <tom@codesourcery.com>
493 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
495 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
496 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
497 * defaults.h (ASM_OUTPUT_LABELREF): Same.
499 2017-11-11 Martin Sebor <msebor@redhat.com>
502 * doc/extend.texi (attribute nonstring): Remove spurious argument.
505 * prefic.c (translate_name): Replace strncpy with memcpy to
506 avoid -Wstringop-truncation.
508 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
510 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
512 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
514 * predict.c (maybe_hot_frequency_p): Do not use cfun.
516 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
518 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
521 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
523 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
524 enable generating XXBRH if the value is in a vector register.
525 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
526 value is in a vector register.
527 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
528 register to register bswap64's instead of doing the GPR sequence
529 used on previous machines.
530 (bswapdi2_xxbrd): New insn.
531 (bswapdi2_reg): Disallow on ISA 3.0.
532 (register to register bswap64 splitter): Do not split the insn on
533 ISA 3.0 systems that use XXBRD.
535 2017-11-10 Martin Sebor <msebor@redhat.com>
538 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
540 (find_subframework_file): Same.
542 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
544 * auto-profile.c (afdo_indirect_call): Drop frequency.
545 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
546 (cgraph_node::create_edge): Drop frequency argument.
547 (cgraph_node::create_indirect_edge): Drop frequency argument.
548 (cgraph_edge::make_speculative): Drop frequency arguments.
549 (cgraph_edge::resolve_speculation): Do not update frequencies
550 (cgraph_edge::dump_edge_flags): Do not dump frequency.
551 (cgraph_node::dump): Check consistency in IPA mode.
552 (cgraph_edge::maybe_hot_p): Use IPA counter.
553 (cgraph_edge::verify_count_and_frequency): Rename to ...
554 (cgraph_edge::verify_count): ... this one; drop frequency checking.
555 (cgraph_node::verify_node): Update.
556 * cgraph.h (struct cgraph_edge): Drop frequency.
557 (cgraph_edge::frequency): New function.
558 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
560 (cgraph_edge::rebuild_edges): Likewise.
561 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
562 (duplicate_thunk_for_node): Do not pass frequency.
563 (cgraph_node::create_clone): Scale only counts.
564 (cgraph_node::create_virtual_clone): Do not pass frequency.
565 (cgraph_node::create_edge_including_clones): Do not pass frequency.
566 (cgraph_node::create_version_clone): Do not pass frequency.
567 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
568 (cgraph_node::expand_thunk): Do not pass frequency.
569 (cgraph_node::create_wrapper): Do not pass frequency.
570 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
572 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
573 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
574 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
575 (gather_caller_stats): Use frequency function.
576 (ipcp_cloning_candidate_p): Use frequency function.
577 (ipcp_propagate_stage): Use frequency function.
578 (get_info_about_necessary_edges): Use frequency function.
579 (update_profiling_info): Update only IPA profile.
580 (update_specialized_profile): Use frequency functoin.
581 (perhaps_add_new_callers): Update only IPA profile.
582 * ipa-devirt.c (ipa_devirt): Use IPA profile.
583 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
584 (dump_ipa_call_summary): Use frequency function.
585 (estimate_edge_size_and_time): Use frequency function.
586 (ipa_merge_fn_summary_after_inlining): Use frequency function.
587 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
588 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
589 (update_noncloned_counts): ... ths one; scale counts only.
590 (clone_inlined_nodes): Do not scale frequency.
591 (inline_call): Do not pass frequency.
592 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
593 (compute_inlined_call_time): Use IPA profile.
594 (want_inline_small_function_p): Use IPA profile.
595 (want_inline_self_recursive_call_p): Use IPA profile.
596 (edge_badness): Use IPA profile.
597 (lookup_recursive_calls): Use IPA profile.
598 (recursive_inlining): Do not pass frequency.
599 (resolve_noninline_speculation): Do not update frequency.
600 (inline_small_functions): Collect max of IPA profile.
601 (dump_overall_stats): Dump IPA porfile.
602 (dump_inline_stats): Dump IPA porfile.
603 (ipa_inline): Collect IPA stats.
604 * ipa-inline.h (clone_inlined_nodes): Update prototype.
605 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
606 (ipa_propagate_frequency): Use frequency function.
607 (ipa_profile): Cleanup.
608 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
609 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
610 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
611 (input_node): Do not stream frequency.
612 (input_edge): Do not stream frequency.
613 (merge_profile_summaries): Scale only IPA profiles.
614 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
615 * predict.c (drop_profile): Do not recompute frequency.
616 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
617 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
618 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
619 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
620 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
621 * tree-inline.c (copy_bb): Do not scale frequency.
622 (expand_call_inline): Do not scale frequency.
623 (tree_function_versioning): Do not scale frequency.
624 * ubsan.c (ubsan_create_edge): Do not pass frequency.
626 2017-11-10 Julia Koval <julia.koval@intel.com>
628 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8,
629 _mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8,
630 _mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8,
631 _mm256_maskz_gf2p8affine_epi64_epi8,
632 _mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8
633 _mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
634 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi,
635 __builtin_ia32_vgf2p8affineqb_v32qi,
636 __builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
637 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
639 2017-11-10 Tamar Christina <tamar.christina@arm.com>
643 (arm_option_override): Refactor.
644 (arm_option_reconfigure_globals): New.
645 (arm_options_perform_arch_sanity_checks): New.
646 * config/arm/arm-protos.h (arm_option_reconfigure_globals): New prototype.
647 (arm_options_perform_arch_sanity_checks): Likewise
649 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
651 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
653 * rs6000/rs6000.md ("type" attr): Add qmul.
654 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
655 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
656 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
658 2017-11-10 Martin Sebor <msebor@redhat.com>
661 * builtins.c (compute_objsize): Handle arrays that
662 compute_builtin_object_size likes to fail for. Make extern.
663 * builtins.h (compute_objsize): Declare.
664 (check_strncpy_sizes): New function.
665 (expand_builtin_strncpy): Call check_strncpy_sizes.
666 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
667 -Wstringop-truncation.
668 (gimple_fold_builtin_strncat): Same.
669 * gimple.c (gimple_build_call_from_tree): Set call location.
670 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
671 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
672 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
673 (handle_builtin_strlen): Use strlen_to_stridx.
674 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
676 Use strlen_to_stridx.
677 (pass_strlen::execute): Release strlen_to_stridx.
678 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
679 (-Wstringop-truncation): Document new option.
681 2017-11-10 Martin Liska <mliska@suse.cz>
683 PR gcov-profile/82702
684 * gcov.c (main): Handle intermediate files in a different
686 (get_gcov_intermediate_filename): New function.
687 (output_gcov_file): Remove support of intermediate files.
688 (generate_results): Allocate intermediate file.
689 (release_structures): Clean-up properly fn_end.
690 (output_intermediate_file): Start iterating with line 1.
692 2017-11-10 Jakub Jelinek <jakub@redhat.com>
694 PR tree-optimization/82929
695 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
696 ops_swapped_p non-static data member.
697 (store_immediate_info::store_immediate_info): Clear it.
698 (imm_store_chain_info::coalesce_immediate_stores): If swapping
699 ops set ops_swapped_p.
700 (count_multiple_uses): Handle ops_swapped_p.
702 2017-11-10 Martin Liska <mliska@suse.cz>
704 * coverage.c (coverage_init): Stream information about
705 support of has_unexecuted_blocks.
706 * doc/gcov.texi: Document that.
707 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
708 * gcov.c (read_graph_file): Likewise.
709 (output_line_beginning): Fix a small issue with
712 2017-11-10 Bin Cheng <bin.cheng@arm.com>
714 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
715 reference of trivial component.
717 2017-11-10 Jakub Jelinek <jakub@redhat.com>
720 * gimple-ssa-store-merging.c
721 (pass_store_merging::terminate_all_aliasing_chains): For
722 gimple_store_p stmts also call refs_output_dependent_p.
724 PR rtl-optimization/82913
725 * compare-elim.c (try_merge_compare): Punt if def_insn is not
728 2017-11-09 Jeff Law <law@redhat.com>
730 * vr-values.h: New file with vr_values class.
731 * tree-vrp.c: Include vr-values.h
732 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
733 data objects into the vr_values class.
734 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
735 (get_value_range): Make it a member function within vr_values class.
736 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
737 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
738 (op_with_constant_singleton_value_range): Likewise.
739 (extract_range_for_var_from_comparison_expr): Likewise.
740 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
741 (extract_range_from_binary_expr): Likewise.
742 (extract_range_from_unary_expr): Likewise.
743 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
744 (check_for_binary_op_overflow, extract_range_basic): Likewise.
745 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
746 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
747 (compare_name_with_value, compare_names): Likewise.
748 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
749 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
750 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
751 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
752 (extract_range_from_phi_node): Likewise.
753 (simplify_truth_ops_using_ranges): Likewise.
754 (simplify_div_or_mod_using_ranges): Likewise.
755 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
756 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
757 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
758 (simplify_float_conversion_using_ranges): Likewise.
759 (simplify_internal_call_using_ranges): Likewise.
760 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
761 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
762 poitner via x_vr_values for calls into gimple folder.
763 (vrp_initialize_lattice): Make this the vr_values ctor.
764 (vrp_free_lattice): Make this the vr_values dtor.
765 (set_vr_value): New function.
766 (class vrp_prop): Add vr_values data member. Add various member
767 functions as well as member functions that delegate to vr_values.
768 (check_array_ref): Make a member function within vrp_prop class.
769 (search_for_addr_array, vrp_initialize): Likewise.
770 (vrp_finalize): Likewise. Revamp to avoid direct access to
771 vr_value, values_propagated, etc.
772 (check_array_bounds): Extract vrp_prop class instance pointer from
773 walk info structure. Use it to call member functions.
774 (check_all_array_refs): Make a member function within vrp_prop class.
775 Smuggle class instance pointer via walk info structure.
776 (x_vr_values): New local static.
777 (vrp_valueize): Use x_vr_values to get class instance.
778 (vr_valueize_1): Likewise.
779 (class vrp_folder): Add vr_values data member. Add various member
780 functions as well as member functions that delegate to vr_values.
781 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
782 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
783 class instance from vr_values. Use it to call member functions.
784 (vrp_dom_walker): Add vr_values data member.
785 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
786 instance via x_vr_values.
787 (identify_jump_threads): Accept vr_values as argument. Store
788 it into the walker structure.
789 (evrp_dom_walker): Add vr_values class data member. Add various
791 (evrp_dom_walker::try_find_new_range): Use vr_values data
792 member to access the memory allocator.
793 (evrp_dom_walker::before_dom_children): Store vr_values class
794 instance into the vrp_folder class.
795 (evrp_dom_walker::push_value_range): Rework to avoid direct
796 access to num_vr_values and vr_value.
797 (evrp_dom_walker::pop_value_range): Likewise.
798 (execute_early_vrp): Remove call to vrp_initialize_lattice.
799 Use vr_values to get to dump_all_value_ranges member function.
800 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
801 and simplify_cond_using_ranges_2 via vrp_prop class instance.
802 Pass vr_values class instance down to identify_jump_threads.
803 Remove call to vrp_free_lattice.
804 (debug_all_value_ranges): Remove.
806 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
807 (vrp_folder): Likewise.
809 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
810 Get it from the existing bitmap instead.
811 (vrp_intersect_ranges_1): Likewise.
813 2017-11-09 Jakub Jelinek <jakub@redhat.com>
815 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
817 (store_immediate_info::store_immediate_info): Add bitnotp argument,
819 (imm_store_chain_info::coalesce_immediate_stores): Break group
820 if bit_not_p is different.
821 (count_multiple_uses, split_group,
822 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
823 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
824 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
825 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
826 to store_immediate_info ctor.
828 2017-11-09 Jim Wilson <jimw@sifive.com>
830 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
831 (scan_prog_file): Likewise.
833 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
835 * bb-reorder.c (max_entry_frequency): Remove.
836 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
837 connect_traces, push_to_next_round_p): Remove prototypes.
838 (find_traces_1_round): Use counts only.
839 (push_to_next_round_p): Likewise.
840 (find_traces): Likewise.
841 (rotate_loop): Likewise.
842 (find_traces_1_round): Likewise.
843 (connect_traces): Likewise.
844 (edge_order): Likewise.
846 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
848 * config/arm/arm.c (output_return_instruction): Add comments to
849 indicate requirement for cmse_nonsecure_entry return to account
850 for the size of clearing instruction output here.
851 (thumb_exit): Likewise.
852 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
853 return in hardfloat mode.
855 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
857 * config/rs6000/rs6000.c (machine_function): Add a bool,
858 "toc_is_wrapped_separately".
859 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
860 if it wasn't explicitly set or unset, we are optimizing for speed, and
861 doing separate shrink-wrapping.
862 (rs6000_get_separate_components): Enable the TOC component if
863 saving the TOC register in the prologue.
864 (rs6000_components_for_bb): Handle the TOC component.
865 (rs6000_emit_prologue_components): Store the TOC register where needed.
866 (rs6000_set_handled_components): Mark TOC as handled, if handled.
867 (rs6000_emit_prologue): Don't save the TOC if that is already done.
869 2017-11-09 Martin Jambor <mjambor@suse.cz>
871 * ipa-param-manipulation.c: New file.
872 * ipa-param-manipulation.h: Likewise.
873 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
874 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
875 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
876 (ipa_parm_adjustment): Likewise.
877 (ipa_parm_adjustment_vec): Likewise.
878 (ipa_get_vector_of_formal_parms): Moved declaration to
879 ipa-param-manipulation.h.
880 (ipa_get_vector_of_formal_parm_types): Likewise.
881 (ipa_modify_formal_parameters): Likewise.
882 (ipa_modify_call_arguments): Likewise.
883 (ipa_combine_adjustments): Likewise.
884 (ipa_dump_param_adjustments): Likewise.
885 (ipa_modify_expr): Likewise.
886 (ipa_get_adjustment_candidate): Likewise.
887 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
888 ipa-param-manipulation.c.
889 (ipa_get_vector_of_formal_parm_types): Likewise.
890 (ipa_modify_formal_parameters): Likewise.
891 (ipa_modify_call_arguments): Likewise.
892 (ipa_modify_expr): Likewise.
893 (get_ssa_base_param): Likewise.
894 (ipa_get_adjustment_candidate): Likewise.
895 (index_in_adjustments_multiple_times_p): Likewise.
896 (ipa_combine_adjustments): Likewise.
897 (ipa_dump_param_adjustments): Likewise.
898 * tree-sra.c: Also include ipa-param-manipulation.h
899 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
902 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
903 Alan Hayward <alan.hayward@arm.com>
904 David Sherwood <david.sherwood@arm.com>
906 * doc/sourcebuild.texi (vect_masked_store): Document.
908 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
909 Alan Hayward <alan.hayward@arm.com>
910 David Sherwood <david.sherwood@arm.com>
912 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
914 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
915 Alan Hayward <alan.hayward@arm.com>
916 David Sherwood <david.sherwood@arm.com>
918 * doc/sourcebuild.texi (vect_variable_length): Document.
920 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
921 Alan Hayward <alan.hayward@arm.com>
922 David Sherwood <david.sherwood@arm.com>
924 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
926 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
927 Alan Hayward <alan.hayward@arm.com>
928 David Sherwood <david.sherwood@arm.com>
930 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
932 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
933 Alan Hayward <alan.hayward@arm.com>
934 David Sherwood <david.sherwood@arm.com>
936 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
937 previously undocumented selectors.
938 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
940 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
942 * doc/rtl.texi (const_vector): Say that elements can be
944 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
945 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
946 (gen_vec_duplicate): Use it instead of CONSTANT_P.
947 * optabs.c (expand_vector_broadcast): Likewise.
949 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
950 Alan Hayward <alan.hayward@arm.com>
951 David Sherwood <david.sherwood@arm.com>
953 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
954 scaled index even if the unscaled address was invalid.
955 Don't increase the complexity of using a scale in that case.
957 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
958 Alan Hayward <alan.hayward@arm.com>
959 David Sherwood <david.sherwood@arm.com>
961 * doc/rtl.texi: Rewrite the subreg rules so that they partition
962 the inner register into REGMODE_NATURAL_SIZE bytes rather than
963 UNITS_PER_WORD bytes.
964 * emit-rtl.c (validate_subreg): Divide subregs into blocks
965 based on REGMODE_NATURAL_SIZE of the inner mode.
966 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
967 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
968 * expmed.c (lowpart_bit_field_p): Divide the value up into
969 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
970 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
971 whether something is likely to occupy more than one register.
973 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
976 * ipa-inline-transform.c (update_noncloned_frequencies): Use
977 profile_count::adjust_for_ipa_scaling.
978 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
979 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
981 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
983 2017-11-09 Jakub Jelinek <jakub@redhat.com>
985 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
986 (split_group): Add total_orig and total_new arguments, estimate the
987 number of statements related to the store group without store merging
988 and with store merging.
989 (imm_store_chain_info::output_merged_store): Adjust split_group
990 callers, punt if estimated number of statements with store merging
991 is not smaller than estimated number of statements without it.
993 (handled_load): Remove has_single_use checks.
994 (pass_store_merging::process_store): Likewise.
996 2017-11-09 Richard Biener <rguenther@suse.de>
998 PR tree-optimization/82902
999 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1001 2017-11-09 Martin Liska <mliska@suse.cz>
1004 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1007 2017-11-09 Martin Liska <mliska@suse.cz>
1009 PR tree-optimization/82669
1010 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1012 2017-11-09 Martin Liska <mliska@suse.cz>
1014 PR gcov-profile/48463
1015 * coverage.c (coverage_begin_function): Output also end locus
1016 of a function and information whether the function is
1018 * gcov-dump.c (tag_function): Parse and print the information.
1019 * gcov.c (INCLUDE_MAP): Add include.
1020 (INCLUDE_SET): Likewise.
1021 (struct line_info): Move earlier in the source file because
1022 of vector<line_info> in function_info structure.
1023 (line_info::line_info): Likewise.
1024 (line_info::has_block): Likewise.
1025 (struct source_info): Add new member index.
1026 (source_info::get_functions_at_location): New function.
1027 (function_info::group_line_p): New function.
1028 (output_intermediate_line): New function.
1029 (output_intermediate_file): Use the mentioned function.
1030 (struct function_start): New.
1031 (struct function_start_pair_hash): Likewise.
1032 (process_file): Add code that identifies group functions.
1033 Assign lines either to global or function scope.
1034 (generate_results): Skip artificial functions.
1035 (find_source): Assign index for each source file.
1036 (read_graph_file): Read new flag artificial and end_line.
1037 (add_line_counts): Assign it either to global of function scope.
1038 (accumulate_line_counts): Isolate core of the function to
1039 accumulate_line_info and call it for both function and global
1041 (accumulate_line_info): New function.
1042 (output_line_beginning): Fix GNU coding style.
1043 (print_source_line): New function.
1044 (output_line_details): Likewise.
1045 (output_function_details): Likewise.
1046 (output_lines): Iterate both source (global) scope and function
1048 (struct function_line_start_cmp): New class.
1049 * doc/gcov.texi: Reflect changes in documentation.
1051 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1054 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1055 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1056 and similarly for not instead of neg.
1058 2017-11-08 Andi Kleen <ak@linux.intel.com>
1060 * config/i386/i386.opt: Add -mforce-indirect-call.
1061 * config/i386/predicates.md: Check for flag_force_indirect_call.
1062 * doc/invoke.texi: Document -mforce-indirect-call
1064 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1066 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1068 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1069 config/riscv/riscv.c (predict.h): New include.
1070 (riscv_slow_unaligned_access_p): No longer static.
1071 (riscv_block_move_straight): Add require.
1072 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1074 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1077 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1078 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1079 nonimmediate_operand predicate for operand 1 instead of
1082 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1084 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1086 * config/aarch64/constraints.md (Uml): New constraint.
1087 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1090 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1092 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1093 of two vec_duplicates into a vec_concat.
1095 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1097 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1098 Simplify vec_merge of vec_duplicate and vec_concat.
1099 * config/aarch64/constraints.md (Utq): New constraint.
1100 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1103 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1105 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1106 Simplify vec_merge of vec_duplicate and const_vector.
1107 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1109 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1110 mode iterator. Update predicate on operand 1 to
1111 handle non-const_vec constants. Delete constraints.
1112 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1114 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1116 PR tree-optimization/78821
1117 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1119 (store_operand_info::store_operand_info): Initialize it to false.
1120 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1121 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1122 store in the group, and if chain_info is non-NULL, to ignore altogether
1124 (compatible_load_p): Fail if bit_not_p does not match.
1125 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1126 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1127 (pass_store_merging::process_store): Adjust
1128 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1129 call terminate_all_aliasing_chains newly when adding a store into
1130 a chain with non-NULL chain_info.
1132 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1134 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1136 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1138 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1140 (aarch64_layout_frame): Initialise emit_frame_chain.
1141 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1142 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1144 2017-11-08 Martin Liska <mliska@suse.cz>
1146 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1147 of gimple_call_internal_p.
1149 2017-11-07 Tom de Vries <tom@codesourcery.com>
1151 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1153 2017-11-07 Tom de Vries <tom@codesourcery.com>
1155 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1158 2017-11-08 Martin Liska <mliska@suse.cz>
1161 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1163 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1165 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1167 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1170 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1171 (*cmp<mode>_ccz_1): New insn with $k alternative.
1174 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1175 RTX_COMM_COMPARE as commutative as well.
1176 (ix86_binary_operator_ok): Formatting fix.
1177 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1178 *<code><mode>3<mask_name><round_saeonly_name>,
1179 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1180 *<s>mul<mode>3_highpart<mask_name>,
1181 *vec_widen_umult_even_v16si<mask_name>,
1182 *vec_widen_umult_even_v8si<mask_name>,
1183 *vec_widen_umult_even_v4si<mask_name>,
1184 *vec_widen_smult_even_v16si<mask_name>,
1185 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1186 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1187 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1188 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1189 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1190 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1191 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1192 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1193 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1194 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1195 ix86_binary_operator_ok. Formatting fixes.
1196 (*<plusminus_insn><mode>3<mask_name><round_name>,
1197 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1200 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1202 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1203 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1204 eq and ne if TARGET_ISEL.
1205 (cmp): New code_iterator.
1206 (UNS, UNSU_, UNSIK): New code_attrs.
1207 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1208 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1210 ("eq<mode>3"): ... this.
1211 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1213 ("ne<mode>3"): ... this.
1215 2017-11-07 Julia Koval <julia.koval@intel.com>
1218 * common/config/i386/i386-common.c
1219 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1220 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1221 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1222 * config/i386/i386.opt: Ditto.
1223 * config/i386/i386.c (ix86_target_string): Ditto.
1224 (ix86_option_override_internal): Ditto.
1225 (ix86_init_mpx_builtins): Move MPX to args2.
1226 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1227 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1228 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1229 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1230 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1231 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1232 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1234 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1237 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1238 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1239 (zero-extendsidi peephole2): Remove peephole.
1241 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1244 * stor-layout.c: Include attribs.h.
1245 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1246 explicit lookup of "aligned" attribute.
1248 2017-11-07 Andrew Waterman <andrew@sifive.com>
1250 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1251 (riscv_expand_block_move): Likewise.
1252 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1254 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1255 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1256 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1257 (riscv_adjust_block_mem): Likewise.
1258 (riscv_block_move_loop): Likewise.
1259 (riscv_expand_block_move): Likewise.
1260 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1262 2017-11-07 Michael Clark <michaeljclark@mac.com>
1264 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1265 (MUSL_DYNAMIC_LINKER): Likewise.
1267 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1269 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1270 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1273 2017-11-07 Richard Biener <rguenther@suse.de>
1275 * match.pd: Fix build.
1277 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1278 Jackson Woodruff <jackson.woodruff@arm.com>
1280 PR tree-optimization/71026
1281 * match.pd: Canonicalize negate in division.
1283 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1286 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1288 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1290 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1291 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1293 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1295 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1296 non-scalar integral types.
1297 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1298 (-(A-B), -(~A)): New transformations.
1300 2017-11-07 Tom de Vries <tom@codesourcery.com>
1302 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1303 semicolon after "do {} while (0)".
1304 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1305 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1306 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1307 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1308 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1310 2017-11-07 Tom de Vries <tom@codesourcery.com>
1312 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1313 after "do {} while (0)".
1314 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1315 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1316 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1317 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1318 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1320 2017-11-07 Tom de Vries <tom@codesourcery.com>
1323 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1325 (arm_rtx_costs_internal): Add missing semicolon after
1326 HANDLE_NARROW_SHIFT_ARITH call.
1328 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1330 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1331 disable isel if it was not set explicitly.
1333 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1335 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1337 (add_type_attribute) likewise.
1339 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1341 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1342 of struct ix86_frame.
1343 (ix86_initial_elimination_offset): Likewise.
1344 (ix86_expand_split_stack_prologue): Likewise.
1346 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1348 * tree-vrp.h (enum value_range_type): Update stale comment.
1350 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1351 Alan Hayward <alan.hayward@arm.com>
1352 David Sherwood <david.sherwood@arm.com>
1354 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1355 (aarch64_expand_vec_perm_const): Take the number of units too.
1356 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1357 (aarch64_expand_vec_perm_const): Likewise.
1358 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1359 (vec_perm<mode>): Update accordingly.
1361 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1362 Alan Hayward <alan.hayward@arm.com>
1363 David Sherwood <david.sherwood@arm.com>
1365 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
1366 Take the number of units too.
1367 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1368 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1369 but check for a vector mode before rather than after the call.
1370 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1371 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1372 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1373 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1374 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1375 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1376 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1377 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1378 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1379 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1380 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1381 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1382 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1383 (aarch64_sqdmull2_n<mode>): Update accordingly.
1385 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1386 Alan Hayward <alan.hayward@arm.com>
1387 David Sherwood <david.sherwood@arm.com>
1389 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
1390 the number of units too.
1391 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
1392 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
1393 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
1394 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
1395 (vec_store_lanesxi<mode>): Update accordingly.
1397 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1398 Alan Hayward <alan.hayward@arm.com>
1399 David Sherwood <david.sherwood@arm.com>
1401 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
1402 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
1403 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
1404 of units rather than the mode.
1405 * config/aarch64/iterators.md (nunits): New mode attribute.
1406 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
1407 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
1408 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
1409 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
1410 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1411 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
1412 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
1413 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
1414 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
1415 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
1416 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1417 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
1418 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
1419 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
1420 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1421 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1422 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
1423 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
1424 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
1425 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
1426 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
1427 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
1428 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
1429 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1430 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
1431 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
1432 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1433 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
1434 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
1435 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
1436 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
1437 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
1438 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
1439 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
1440 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
1441 (aarch64_simd_vec_setv2di): Likewise.
1443 2017-11-06 Carl Love <cel@us.ibm.com>
1445 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
1447 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
1448 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
1450 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
1452 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
1453 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
1454 Add power 8 macro expansions.
1455 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
1456 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
1457 power 8 instructions. (VSX_XXBR): Add iterator.
1459 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
1461 * config/arm/arm.md (predicable_short_it): Change default to "no",
1462 improve documentation, remove uses that are identical to the default.
1463 (enabled_for_depr_it): Rename to enabled_for_short_it.
1464 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
1465 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
1466 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
1467 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
1468 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
1470 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1473 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
1474 float128 helper macros, which are no longer used after deleting
1475 the old 'q' built-in functions, and moving the round to odd
1476 built-in functions to being special built-in functions.
1477 (BU_FLOAT128_2): Likewise.
1478 (BU_FLOAT128_1_HW): Likewise.
1479 (BU_FLOAT128_2_HW): Likewise.
1480 (BU_FLOAT128_3_HW): Likewise.
1481 (FABSQ): Delete old 'q' built-in functions.
1482 (COPYSIGNQ): Likewise.
1483 (SQRTF128_ODD): Move round to odd built-in functions to be
1484 special built-in functions, so that we can handle
1485 -mabi=ieeelongdouble.
1486 (TRUNCF128_ODD): Likewise.
1487 (ADDF128_ODD): Likewise.
1488 (SUBF128_ODD): Likewise.
1489 (MULF128_ODD): Likewise.
1490 (DIVF128_ODD): Likewise.
1491 (FMAF128_ODD): Likewise.
1492 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
1493 built-in names to 'f128'.
1494 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
1495 old 'q' built-in functions, as the machine independent code for
1496 'f128' built-in functions handles this.
1497 (rs6000_expand_builtin): Add expansion for float128 round to odd
1498 functions, keying off on -mabi=ieeelongdouble of whether to use
1499 the KFmode or TFmode variant.
1500 (rs6000_init_builtins): Initialize the _Float128 round to odd
1502 * doc/extend.texi (PowerPC Built-in Functions): Document the old
1503 _Float128 'q' built-in functions are now mapped into the new
1504 'f128' built-in functions.
1506 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
1508 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
1509 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
1511 2017-11-06 David Malcolm <dmalcolm@redhat.com>
1514 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
1515 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
1516 * toplev.c: Include "ipa-fnsummary.h".
1517 (toplev::finalize): Call ipa_fnsummary_c_finalize.
1519 2017-11-06 Jakub Jelinek <jakub@redhat.com>
1521 PR tree-optimization/82838
1522 * gimple-ssa-store-merging.c
1523 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
1524 on a separate gimple_seq which is then appended to seq.
1526 2017-11-06 Jeff Law <law@redhat.com>
1529 * config/i386/i386.c (PROBE_INTERVAL): Remove.
1530 (get_probe_interval): New functions.
1531 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
1532 (ix86_adjust_stack_and_probe): Likewise.
1533 (output_adjust_stack_and_probe): Likewise.
1534 (ix86_emit_probe_stack_range): Likewise.
1535 (ix86_expand_prologue): Likewise.
1537 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1539 PR tree-optimization/82816
1540 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
1541 if the modes of the two types are the same.
1542 (convert_plusminus_to_widen): Likewise.
1544 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1546 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
1548 (usadv16qi): New define_expand.
1549 (usadv8hi): New define_expand.
1551 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
1554 * ipa-inline-transform.c (update_noncloned_frequencies): Always
1556 (inline_transform): Likewise.
1557 * predict.c (counts_to_freqs): Remove useless conditional.
1558 * profile-count.h (profile_count::apply_scale): Move sanity check.
1559 * tree-inline.c (copy_bb): Always scale.
1560 (copy_cfg_body): Likewise.
1562 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
1565 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
1567 (*cmp_ite0): Add enabled_for_depr_it attribute.
1568 (*cmp_ite1): Likewise.
1570 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1572 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
1575 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1577 * tree-vrp.c (vrp_int_const_binop): Return true on success and
1578 return the value by pointer.
1579 (extract_range_from_multiplicative_op_1): Update accordingly.
1580 Return as soon as an operation fails.
1582 2017-11-05 Tom de Vries <tom@codesourcery.com>
1585 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
1586 (DEF_SANITIZER_BUILTIN): ... here.
1587 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
1588 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
1590 2017-11-05 Tom de Vries <tom@codesourcery.com>
1593 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
1595 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
1596 ASM_OUTPUT_BEFORE_CASE_LABEL call.
1597 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
1599 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1600 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1601 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1603 2017-11-05 Tom de Vries <tom@codesourcery.com>
1606 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
1609 2017-11-04 Michael Clark <michaeljclark@mac.com>
1611 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
1612 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
1614 (*addsi3_extended): Likewise.
1615 (*addsi3_extended2): Likewise.
1616 (<optab>si3): Likewise.
1617 (<optab>di3): Likewise.
1618 (<optab><mode>3): Likewise.
1619 (<*optabe>si3_internal): Likewise.
1620 (zero_extendqi<SUPERQI:mode>2): Likewise.
1621 (*add<mode>hi3): Likewise.
1622 (*xor<mode>hi3): Likewise.
1623 (<optab>di3): Likewise.
1624 (*<optab>si3_extend): Likewise.
1625 (*sge<u>_<X:mode><GPR:mode>): Likewise.
1626 (*slt<u>_<X:mode><GPR:mode>): Likewise.
1627 (*sle<u>_<X:mode><GPR:mode>): Likewise.
1629 2017-11-04 Andrew Waterman <andrew@sifive.com>
1631 * config/riscv/riscv.c (riscv_option_override): Conditionally set
1632 TARGET_STRICT_ALIGN based upon -mtune argument.
1634 2017-11-04 Andrew Waterman <andrew@sifive.com>
1636 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
1638 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
1640 config/i386/i386.c (choose_basereg): Use optional scratch
1641 register and add assertion.
1642 (x86_emit_outlined_ms2sysv_save): Use scratch register when
1643 needed, and don't allocate stack.
1644 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
1645 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
1646 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
1648 2017-11-03 Jeff Law <law@redhat.com>
1650 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
1651 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
1652 to probe at the start of a noreturn function.
1654 2017-11-03 Jakub Jelinek <jakub@redhat.com>
1656 PR tree-optimization/78821
1657 * gimple-ssa-store-merging.c: Update the file comment.
1658 (MAX_STORE_ALIAS_CHECKS): Define.
1659 (struct store_operand_info): New type.
1660 (store_operand_info::store_operand_info): New constructor.
1661 (struct store_immediate_info): Add rhs_code and ops data members.
1662 (store_immediate_info::store_immediate_info): Add rhscode, op0r
1663 and op1r arguments to the ctor, initialize corresponding data members.
1664 (struct merged_store_group): Add load_align_base and load_align
1666 (merged_store_group::merged_store_group): Initialize them.
1667 (merged_store_group::do_merge): Update them.
1668 (merged_store_group::apply_stores): Pick the constant for
1669 encode_tree_to_bitpos from one of the two operands, or skip
1670 encode_tree_to_bitpos if neither operand is a constant.
1671 (class pass_store_merging): Add process_store method decl. Remove
1672 bool argument from terminate_all_aliasing_chains method decl.
1673 (pass_store_merging::terminate_all_aliasing_chains): Remove
1674 var_offset_p argument and corresponding handling.
1675 (stmts_may_clobber_ref_p): New function.
1676 (compatible_load_p): New function.
1677 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
1678 if there is overlap and rhs_code is not INTEGER_CST. For
1679 non-overlapping stores terminate group if rhs is not mergeable.
1680 (get_alias_type_for_stmts): Change first argument from
1681 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
1682 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
1683 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
1685 (get_location_for_stmts): Change first argument from
1686 auto_vec<gimple *> & to vec<gimple *> &.
1687 (struct split_store): Remove orig_stmts data member, add orig_stores.
1688 (split_store::split_store): Create orig_stores rather than orig_stmts.
1689 (find_constituent_stmts): Renamed to ...
1690 (find_constituent_stores): ... this. Change second argument from
1691 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
1692 to info structures rather than the statements.
1693 (split_group): Rename ALLOW_UNALIGNED argument to
1694 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
1695 it. Adjust find_constituent_stores caller.
1696 (imm_store_chain_info::output_merged_store): Handle rhs_code other
1697 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
1698 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
1699 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
1700 (mem_valid_for_store_merging): New function.
1701 (handled_load): New function.
1702 (pass_store_merging::process_store): New method.
1703 (pass_store_merging::execute): Use process_store method. Adjust
1704 terminate_all_aliasing_chains caller.
1706 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1708 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
1709 Return true for more constants, symbols and label references.
1710 (aarch64_valid_floating_const): Remove unused function.
1712 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1715 * config/aarch64/aarch64.c (aarch64_layout_frame):
1716 Undo forcing of LR at bottom of frame.
1718 2017-11-03 Jeff Law <law@redhat.com>
1721 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
1722 for int_registers_saved.
1724 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
1725 extracted from tree-ssa-dom.c.
1726 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
1727 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
1728 (record_equivalences_from_incoming_edge): Add additional argument
1729 to single_pred_edge_ignoring_loop_edges call.
1730 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
1731 (uncprop_dom_walker::before_dom_children): Add additional argument
1732 to single_pred_edge_ignoring_loop_edges call.
1733 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
1734 single_pred_edge_ignoring_loop_edges rather than open coding.
1735 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
1737 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
1739 * match.pd (-(-A)): Rewrite.
1741 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1743 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
1744 (rs6000_emit_int_cmove): New declaration.
1745 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
1746 (rs6000_emit_sISEL): Delete.
1747 (rs6000_emit_int_cmove): Make non-static.
1748 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
1749 instead of rs6000_emit_sISEL.
1751 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
1753 * asan.c (create_cond_insert_point): Maintain profile.
1754 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
1756 * basic-block.h (struct basic_block_def): Remove frequency.
1757 (EDGE_FREQUENCY): Use to_frequency
1758 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
1760 (find_traces): Update to use to_frequency.
1761 (find_traces_1_round): Likewise; use only IPA counts.
1762 (bb_to_key): Likewise.
1763 (connect_traces): Use IPA counts only.
1764 (copy_bb_p): Update to use to_frequency.
1765 (fix_up_crossing_landing_pad): Likewise.
1766 (sanitize_hot_paths): Likewise.
1767 * bt-load.c (basic_block_freq): Likewise.
1768 * cfg.c (init_flow): Set count_max to uninitialized.
1769 (check_bb_profile): Remove frequencies; check counts.
1770 (dump_bb_info): Do not dump frequencies.
1771 (update_bb_profile_for_threading): Update counts only.
1772 (scale_bbs_frequencies_int): Likewise.
1773 (MAX_SAFE_MULTIPLIER): Remove.
1774 (scale_bbs_frequencies_gcov_type): Update counts only.
1775 (scale_bbs_frequencies_profile_count): Update counts only.
1776 (scale_bbs_frequencies): Update counts only.
1777 * cfg.h (struct control_flow_graph): Add count-max.
1778 (update_bb_profile_for_threading): Update prototype.
1779 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
1780 (find_many_sub_basic_blocks): Likewise.
1781 * cfgcleanup.c (try_forward_edges): Likewise.
1782 (try_crossjump_to_edge): Likewise.
1783 * cfgexpand.c (expand_gimple_cond): Likewise.
1784 (expand_gimple_tailcall): Likewise.
1785 (construct_init_block): Likewise.
1786 (construct_exit_block): Likewise.
1787 * cfghooks.c (verify_flow_info): Check consistency of counts.
1788 (dump_bb_for_graph): Do not dump frequencies.
1789 (split_block_1): Do not update frequencies.
1790 (split_edge): Do not update frequencies.
1791 (make_forwarder_block): Do not update frequencies.
1792 (duplicate_block): Do not update frequencies.
1793 (account_profile_record): Do not update frequencies.
1794 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
1795 for global heuristics.
1796 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
1797 (expected_loop_iterations_unbounded): Use counts only.
1798 * cfgloopmanip.c (scale_loop_profile): Simplify.
1799 (create_empty_loop_on_edge): Simplify
1801 (duplicate_loop_to_header_edge): Simplify
1802 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
1803 (update_br_prob_note): Take care of removing note when profile
1805 (relink_block_chain): Do not dump frequency.
1806 (rtl_account_profile_record): Use to_frequency.
1807 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
1808 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
1809 (cgraph_update_edges_for_call_stmt_node): Likewise.
1810 (cgraph_edge::verify_count_and_frequency): Update.
1811 (cgraph_node::verify_node): Temporarily disable frequency verification.
1812 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
1813 to_cgraph_frequency.
1814 (cgraph_edge::rebuild_edges): Convert to ipa counts.
1815 * cgraphunit.c (init_lowered_empty_function): Do not initialize
1817 (cgraph_node::expand_thunk): Update profile.
1818 * except.c (dw2_build_landing_pads): Do not update frequency.
1819 * final.c (compute_alignments): Use to_frequency.
1820 (dump_basic_block_info): Do not dump frequency.
1821 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
1822 (dump_gimple_bb_header): Do not dump frequency.
1823 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
1825 * gimple-streamer-in.c (input_bb): Do not stream frequency.
1826 * gimple-streamer-out.c (output_bb): Do not stream frequency.
1827 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
1828 (init_before_recovery): Do not update frequency.
1829 (sched_create_recovery_edges): Do not update frequency.
1830 * hsa-gen.c (convert_switch_statements): Do not update frequency.
1831 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1832 (ipa_cp_c_finalize): Set max_count to uninitialized.
1833 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1834 (param_change_prob): Use counts.
1835 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1837 * ipa-split.c (consider_split): Use to_frequency.
1838 (split_function): Use to_frequency.
1839 * ira-build.c (loop_compare_func): Likewise.
1840 (mark_loops_for_removal): Likewise.
1841 (mark_all_loops_for_removal): Likewise.
1842 * loop-doloop.c (doloop_modify): Do not update frequency.
1843 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1845 * lto-streamer-in.c (input_function): Update count_max.
1846 * omp-expand.c (expand_omp_taskreg): Update count_max.
1847 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1848 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1849 (maybe_hot_count_p): Use ipa counts only.
1850 (maybe_hot_bb_p): Simplify.
1851 (maybe_hot_edge_p): Simplify.
1852 (probably_never_executed): Do not take frequency argument.
1853 (probably_never_executed_bb_p): Do not pass frequency.
1854 (probably_never_executed_edge_p): Likewise.
1855 (combine_predictions_for_bb): Check that profile is nonzero.
1856 (propagate_freq): Do not set frequency.
1857 (drop_profile): Simplify.
1858 (counts_to_freqs): Simplify.
1859 (expensive_function_p): Use to_frequency.
1860 (propagate_unlikely_bbs_forward): Simplify.
1861 (determine_unlikely_bbs): Simplify.
1862 (estimate_bb_frequencies): Add hack to silence graphite issues.
1863 (compute_function_frequency): Use ipa counts.
1864 (pass_profile::execute): Update.
1865 (rebuild_frequencies): Use counts only.
1866 (force_edge_cold): Use counts only.
1867 * profile-count.c (profile_count::dump): Dump new count types.
1868 (profile_count::differs_from_p): Check compatiblity.
1869 (profile_count::to_frequency): New function.
1870 (profile_count::to_cgraph_frequency): New function.
1871 * profile-count.h (struct function): Declare.
1872 (enum profile_quality): Add profile_guessed_local and
1873 profile_guessed_global0.
1874 (class profile_proability): Decrease number of bits to 29;
1875 update from_reg_br_prob_note and to_reg_br_prob_note.
1876 (class profile_count: Update comment; decrease number of bits
1877 to 61. Check compatibility.
1878 (profile_count::compatible_p): New private member function.
1879 (profile_count::ipa_p): New member function.
1880 (profile_count::operator<): Handle global zero correctly.
1881 (profile_count::operator>): Handle global zero correctly.
1882 (profile_count::operator<=): Handle global zero correctly.
1883 (profile_count::operator>=): Handle global zero correctly.
1884 (profile_count::nonzero_p): New member function.
1885 (profile_count::force_nonzero): New member function.
1886 (profile_count::max): New member function.
1887 (profile_count::apply_scale): Handle IPA scalling.
1888 (profile_count::guessed_local): New member function.
1889 (profile_count::global0): New member function.
1890 (profile_count::ipa): New member function.
1891 (profile_count::to_frequency): Declare.
1892 (profile_count::to_cgraph_frequency): Declare.
1893 * profile.c (OVERLAP_BASE): Delete.
1894 (compute_frequency_overlap): Delete.
1895 (compute_branch_probabilities): Do not use compute_frequency_overlap.
1896 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
1897 * sched-ebb.c (rank): Use counts only.
1898 * shrink-wrap.c (handle_simple_exit): Use counts only.
1899 (try_shrink_wrapping): Use counts only.
1900 (place_prologue_for_one_component): Use counts only.
1901 * tracer.c (find_best_predecessor): Use to_frequency.
1902 (find_trace): Use to_frequency.
1903 (tail_duplicate): Use to_frequency.
1904 * trans-mem.c (expand_transaction): Do not update frequency.
1905 * tree-call-cdce.c: Do not update frequency.
1906 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1907 (gimple_merge_blocks): Likewise.
1908 (gimple_split_edge): Likewise.
1909 (gimple_duplicate_sese_region): Likewise.
1910 (gimple_duplicate_sese_tail): Likewise.
1911 (move_sese_region_to_fn): Likewise.
1912 (gimple_account_profile_record): Likewise.
1913 (insert_cond_bb): Likewise.
1914 * tree-complex.c (expand_complex_div_wide): Likewise.
1915 * tree-eh.c (lower_resx): Update profile.
1916 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1918 (initialize_cfun): Do not initialize frequencies
1919 (freqs_to_counts): Delete.
1920 (copy_cfg_body): Ignore count parameter.
1921 (copy_body): Update.
1922 (expand_call_inline): Update count_max.
1923 (optimize_inline_calls): Update count_max.
1924 (tree_function_versioning): Update count_max.
1925 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1926 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1928 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1929 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1930 (try_peel_loop): Likewise.
1931 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1933 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1934 (tree_transform_and_unroll_loop): Do not use frequencies
1935 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
1936 Use reliable prediction only.
1937 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
1938 * tree-ssa-sink.c (select_best_block): Use to_frequency.
1939 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
1940 probability scaling.
1941 * tree-ssa-threadupdate.c (create_block_for_threading): Do
1942 not update frequency
1943 (any_remaining_duplicated_blocks): Likewise.
1944 (update_profile): Likewise.
1945 (estimated_freqs_path): Delete.
1946 (freqs_to_counts_path): Delete.
1947 (clear_counts_path): Delete.
1948 (ssa_fix_duplicate_block_edges): Likewise.
1949 (duplicate_thread_path): Likewise.
1950 * tree-switch-conversion.c (gen_inbound_check): Use counts.
1951 * tree-tailcall.c (decrease_profile): Do not update frequency.
1952 (eliminate_tail_call): Likewise.
1953 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
1954 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
1955 (optimize_mask_stores): Likewise.
1956 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
1957 * ubsan.c (ubsan_expand_null_ifn): Update profile.
1958 (ubsan_expand_ptr_ifn): Update profile.
1959 * value-prof.c (gimple_ic): Simplify.
1960 * value-prof.h (gimple_ic): Update prototype.
1961 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
1962 * ipa-inline.c (compute_uninlined_call_time): Be sure that
1964 (want_inline_self_recursive_call_p): Likewise.
1965 (resolve_noninline_speculation): Only cummulate defined counts.
1966 (inline_small_functions): Use nonzero_p.
1967 (ipa_inline): Do not access freed node.
1969 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1971 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1972 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
1974 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
1976 * config/riscv/riscv.c (riscv_legitimize_move): Handle
1977 non-legitimate address.
1979 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1981 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1982 (*lt0_<mode>di, *lt0_<mode>si): New.
1984 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1986 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
1987 TARGET_PAIRED_FLOAT.
1989 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
1990 Jim Wilson <jim.wilson@linaro.org>
1992 * config/aarch64/aarch64-cores.def (saphira): New CPU.
1993 * config/aarch64/aarch64-tune.md: Regenerated.
1994 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
1995 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
1997 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
1999 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2000 (arc_expand_prologue): Restore blink for millicode.
2001 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2003 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2006 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2007 gen_vec_duplicate after forcing the scalar into a register.
2009 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2011 * combine (try_combine): Print the insns input to try_combine to the
2014 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2017 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2018 Remove second argument from aarch64_process_target_attr call.
2019 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2021 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2023 (aarch64_handle_attr_arch): Remove second argument.
2024 (aarch64_handle_attr_cpu): Ditto.
2025 (aarch64_handle_attr_tune): Ditto.
2026 (aarch64_handle_attr_isa_flags): Ditto.
2027 (aarch64_process_one_target_attr): Ditto.
2028 (aarch64_process_target_attr): Ditto.
2029 (aarch64_option_valid_attribute_p): Remove second argument.
2030 on aarch64_process_target_attr call.
2032 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2034 * diagnostic.c: Include "selftest-diagnostic.h".
2035 (selftest::assert_location_text): New function.
2036 (selftest::test_diagnostic_get_location_text): New function.
2037 (selftest::diagnostic_c_tests): Call it.
2039 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2041 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2042 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2043 (class selftest::test_diagnostic_context): Move to...
2044 * selftest-diagnostic.c: New file.
2045 * selftest-diagnostic.h: New file.
2047 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2049 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2050 offset range for FT32B.
2051 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2052 * config/ft32/ft32.md: Add TARGET_NOPM.
2053 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2054 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2056 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2058 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2060 2017-11-02 Jeff Law <law@redhat.com>
2062 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2063 virtual keyword on FINAL OVERRIDE members.
2065 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2066 virtuals together. Add virtual destructor.
2067 (substitute_and_fold_engine): Similarly.
2069 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2071 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2073 2017-11-02 Richard Biener <rguenther@suse.de>
2075 PR tree-optimization/82795
2076 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2078 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2080 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2082 (gcc_GAS_CHECK_FEATURE): Remove.
2083 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2084 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2085 * configure: Regenerate.
2087 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2089 * config/arc/arc.c (hwloop_optimize): Account for empty
2092 2017-11-02 Richard Biener <rguenther@suse.de>
2095 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2096 Truncate ARRAY_REF index and element size.
2098 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2100 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2102 2017-11-01 Jeff Law <law@redhat.com>
2104 * tree-ssa-ccp.c (ccp_folder): New class derived from
2105 substitute_and_fold_engine.
2106 (ccp_folder::get_value): New member function.
2107 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2108 (ccp_fold_stmt): Remove prototype.
2109 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2110 * tree-ssa-copy.c (copy_folder): New class derived from
2111 substitute_and_fold_engine.
2112 (copy_folder::get_value): Renamed from get_value.
2113 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2114 * tree-vrp.c (vrp_folder): New class derived from
2115 substitute_and_fold_engine.
2116 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2117 (vrp_folder::get_value): New member function.
2118 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2119 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2120 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2121 provide a class interface to folder/substitute routines.
2122 (ssa_prop_fold_stmt_fn): Remove typedef.
2123 (ssa_prop_get_value_fn): Likewise.
2124 (subsitute_and_fold): Remove prototype.
2125 (replace_uses_in): Likewise.
2126 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2127 Renamed from replace_uses_in. Call the virtual member function
2128 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2129 (substitute_and_fold_dom_walker): Remove initialization of
2130 data member entries for calbacks. Add substitute_and_fold_engine
2131 member and initialize it.
2132 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2133 member functions for get_value, replace_phi_args_in c
2134 replace_uses_in, and fold_stmt calls.
2135 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2136 substitute_and_fold. Remove assert. Update ctor call.
2138 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2139 (ssa_prop_visit_phi_fn): Likewise.
2140 (class ssa_propagation_engine): New class to provide an interface
2142 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2144 (ssa_prop_visit_phi): Likewise.
2145 (ssa_propagation_engine::simulate_stmt): Moved into class.
2146 Call visit_phi/visit_stmt from the class rather than via
2147 file scoped static variables.
2148 (ssa_propagation_engine::simulate_block): Moved into class.
2149 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2150 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2151 set file scoped statics for the visit_stmt/visit_phi callbacks.
2152 * tree-complex.c (complex_propagate): New class derived from
2153 ssa_propagation_engine.
2154 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2155 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2156 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2158 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2159 ssa_propagation_engine.
2160 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2161 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2162 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2163 * tree-ssa-copy.c (copy_prop): New class derived from
2164 ssa_propagation_engine.
2165 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2166 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2167 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2168 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2169 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2170 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2171 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2173 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2175 PR rtl-optimization/82778
2176 PR rtl-optimization/82597
2177 * compare-elim.c (struct comparison): Add in_a_setter field.
2178 (find_comparison_dom_walker::before_dom_children): Remove killed
2179 bitmap and df_simulate_find_defs call, instead walk the defs.
2180 Compute last_setter and initialize in_a_setter. Merge definitions
2181 with first initialization for a few variables.
2182 (try_validate_parallel): Use insn_invalid_p instead of
2183 recog_memoized. Return insn rather than just the pattern.
2184 (try_merge_compare): Fix up comment. Don't uselessly test if
2185 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2187 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2190 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2191 Alan Hayward <alan.hayward@arm.com>
2192 David Sherwood <david.sherwood@arm.com>
2194 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2195 aarch64_hard_regno_nregs to get the number of registers
2198 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2199 Alan Hayward <alan.hayward@arm.com>
2200 David Sherwood <david.sherwood@arm.com>
2202 * config/aarch64/constraints.md (Upl): Rename to...
2204 * config/aarch64/aarch64.md
2205 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2208 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2209 Alan Hayward <alan.hayward@arm.com>
2210 David Sherwood <david.sherwood@arm.com>
2212 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2213 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2216 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2217 Alan Hayward <alan.hayward@arm.com>
2218 David Sherwood <david.sherwood@arm.com>
2220 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2221 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2222 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2224 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2225 of a single element.
2226 * config/aarch64/iterators.md: Add a comment above the permute
2227 unspecs to say that they are generated directly by
2228 aarch64_expand_vec_perm_const.
2229 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2231 2017-11-01 Nathan Sidwell <nathan@acm.org>
2233 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2235 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2237 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2238 and medany code models, and describe what they do.
2240 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2242 Revert accidental duplicate:
2244 * combine.c (can_change_dest_mode): Reject changes in
2245 REGMODE_NATURAL_SIZE.
2247 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2249 PR rtl-optimization/64682
2250 PR rtl-optimization/69567
2251 PR rtl-optimization/69737
2252 PR rtl-optimization/82683
2253 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2254 register mentioned in the note, drop the note, unless it came from I3,
2255 in which case it should go to I3 again.
2257 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2259 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2260 and return false if not.
2261 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2263 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2265 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2266 range is known to be empty.
2268 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2269 Alan Hayward <alan.hayward@arm.com>
2270 David Sherwood <david.sherwood@arm.com>
2272 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2273 and CONST_VECTOR_NUNITS instead of computing the number of units from
2274 the byte sizes of the vector and element.
2275 (simplify_binary_operation_1): Likewise.
2276 (simplify_const_binary_operation): Likewise.
2277 (simplify_ternary_operation): Likewise.
2279 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2280 Alan Hayward <alan.hayward@arm.com>
2281 David Sherwood <david.sherwood@arm.com>
2283 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2284 (int_mem_offset): ...this new function.
2285 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2286 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2287 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
2290 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2291 Alan Hayward <alan.hayward@arm.com>
2292 David Sherwood <david.sherwood@arm.com>
2294 * lower-subreg.c (interesting_mode_p): New function.
2295 (compute_costs, find_decomposable_subregs, decompose_register)
2296 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2297 (resolve_clobber, dump_choices): Use it.
2299 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2300 Alan Hayward <alan.hayward@arm.com>
2301 David Sherwood <david.sherwood@arm.com>
2303 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
2305 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2306 Alan Hayward <alan.hayward@arm.com>
2307 David Sherwood <david.sherwood@arm.com>
2309 * alias.c (find_base_value, find_base_term): Only process integer
2310 truncations. Check the precision rather than the size.
2312 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2313 Alan Hayward <alan.hayward@arm.com>
2314 David Sherwood <david.sherwood@arm.com>
2316 * machmode.h (is_narrower_int_mode): New function
2317 * optabs.c (expand_float, expand_fix): Use it.
2318 * dwarf2out.c (rotate_loc_descriptor): Likewise.
2320 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2321 Alan Hayward <alan.hayward@arm.com>
2322 David Sherwood <david.sherwood@arm.com>
2324 * rtl.h (narrower_subreg_mode): New function.
2325 * ira-color.c (update_costs_from_allocno): Use it.
2327 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2328 Alan Hayward <alan.hayward@arm.com>
2329 David Sherwood <david.sherwood@arm.com>
2331 * optabs-query.h (convert_optab_p): New function, split out from...
2332 (convert_optab_handler): ...here.
2333 (widening_optab_handler): Delete.
2334 (find_widening_optab_handler): Remove permit_non_widening parameter.
2335 (find_widening_optab_handler_and_mode): Likewise. Provide an
2336 override that operates on mode class wrappers.
2337 * optabs-query.c (widening_optab_handler): Delete.
2338 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2339 parameter. Assert that the two modes are the same class and that
2340 the "from" mode is narrower than the "to" mode. Use
2341 convert_optab_handler instead of widening_optab_handler.
2342 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2343 instead of widening_optab_handler.
2344 * expr.c (expand_expr_real_2): Update calls to
2345 find_widening_optab_handler.
2346 * optabs.c (expand_widen_pattern_expr): Likewise.
2347 (expand_binop_directly): Take the insn_code as a parameter.
2348 (expand_binop): Only call find_widening_optab_handler for
2349 conversion optabs; use optab_handler otherwise. Update calls
2350 to find_widening_optab_handler and expand_binop_directly.
2351 Use convert_optab_handler instead of widening_optab_handler.
2352 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2353 find_widening_optab_handler and use scalar_mode rather than
2355 (convert_plusminus_to_widen): Likewise.
2357 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2358 Alan Hayward <alan.hayward@arm.com>
2359 David Sherwood <david.sherwood@arm.com>
2361 * machmode.h (fixed_size_mode): New class.
2362 * rtl.h (get_pool_mode): Return fixed_size_mode.
2363 * gengtype.c (main): Add fixed_size_mode.
2364 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2365 (get_raw_arg_mode): Likewise.
2366 * doc/tm.texi: Regenerate.
2367 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2368 * targhooks.c (default_get_reg_raw_mode): Likewise.
2369 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2370 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2371 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2372 (msp430_get_raw_result_mode): Likewise.
2373 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2374 * dbxout.c (dbxout_parms): Require fixed-size modes.
2375 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2376 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2377 * omp-low.c (lower_oacc_reductions): Likewise.
2378 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2379 (simplify_subreg): Update accordingly.
2380 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2381 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2382 that aren't fixed-size.
2383 (get_pool_mode): Return a fixed_size_mode.
2384 (output_constant_pool_2): Take a fixed_size_mode.
2386 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2387 Alan Hayward <alan.hayward@arm.com>
2388 David Sherwood <david.sherwood@arm.com>
2390 * doc/rtl.texi (vec_series): Document.
2391 (const): Say that the operand can be a vec_series.
2392 * rtl.def (VEC_SERIES): New rtx code.
2393 * rtl.h (const_vec_series_p_1): Declare.
2394 (const_vec_series_p): New function.
2395 * emit-rtl.h (gen_const_vec_series): Declare.
2396 (gen_vec_series): Likewise.
2397 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
2398 (gen_vec_series): Likewise.
2399 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
2400 * simplify-rtx.c (simplify_unary_operation): Handle negations
2402 (simplify_binary_operation_series): New function.
2403 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
2404 (test_vector_ops_series): New function.
2405 (test_vector_ops): Call it.
2406 * config/powerpcspe/altivec.md (altivec_lvsl): Use
2407 gen_const_vec_series.
2408 (altivec_lvsr): Likewise.
2409 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
2411 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2412 Alan Hayward <alan.hayward@arm.com>
2413 David Sherwood <david.sherwood@arm.com>
2415 * doc/rtl.texi (const): Update description of address constants.
2416 Say that vector constants are allowed too.
2417 * common.md (E, F): Use CONSTANT_P instead of checking for
2419 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
2420 checking for CONST_VECTOR.
2421 * expmed.c (make_tree): Use build_vector_from_val for a CONST
2423 * expr.c (expand_expr_real_2): Check for vector modes instead
2424 of checking for CONST_VECTOR.
2425 * rtl.h (const_vec_p): New function.
2426 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
2427 (unwrap_const_vec_duplicate): Handle them here too.
2429 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2430 David Malcolm <dmalcolm@redhat.com>
2431 Alan Hayward <alan.hayward@arm.com>
2432 David Sherwood <david.sherwood@arm.com>
2434 * rtl.h (vec_duplicate_p): New function.
2435 * selftest-rtl.c (assert_rtx_eq_at): New function.
2436 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
2437 (assert_rtx_eq_at): Declare.
2438 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
2439 * selftest-run-tests.c (selftest::run_tests): Call it.
2440 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
2441 (simplify_unary_operation_1): Recursively handle vector duplicates.
2442 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
2444 (simplify_subreg): Handle subregs of vector duplicates.
2445 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
2446 (selftest::simplify_rtx_c_tests): New functions.
2448 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2449 Alan Hayward <alan.hayward@arm.com>
2450 David Sherwood <david.sherwood@arm.com>
2452 * emit-rtl.h (gen_const_vec_duplicate): Declare.
2453 (gen_vec_duplicate): Likewise.
2454 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
2456 (gen_const_vector): ...here.
2457 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
2458 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
2459 whose elements are all equal.
2460 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
2461 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
2462 (simplify_relational_operation): Likewise.
2463 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2465 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
2466 (aarch64_expand_vector_init): Likewise.
2467 * config/arm/arm.c (neon_vdup_constant): Likewise.
2468 (neon_expand_vector_init): Likewise.
2469 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
2470 (arm_block_set_unaligned_vect): Likewise.
2471 (arm_block_set_aligned_vect): Likewise.
2472 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
2473 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
2474 (expand_vec_perm_even_odd_pack): Likewise.
2475 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
2476 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
2477 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
2478 gen_const_vec_duplicate.
2479 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
2480 * config/mips/mips.c (mips_gen_const_int_vector): Use
2481 gen_const_vec_duplicate.
2482 (mips_expand_vector_init): Use CONST0_RTX.
2483 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
2484 (define_split): Use gen_const_vec_duplicate.
2485 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
2486 (define_split): Use gen_const_vec_duplicate.
2487 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
2488 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
2489 * config/spu/spu.c (spu_const): Likewise.
2491 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2492 Alan Hayward <alan.hayward@arm.com>
2493 David Sherwood <david.sherwood@arm.com>
2495 * combine.c (can_change_dest_mode): Reject changes in
2496 REGMODE_NATURAL_SIZE.
2498 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
2500 * configure.ac (--enable-libssp): New.
2501 (gcc_cv_libc_provides_ssp): Check for explicit setting before
2502 trying to determine target-specific default. Adjust indentation.
2503 * configure: Regenerated.
2504 * doc/install.texi (Configuration): Expand --disable-libssp
2507 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
2509 config/i386/i386.c (ix86_expand_epilogue): Correct stack
2512 2017-10-31 Martin Jambor <mjambor@suse.cz>
2515 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
2517 2017-10-31 David Malcolm <dmalcolm@redhat.com>
2519 * auto-profile.c (autofdo_source_profile::read): Use
2520 UNKNOWN_LOCATION rather than 0.
2521 * diagnostic-core.h (warning_at_rich_loc): Rename to...
2522 (warning_at): ...this overload.
2523 (warning_at_rich_loc_n): Rename to...
2524 (warning_n): ...this overload.
2525 (error_at_rich_loc): Rename to...
2526 (error_at): ...this overload.
2527 (pedwarn_at_rich_loc): Rename to...
2528 (pedwarn): ...this overload.
2529 (permerror_at_rich_loc): Rename to...
2530 (permerror): ...this overload.
2531 (inform_at_rich_loc): Rename to...
2532 (inform): ...this overload.
2533 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
2534 (diagnostic_n_impl_richloc): Rename to...
2535 (diagnostic_n_impl): ...this rich_location *-based decl.
2536 (inform_at_rich_loc): Rename to...
2537 (inform): ...this, and add an assertion.
2538 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
2539 (warning_at_rich_loc): Rename to...
2540 (warning_at): ...this, and add an assertion.
2541 (warning_at_rich_loc_n): Rename to...
2542 (warning_n): ...this, and add an assertion.
2543 (warning_n): Update location_t-based implementation for removal of
2544 location_t-based diagnostic_n_impl.
2545 (pedwarn_at_rich_loc): Rename to...
2546 (pedwarn): ...this, and add an assertion.
2547 (permerror_at_rich_loc): Rename to...
2548 (permerror): ...this, and add an assertion.
2549 (error_n): Update for removal of location_t-based diagnostic_n_impl.
2550 (error_at_rich_loc): Rename to...
2551 (error_at): ...this, and add an assertion.
2552 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
2553 (driver::do_spec_on_infiles): Likewise.
2554 * substring-locations.c (format_warning_va): Update for renaming
2555 of inform_at_rich_loc.
2557 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2559 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
2560 _Float<N>X built-in functions so that the variant without the
2561 "__builtin_" prefix is only enabled for the GNU C and Objective C
2562 languages when they are in non-strict ANSI/ISO mode.
2563 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
2564 * target.def (floatn_builtin_p): Add a target hook to control
2565 whether _Float<N> and _Float<N>X built-in functions without the
2566 "__builtin_" prefix are enabled, and return true for C and
2567 Objective C in the default hook. Include langhooks.h in
2569 * targhooks.h (default_floatn_builtin_p): Likewise.
2570 * targhooks.c (default_floatn_builtin_p): Likewise.
2571 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
2572 floatn_builtin_p target hook.
2573 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
2575 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
2576 Eric Botcazou <ebotcazou@adacore.com>
2578 PR rtl-optimization/81803
2579 * lra-constraints.c (curr_insn_transform): Also reload the whole
2580 register for a strict subreg no wider than a word if this is for
2581 a WORD_REGISTER_OPERATIONS target.
2583 2017-10-31 Jason Merrill <jason@redhat.com>
2585 * gdbinit.in: Skip over inlines from timevar.h.
2587 2017-10-31 Martin Liska <mliska@suse.cz>
2589 * doc/gcov.texi: Document new option.
2590 * gcov.c (print_usage): Likewise print it.
2591 (process_args): Support the argument.
2592 (format_count): New function.
2593 (format_gcov): Use the function.
2595 2017-10-31 Martin Liska <mliska@suse.cz>
2597 * gcov.c (struct name_map): do not use typedef.
2598 Define operator== and operator<.
2599 (name_search): Remove.
2600 (name_sort): Remove.
2601 (main): Do not allocate names.
2602 (process_file): Add vertical space.
2603 (generate_results): Use std::find.
2604 (release_structures): Do not release memory.
2605 (find_source): Use std::find.
2607 2017-10-31 Martin Liska <mliska@suse.cz>
2609 * gcov.c (struct line_info): Remove it's typedef.
2610 (line_info::line_info): Add proper ctor.
2611 (line_info::has_block): Do not use a typedef.
2612 (struct source_info): Do not use typedef.
2613 (circuit): Likewise.
2614 (get_cycles_count): Likewise.
2615 (output_intermediate_file): Iterate via vector iterator.
2616 (add_line_counts): Use std::vector methods.
2617 (accumulate_line_counts): Likewise.
2618 (output_lines): Likewise.
2620 2017-10-31 Martin Liska <mliska@suse.cz>
2622 * gcov.c (struct source_info): Remove typedef.
2623 (source_info::source_info): Add proper ctor.
2624 (accumulate_line_counts): Use struct, not it's typedef.
2625 (output_gcov_file): Likewise.
2626 (output_lines): Likewise.
2627 (main): Do not allocate an array.
2628 (output_intermediate_file): Use size of vector container.
2629 (process_file): Resize the vector.
2630 (generate_results): Do not preallocate, use newly added vector
2632 (release_structures): Do not release sources.
2633 (find_source): Use vector methods.
2634 (add_line_counts): Do not use typedef.
2636 2017-10-31 Martin Liska <mliska@suse.cz>
2638 * doc/gcov.texi: Document that.
2639 * gcov.c (add_line_counts): Mark lines with a non-executed
2641 (output_line_beginning): Handle such lines.
2642 (output_lines): Pass new argument.
2643 (output_intermediate_file): Print it in intermediate format.
2645 2017-10-31 Martin Liska <mliska@suse.cz>
2647 * color-macros.h: New file.
2648 * diagnostic-color.c: Factor out color related to macros to
2650 * doc/gcov.texi: Document -k option.
2651 * gcov.c (INCLUDE_STRING): Include string.h.
2652 (print_usage): Add -k option.
2653 (process_args): Parse it.
2654 (pad_count_string): New function.
2655 (output_line_beginning): Likewise.
2656 (DEFAULT_LINE_START): New macro.
2657 (output_lines): Support color output.
2659 2017-10-31 Martin Liska <mliska@suse.cz>
2661 PR gcov-profile/82633
2662 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
2663 their interaction with GCOV infrastructure.
2664 * configure.ac: Add -fkeep-{inline,static}-functions to
2666 * configure: Regenerate.
2668 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
2671 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
2673 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
2676 * config/rs6000/rs6000.md (allocate_stack): Force update interval
2677 into a register if it does not fit into an immediate offset field.
2679 2017-10-31 Olivier Hainque <hainque@adacore.com>
2681 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
2683 2017-10-31 Julia Koval <julia.koval@intel.com>
2685 * config.gcc: Add gfniintrin.h.
2686 * config/i386/gfniintrin.h: New.
2687 * config/i386/i386-builtin-types.def
2688 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
2689 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
2690 __builtin_ia32_vgf2p8affineinvqb_v32qi,
2691 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
2692 __builtin_ia32_vgf2p8affineinvqb_v16qi,
2693 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
2694 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
2695 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
2696 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
2697 V64QI_FTYPE_V64QI_V64QI_INT): New types.
2698 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2699 * config/i386/immintrin.h: Include gfniintrin.h.
2700 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
2702 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
2704 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
2706 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2708 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
2709 (arm_ashldi3_1bit): Remove pattern.
2710 (ashrdi3): Remove shift by 1 expansion.
2711 (arm_ashrdi3_1bit): Remove pattern.
2712 (lshrdi3): Remove shift by 1 expansion.
2713 (arm_lshrdi3_1bit): Remove pattern.
2714 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
2715 cost of ashldi3 by 1.
2716 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
2717 (<shift>di3_neon): Likewise.
2719 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2721 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
2722 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
2723 and (*aarch64_simd_mov<VQ:mode>).
2724 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
2725 pattern alternative.
2726 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
2727 attributes to match pattern alternative.
2729 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
2731 * config.gcc (powerpc*-*-*): Add emmintrin.h.
2732 * config/rs6000/emmintrin.h: New file.
2733 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
2735 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
2737 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
2738 * (movdi_vfp_cortexa8): Remove pattern.
2740 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2742 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
2744 (Specific, alpha*-dec-osf5.1): Remove.
2745 (Specific, mips-sgi-irix5): Remove.
2746 (Specific, mips-sgi-irix6): Remove.
2748 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2751 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
2752 arguments to clear_bit_region_be.
2754 2017-10-30 Jim Wilson <wilson@tuliptree.org>
2756 * gimplify.c: Include memmodel.h.
2758 2017-10-30 Martin Jambor <mjambor@suse.cz>
2760 * omp-grid.c (grid_attempt_target_gridification): Also insert a
2761 condition whether loop should be executed at all.
2763 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
2765 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
2766 gimple folding of vec_madd() intrinsics.
2767 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
2768 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
2769 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
2771 2017-10-30 Richard Biener <rguenther@suse.de>
2773 PR tree-optimization/82762
2775 2017-10-23 Richard Biener <rguenther@suse.de>
2777 PR tree-optimization/82129
2779 2017-08-01 Richard Biener <rguenther@suse.de>
2781 PR tree-optimization/81181
2782 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2783 (compute_antic): ... end of iteration here.
2785 2017-10-30 Joseph Myers <joseph@codesourcery.com>
2787 * doc/invoke.texi (C Dialect Options): Document -std=c17,
2788 -std=iso9899:2017 and -std=gnu17.
2789 * doc/standards.texi (C Language): Document C17 support.
2790 * doc/cpp.texi (Overview): Mention -std=c17.
2791 (Standard Predefined Macros): Document C11 and C17 values of
2792 __STDC_VERSION__. Do not refer to C99 support as incomplete.
2793 * doc/extend.texi (Inline): Do not list individual options for
2794 standards newer than C99.
2795 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
2797 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
2800 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
2802 * asan.c (asan_finish_file): Align asan globals array by shadow
2805 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2808 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
2809 (struct store_immediate_info): Add bitregion_start and bitregion_end
2811 (store_immediate_info::store_immediate_info): Add brs and bre
2812 arguments and initialize bitregion_{start,end} from those.
2813 (struct merged_store_group): Add bitregion_start, bitregion_end,
2814 align_base and mask fields. Drop unnecessary struct keyword from
2815 struct store_immediate_info. Add do_merge method.
2816 (clear_bit_region_be): Use memset instead of loop storing zeros.
2817 (merged_store_group::do_merge): New method.
2818 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
2819 stores as long as the surrounding bitregions have no gaps.
2820 (merged_store_group::merge_overlapping): Use do_merge.
2821 (merged_store_group::apply_stores): Test that bitregion_{start,end}
2822 is byte aligned, rather than requiring that start and width are
2823 byte aligned. Drop unnecessary struct keyword from
2824 struct store_immediate_info. Allocate and populate also mask array.
2825 Make start of the arrays relative to bitregion_start rather than
2826 start and size them according to bitregion_{end,start} difference.
2827 (struct imm_store_chain_info): Drop unnecessary struct keyword from
2828 struct store_immediate_info.
2829 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
2830 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
2831 struct keyword from struct store_immediate_info.
2832 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2833 between stores as long as the surrounding bitregions have no gaps.
2835 (struct split_store): Add orig non-static data member.
2836 (split_store::split_store): Initialize orig to false.
2837 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2838 if there is exactly a single original stmt. Change stmts argument
2839 to pointer from reference, if NULL, don't push anything to it. Add
2840 first argument, use it to optimize skipping over orig stmts that
2841 are known to be before bitpos already. Simplify.
2842 (split_group): Return unsigned int count how many stores are or
2843 would be needed rather than a bool. Add allow_unaligned argument.
2844 Change split_stores argument from reference to pointer, if NULL,
2845 only do a dry run computing how many stores would be produced.
2846 Rewritten algorithm to use both alignment and misalign if
2847 !allow_unaligned and handle bitfield stores with gaps.
2848 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2849 from bitregion_start instead of start. Compute allow_unaligned
2850 here, if true, do 2 split_group dry runs to compute which one
2851 produces fewer stores and prefer aligned if equal. Punt if
2852 new count is bigger or equal than original before emitting any
2853 statements, rather than during that. Remove no longer needed
2854 new_ssa_names tracking. Replace num_stmts with
2855 split_stores.length (). Use 32-bit stack allocated entries
2856 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2857 if possible. Handle bitfields with gaps.
2858 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2859 Compute bitregion_{start,end} for the stores and construct
2860 store_immediate_info with that. Formatting fixes.
2862 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2865 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2866 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2868 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2870 * gimplify.c: Include tm_p.h.
2872 * common.opt (gcoff): Re-add as ignored option.
2873 (gcoff1, gcoff2, gcoff3): Likewise.
2875 * Makefile.in (OBJS): Delete sdbout.o.
2876 (GTFILES): Delete $(srcdir)/sdbout.c.
2877 * debug.h: Delete sdb_debug_hooks.
2878 * final.c: Delete sdbout.h include.
2879 (final_scan_insn): Delete SDB_DEBUG check.
2880 (rest_of_clean_state): Likewise.
2881 * output.h: Delete sdb_begin_function_line.
2884 * toplev.c: Delete sdbout.h include.
2885 (process_options): Delete SDB_DEBUG check.
2886 * tree-core.h (tree_type_common): Delete pointer field of
2888 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
2889 TYPE_SYMTAB_POINTER.
2890 * tree.h (TYPE_SYMTAB_POINTER): Delete.
2891 (TYPE_SYMTAB_IS_POINTER): Delete.
2892 (TYPE_SYMTAB_IS_DIE): Renumber.
2893 * xcoffout.c: Refer to former sdbout.c file.
2894 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
2896 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
2897 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
2898 (Debugging Options): Delete -gcoff.
2899 (-gstabs): Delete SDB reference.
2901 (-gcoff@var{level}): Delete.
2902 * doc/passes.texi (Debugging information output): Delete SDB and
2903 sdbout.c references.
2904 * doc/tm.texi: Regenerate.
2905 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2906 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2908 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2909 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2910 (SDB_DEBUGGING_INFO): Delete.
2911 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2912 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2913 * target.def (output_source_filename): Delete COFF reference.
2915 * common.opt (gcoff): Delete.
2916 (gxcoff+): Update Negative chain.
2917 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2919 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2920 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2921 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2923 (expand_function_start): Change sdb reference to past tense.
2924 (expand_function_end): Change sdb reference to past tense.
2925 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2926 * opts.c (debug_type_names): Delete coff entry.
2927 (common_handle_option): Delete OPT_gcoff case.
2928 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2930 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2931 * config/cris/cris.h: Delete SDB reference in comment.
2932 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2933 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2934 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
2935 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
2937 (ix86_expand_prologue): Likewise.
2938 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
2939 * config/ia64/ia64.h: Likewise.
2940 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
2941 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
2943 * config/mmix/mmix.h: Likewise.
2944 * config/nds32/nds32.c: Likewise.
2945 * config/stormy/storym16.h: Likewise.
2946 * config/visium/visium.h: Likewise.
2947 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
2949 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
2951 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
2952 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
2953 FRAME_POINTER_REGNUM point at high end of local var area.
2955 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
2957 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
2958 Move comment around. Do not reset best_edge for a copiable
2959 destination if the copy would cause a partition change.
2960 (better_edge_p): Remove redundant check.
2962 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2964 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
2966 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2968 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
2969 for math functions that have _Float<N> and _Float<N>X variants.
2970 (mathfn_built_in_2): Add support for math functions that have
2971 _Float<N> and _Float<N>X variants.
2972 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
2973 (expand_builtin_mathfn_ternary): Add support for fma with
2974 _Float<N> and _Float<N>X variants.
2975 (expand_builtin): Likewise.
2976 (fold_builtin_3): Likewise.
2977 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
2978 create math function _Float<N> and _Float<N>X variants as external
2980 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
2981 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2982 the __builtin_ prefix and if not strict ansi, without the prefix.
2983 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2984 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2985 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2986 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
2987 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
2988 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
2989 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
2990 function signatures for fma _Float<N> and _Float<N>X variants.
2991 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
2992 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
2993 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
2994 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
2995 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
2996 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
2997 * gencfn-macros.c (print_case_cfn): Add support for math functions
2998 that have _Float<N> and _Float<N>X variants.
2999 (print_define_operator_list): Likewise.
3000 (fltfn_suffixes): Likewise.
3002 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3003 for math functions that have _Float<N> and _Float<N>X variants.
3004 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3005 and _Float<N>X variants.
3006 (COPYSIGN): Likewise.
3009 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3010 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3012 (integer_valued_read_call_p): Likewise.
3013 * fold-const-call.c (fold_const_call_ss): Likewise.
3014 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3015 _Float<N> and _Float<N>X variants.
3016 (fold_const_call_ssss): Add support for fma _Float<N> and
3017 _Float<N>X variants.
3018 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3019 support for copysign and fma _Float<N> and _Float<N>X variants.
3020 (backprop::process_builtin_call_use): Likewise.
3021 * tree-call-cdce.c (can_test_argument_range); Add support for
3022 sqrt _Float<N> and _Float<N>X variants.
3023 (edom_only_function): Likewise.
3024 (get_no_error_domain): Likewise.
3025 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3026 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3027 copysign _Float<N> and _Float<N>X variants.
3028 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3029 handled by machine independent code.
3030 (FMAF128): Likewise.
3031 * doc/cpp.texi (Common Predefined Macros): Document defining
3032 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3033 fma _Float<N> and _Float<N>X variants.
3035 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3038 * config/i386/i386-modes.def (CCFPU): Remove definition.
3039 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3040 (ix86_cc_modes_compatible): Ditto.
3041 (ix86_expand_carry_flag_compare): Ditto.
3042 (ix86_expand_int_movcc): Ditto.
3043 (ix86_expand_int_addcc): Ditto.
3044 (ix86_reverse_condition): Ditto.
3045 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3046 Return true/false for unordered/ordered fp comparisons.
3047 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3048 (ix86_prepare_fp_compare_args): Update for rename.
3049 (ix86_expand_fp_compare): Update for rename. Generate unordered
3050 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3051 (ix86_expand_sse_compare_and_jump): Ditto.
3052 * config/i386/predicates.md (fcmov_comparison_operator):
3053 Remove CCFPU mode handling.
3054 (ix86_comparison_operator): Ditto.
3055 (ix86_carry_flag_operator): Ditto.
3056 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3057 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3058 (*cmpu<mode>_cc_i387): Ditto.
3059 (FPCMP): Remove mode iterator.
3060 (unord): Remove mode attribute.
3061 (unord_subst): New define_subst transformation
3062 (unord): New define_subst attribute.
3064 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3065 (*cmpi<unord>xf_i387): Ditto.
3066 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3067 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3068 using unord_subst transformation.
3069 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3070 (round_saeonly): Also handle CCFP mode.
3071 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3072 Remove UNSPEC_SAHF unspec handling.
3074 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3076 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3078 2017-10-27 Jeff Law <law@redhat.com>
3080 * gimple-ssa-sprintf.c: Include domwalk.h.
3081 (class sprintf_dom_walker): New class, derived from dom_walker.
3082 (sprintf_dom_walker::before_dom_children): New function.
3083 (struct call_info): Moved into sprintf_dom_walker class
3084 (compute_formath_length, handle_gimple_call): Likewise.
3085 (sprintf_length::execute): Call the dominator walker rather
3086 than walking the statements.
3088 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3089 gimple statement locations.
3090 (check_array_bounds): Corresponding changes. Get the statement's
3091 location directly from wi->stmt.
3093 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3096 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3098 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3100 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3101 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3103 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3106 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3107 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3108 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3109 maybe_get_pool_constant.
3110 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3113 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3115 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3116 2.26 caveat. Update gas and gld versions.
3117 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3120 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3122 * cgraph.h (set_malloc_flag): Declare.
3123 * cgraph.c (set_malloc_flag_1): New function.
3124 (set_malloc_flag): Likewise.
3125 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3126 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3128 (read_ipa_call_summary): Add support for reading is_return_callee.
3129 (write_ipa_call_summary): Stream is_return_callee.
3130 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3131 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3132 ipa-prop.h, ipa-fnsummary.h.
3133 (pure_const_names): Change to static.
3134 (malloc_state_e): Define.
3135 (malloc_state_names): Define.
3136 (funct_state_d): Add field malloc_state.
3137 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3138 (check_retval_uses): New function.
3139 (malloc_candidate_p): Likewise.
3140 (analyze_function): Add support for malloc attribute.
3141 (pure_const_write_summary): Stream malloc_state.
3142 (pure_const_read_summary): Add support for reading malloc_state.
3143 (dump_malloc_lattice): New function.
3144 (propagate_malloc): New function.
3145 (warn_function_malloc): New function.
3146 (ipa_pure_const::execute): Call propagate_malloc and
3147 ipa_free_fn_summary.
3148 (pass_local_pure_const::execute): Add support for malloc attribute.
3149 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3150 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3152 2017-10-27 Martin Liska <mliska@suse.cz>
3154 PR gcov-profile/82457
3155 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3156 for fork-like functions to be properly instrumented.
3158 2017-10-27 Richard Biener <rguenther@suse.de>
3161 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3162 info when we redirected EH.
3164 2017-10-26 Michael Collison <michael.collison@arm.com>
3166 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3168 (<optab>_trunchf<GPI:mode>2: New pattern.
3169 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3170 * config/aarch64/iterators.md (wv): New mode attribute.
3171 (vf, VF): New mode attributes.
3172 (vgp, VGP): New mode attributes.
3173 (s): Update attribute with SImode and DImode prefixes.
3175 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3177 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3178 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3179 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3180 (nios2_option_overide): Initialize it. Don't allow R0-relative
3181 addressing with PIC.
3182 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3183 (nios2_symbolic_constant_p): Likewise.
3184 (nios2_legitimate_address_p): Likewise.
3185 (nios2_r0rel_section_name_p): New.
3186 (nios2_symbol_ref_in_r0rel_data_p): New.
3187 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3188 (r0rel_constant_p): New.
3189 (nios2_print_operand_address): Handle r0rel_constant_p.
3190 (nios2_cdx_narrow_form_p): Likewise.
3191 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3192 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3193 (Nios II Options): Document -mr0rel-sec.
3195 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3197 * config/nios2/nios2.c: Include xregex.h.
3198 (nios2_gprel_sec_regex): New.
3199 (nios2_option_overide): Initialize it. Don't allow GP-relative
3200 addressing with PIC.
3201 (nios2_small_section_name_p): Check for regex match.
3202 * config/nios2/nios2.opt (mgprel-sec=): New option.
3203 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3204 (Nios II Options): Document -mgprel-sec.
3206 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3208 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3210 2017-10-26 Tom de Vries <tom@codesourcery.com>
3212 PR tree-optimization/82707
3213 * gimple.c (gimple_copy): Fix unsharing of
3214 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3216 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3218 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3219 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3220 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3221 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3222 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3223 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3224 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3225 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3226 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3227 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3228 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3229 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3230 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3231 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3232 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3233 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3234 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3236 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3238 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3240 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3241 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3242 warning to rs6000.c. Remove the Undocumented flag, since it has
3244 (-mabi=ibmlongdouble): Likewise.
3245 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3246 already set, set the default format for long double.
3247 (rs6000_debug_reg_global): Print whether long double is IBM or
3249 (rs6000_option_override_internal): Rework setting long double
3250 format. Only warn if the user is changing the long double default
3251 and they did not use -Wno-psabi.
3252 * doc/invoke.texi (PowerPC options): Update the documentation for
3253 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3255 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3256 Alan Hayward <alan.hayward@arm.com>
3257 David Sherwood <david.sherwood@arm.com>
3259 * rtl.h (wider_subreg_mode): New function.
3260 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3261 rather than an unsigned int *.
3262 * ira-color.c (regno_max_ref_width): Replace with...
3263 (regno_max_ref_mode): ...this new variable.
3264 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3265 Use wider_subreg_mode.
3266 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3267 rather than an unsigned int *.
3268 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3269 (process_alt_operands): Likewise.
3270 (invariant_p): Likewise.
3271 * lra-spills.c (assign_mem_slot): Likewise.
3272 (add_pseudo_to_slot): Likewise.
3273 * lra.c (collect_non_operand_hard_regs): Likewise.
3274 (add_regs_to_insn_regno_info): Likewise.
3275 * reload1.c (regno_max_ref_width): Replace with...
3276 (regno_max_ref_mode): ...this new variable.
3277 (reload): Update accordingly. Update call to
3278 ira_sort_regnos_for_alter_reg.
3279 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3280 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3281 (scan_paradoxical_subregs): Likewise.
3283 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3285 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3286 (aarch64_frame): Add emit_frame_chain boolean.
3287 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3288 Move eh_return case to aarch64_layout_frame.
3289 (aarch64_layout_frame): Initialize emit_frame_chain.
3290 (aarch64_expand_prologue): Use emit_frame_chain.
3292 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3294 * config/aarch64/aarch64.c (aarch64_layout_frame):
3295 Ensure LR is always stored at the bottom of the callee-saves.
3296 Remove rarely used frame layout which saves callee-saves at top of
3297 frame, so the store of LR can be used as a valid probe in all cases.
3299 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3301 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3302 Improve unaligned TImode/TFmode base/offset split.
3304 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3305 Alan Hayward <alan.hayward@arm.com>
3306 David Sherwood <david.sherwood@arm.com>
3308 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3309 * combine.c (find_single_use_1): Likewise.
3310 (expand_field_assignment): Likewise.
3311 (move_deaths): Likewise.
3312 * lra-constraints.c (simplify_operand_subreg): Likewise.
3313 (curr_insn_transform): Likewise.
3314 * lra.c (collect_non_operand_hard_regs): Likewise.
3315 (add_regs_to_insn_regno_info): Likewise.
3316 * rtlanal.c (reg_referenced_p): Likewise.
3317 (covers_regno_no_parallel_p): Likewise.
3319 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3321 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3322 Don't print any bits outside the precision of the value.
3323 * wide-int.cc (test_printing): Add some new tests.
3325 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3327 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3328 supports -xbrace_comment option.
3329 * configure: Regenerate.
3330 * config.in: Regenerate.
3331 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3332 (ASM_CPU_SPEC): Use it.
3334 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3336 * target.def (static_rtx_alignment): New hook.
3337 * targhooks.h (default_static_rtx_alignment): Declare.
3338 * targhooks.c (default_static_rtx_alignment): New function.
3339 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3340 * doc/tm.texi: Regenerate.
3341 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3342 instead of targetm.constant_alignment. Remove call to
3344 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3345 (cris_preferred_mininum_alignment): New function, split out from...
3346 (cris_constant_alignment): ...here.
3347 (cris_static_rtx_alignment): New function.
3348 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3350 (ix86_constant_alignment): ...here.
3351 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3352 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3353 (mmix_static_rtx_alignment): New function.
3354 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3355 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3357 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3360 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3361 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3363 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3365 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3366 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3367 mark as a sign-extending load.
3368 (local_pic_load_u): Define.
3370 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3373 * fold-const.c (operand_equal_for_comparison_p): Also return true
3374 if ARG0 is a simple variant of ARG1 with narrower precision.
3375 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3377 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3379 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3381 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3382 scatter_static, scatter_per_elt.
3383 * x86-tune-costs.h: Add new cost entries.
3385 2017-10-25 Richard Biener <rguenther@suse.de>
3387 * tree-ssa-sccvn.h (vn_eliminate): Declare.
3388 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
3389 class pass_fre): Move to ...
3390 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
3391 class pass_fre): ... here and adjust for statistics.
3393 2017-10-25 Jakub Jelinek <jakub@redhat.com>
3396 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
3397 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
3398 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
3399 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
3402 2017-10-25 Richard Biener <rguenther@suse.de>
3404 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
3405 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
3406 eliminate_push_avail, eliminate_insert): Move inside...
3407 (class eliminate_dom_walker): ... this class in preparation
3409 (fini_eliminate): Remove by merging with ...
3410 (eliminate): ... this function. Adjust for class changes.
3411 (pass_pre::execute): Remove fini_eliminate call.
3412 (pass_fre::execute): Likewise.
3414 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3417 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
3418 (VPERMI2, VPERMI2I): New mode iterators.
3419 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
3420 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
3422 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
3423 mode iterator. Remove 3 old define_insn patterns.
3424 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
3425 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
3426 VPERMI2 mode iterator, remove the other two expanders.
3427 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
3428 to use VPERMI2 mode iterator, add another alternative for vpermi2*
3429 instructions, remove the other two patterns.
3430 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
3431 mode iterator, remove the other two patterns.
3432 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
3433 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
3434 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
3435 and adjust argument order accordingly.
3436 (ix86_expand_vec_perm): Adjust caller.
3437 (expand_vec_perm_1): Likewise.
3438 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
3439 (expand_vec_perm_vpermt2_vpshub2): ... this.
3440 (ix86_expand_vec_perm_const_1): Adjust caller.
3441 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
3444 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
3445 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
3446 (vec_shl_<mode>): Remove unused expander.
3447 (avx512bw_<shift_insn><mode>3): New define_insn.
3448 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
3449 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
3451 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
3454 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
3457 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3459 PR rtl-optimization/82396
3460 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
3461 (autopref_multipass_init): Simplify initialization.
3462 (autopref_rank_data): Simplify sort order.
3463 * gcc/sched-int.h (autopref_multipass_data_): Remove
3464 multi_mem_insn_p, min_offset and max_offset.
3466 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3469 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3470 Check special value of flag_omit_frame_pointer.
3471 (aarch64_can_eliminate): Likewise.
3472 (aarch64_override_options_after_change_1): Simplify handling of
3473 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
3475 2017-10-24 Richard Biener <rguenther@suse.de>
3477 PR tree-optimization/82697
3478 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
3479 zero for conditional load and unconditional store.
3481 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3483 * doc/install.texi: Document bootstrap-cet.
3485 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3488 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
3489 ENDBR instruction at function entrance if function is only
3492 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3495 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
3496 patterns to better describe from which operation the CF is computed.
3497 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
3498 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
3499 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
3500 is 0, use _0 suffixed expanders instead of emitting a comparison
3503 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
3505 * config/i386/i386.md(*movsf_internal, *movdf_internal):
3506 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
3508 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
3511 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
3512 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
3513 * loop-iv.c (iv_get_reaching_def): Likewise.
3514 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
3515 variable is promoted and the partition contains undefined values.
3517 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3519 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
3521 (nios2_address_cost): Define.
3522 (nios2_legitimize_address): Recognize (exp + constant) directly.
3523 (TARGET_ADDRESS_COST): Define.
3525 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3527 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
3528 (nios2_symbolic_memory_operand_p): Declare.
3529 (nios2_split_large_constant): Declare.
3530 (nios2_split_symbolic_memory_operand): Declare.
3531 * config/nios2/nios2.c: Adjust includes.
3532 (nios2_symbolic_constant_allowed): New.
3533 (nios2_symbolic_constant_p): New.
3534 (nios2_plus_symbolic_constant_p): New.
3535 (nios2_valid_addr_expr_p): Recognize addresses involving
3537 (nios2_legitimate_address_p): Likewise, also LO_SUM.
3538 (nios2_symbolic_memory_operand_p): New.
3539 (nios2_large_constant_p): New.
3540 (nios2_split_large_constant): New.
3541 (nios2_split_plus_large_constant): New.
3542 (nios2_split_symbolic_memory_operand): New.
3543 (nios2_legitimize_address): Code refactoring. Handle addresses
3544 involving symbolic constants.
3545 (nios2_emit_move_sequence): Likewise.
3546 (nios2_print_operand): Improve error output.
3547 (nios2_print_operand_address): Handle LO_SUM.
3548 (nios2_cdx_narrow_form_p): Likewise.
3549 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
3550 operands involving symbolic constants.
3551 (movhi_internal, movsi_internal): Likewise.
3552 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
3553 (extendhisi2, extendqi<mode>2): Likewise.
3555 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3557 * tree-pass.h (PROP_rtl_split_insns): Define.
3558 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
3560 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3562 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
3564 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3567 * target.def (const_not_ok_for_debug_p): Default to
3568 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
3569 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
3570 * targhooks.c (default_const_not_ok_for_debug_p): New function.
3571 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
3572 which targetm.const_not_ok_for_debug_p returned true.
3573 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
3575 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
3577 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
3578 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
3579 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
3581 (ix86_const_not_ok_for_debug_p): New function.
3582 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
3583 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
3585 2017-10-23 David Malcolm <dmalcolm@redhat.com>
3588 * system.h: Conditionally include "unique-ptr.h" if
3589 INCLUDE_UNIQUE_PTR is defined.
3590 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
3591 of defining INCLUDE_UNIQUE_PTR before including "system.h".
3593 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
3595 * config/rl78/rl78.md: New define_expand "subdi3".
3597 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
3600 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
3601 DF_REF_INSN if DF_REF_INSN_INFO is false.
3603 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
3605 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
3606 xmm_move instead of sse_move.
3607 (sse_store_index): New function.
3608 (ix86_register_move_cost): Be more sensible about mismatch stall;
3609 model AVX moves correctly; make difference between sse->integer and
3611 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
3612 moves; make difference between SSE and AVX.
3613 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
3614 and zmm_move. Increase size of sse load and store tables;
3615 add unaligned load and store tables; add ssemmx_to_integer.
3616 * x86-tune-costs.h: Update all entries according to real
3617 move latencies from Agner Fog's manual and chip documentation.
3619 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3622 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
3623 * config/i386/constraints.md (Wf): New constraint.
3624 * config/i386/i386.md (UNSPEC_SBB): New unspec.
3625 (cmp<dwi>_doubleword): Removed.
3626 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
3627 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
3628 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
3629 expand with cmp<dwi>_doubleword. For LTU and GEU use
3630 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
3632 * common.opt (gcolumn-info): Enable by default.
3633 * doc/invoke.texi (gcolumn-info): Document new default.
3635 2017-10-23 Richard Biener <rguenther@suse.de>
3637 PR tree-optimization/82672
3638 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
3639 Fold the stmt if we propagated into it.
3641 2017-10-23 Richard Biener <rguenther@suse.de>
3643 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
3644 (bitmap_remove_expr_from_set): ... this. All callers call this
3645 for non-constant values.
3646 (bitmap_set_subtract): Rename to...
3647 (bitmap_set_subtract_expressions): ... this. Adjust and
3649 (bitmap_set_contains_value): Remove superfluous check.
3650 (bitmap_set_replace_value): Inline into single caller ...
3651 (bitmap_value_replace_in_set): ... here and simplify.
3652 (dependent_clean): Merge into ...
3653 (clean): ... this using an overload. Adjust.
3654 (prune_clobbered_mems): Adjust.
3655 (compute_antic_aux): Likewise.
3656 (compute_partial_antic_aux): Likewise.
3658 2017-10-23 Richard Biener <rguenther@suse.de>
3660 PR tree-optimization/82129
3662 2017-08-01 Richard Biener <rguenther@suse.de>
3664 PR tree-optimization/81181
3665 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3666 (compute_antic): ... end of iteration here.
3668 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3670 * target.def (starting_frame_offset): New hook.
3671 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
3672 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
3673 * doc/tm.texi.in: Regenerate.
3674 * hooks.h (hook_hwi_void_0): Declare.
3675 * hooks.c (hook_hwi_void_0): New function.
3676 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
3677 STARTING_FRAME_OFFSET.
3678 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
3679 * reload1.c (reload): Likewise.
3680 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
3681 instead of STARTING_FRAME_OFFSET.
3682 * function.c (try_fit_stack_local): Likewise.
3683 (assign_stack_local_1): Likewise
3684 (instantiate_virtual_regs): Likewise.
3685 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
3686 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
3687 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
3688 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
3689 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
3690 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
3691 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
3692 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
3693 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
3694 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
3695 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
3696 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
3697 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
3698 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
3699 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
3700 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
3701 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
3702 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
3703 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
3704 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
3705 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
3706 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
3707 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
3708 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
3709 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
3710 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
3711 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
3712 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
3713 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
3714 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
3715 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
3716 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
3717 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
3718 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
3719 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
3720 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
3721 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
3722 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
3723 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
3724 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
3725 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
3726 * config/avr/avr.c (avr_starting_frame_offset): Make static and
3727 return a HOST_WIDE_INT.
3728 (avr_builtin_setjmp_frame_value): Use it instead of
3729 STARTING_FRAME_OFFSET.
3730 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3731 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
3732 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
3734 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3735 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
3736 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
3737 (TARGET_CONSTANT_ALIGNMENT): Redefine.
3738 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
3739 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
3740 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3741 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
3742 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
3743 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3744 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
3745 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
3747 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3748 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
3749 * config/mips/mips.c (mips_compute_frame_info): Refer to
3750 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3751 (mips_starting_frame_offset): New function.
3752 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3753 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
3754 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
3755 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
3756 and return a HOST_WIDE_INT.
3757 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3758 (mmix_initial_elimination_offset): Refer to
3759 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
3760 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
3761 * config/pa/pa.c (pa_starting_frame_offset): New function.
3762 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
3763 (pa_expand_prologue): Likewise.
3764 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3765 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
3766 !FRAME_GROWS_DOWNWARD handling to...
3767 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3768 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
3769 !FRAME_GROWS_DOWNWARD handling to...
3770 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3771 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
3772 !FRAME_GROWS_DOWNWARD handling to...
3773 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3774 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
3776 (rs6000_starting_frame_offset): New function.
3777 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
3778 !FRAME_GROWS_DOWNWARD handling to...
3779 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3780 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
3781 !FRAME_GROWS_DOWNWARD handling to...
3782 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3783 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
3784 !FRAME_GROWS_DOWNWARD handling to...
3785 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
3786 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
3787 (rs6000_starting_frame_offset): New function.
3788 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
3789 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
3790 * config/vax/vax.c (vax_starting_frame_offset): New function.
3791 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
3792 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3793 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
3794 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
3795 (TARGET_STARTING_FRAME_OFFSET): Redefine.
3796 * system.h (STARTING_FRAME_OFFSET): Poison.
3798 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3800 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
3801 SCALAR_TYPE_MODE instead of TYPE_MODE.
3803 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3804 Alan Hayward <alan.hayward@arm.com>
3805 David Sherwood <david.sherwood@arm.com>
3807 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
3809 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3810 Alan Hayward <alan.hayward@arm.com>
3811 David Sherwood <david.sherwood@arm.com>
3813 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
3815 2017-10-23 Richard Biener <rguenther@suse.de>
3817 PR tree-optimization/82129
3818 * tree-ssa-pre.c (bitmap_set_and): Remove.
3819 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
3820 canonicalizing expressions in the set to those with lowest
3821 ID rather than taking that from the first edge.
3823 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3825 * combine.c (rtx_equal_for_field_assignment_p): Use
3826 byte_lowpart_offset.
3828 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3829 Alan Hayward <alan.hayward@arm.com>
3830 David Sherwood <david.sherwood@arm.com>
3832 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3833 to a SUBREG_PROMOTED_VAR.
3835 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3836 Alan Hayward <alan.hayward@arm.com>
3837 David Sherwood <david.sherwood@arm.com>
3839 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3840 (expand_debug_source_expr): Likewise.
3841 * combine.c (combine_simplify_rtx): Likewise.
3842 * cse.c (fold_rtx): Likewise.
3843 * optabs.c (expand_float): Likewise.
3844 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3845 (simplify_binary_operation_1): Likewise.
3847 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3848 Alan Hayward <alan.hayward@arm.com>
3849 David Sherwood <david.sherwood@arm.com>
3851 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3852 (record_promoted_value): Likewise.
3853 * expr.c (expand_expr_real_2): Likewise.
3854 * ree.c (update_reg_equal_equiv_notes): Likewise.
3855 (combine_set_extension): Likewise.
3856 * rtlanal.c (low_bitmask_len): Likewise.
3857 * simplify-rtx.c (neg_const_int): Likewise.
3858 (simplify_binary_operation_1): Likewise.
3860 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3861 Alan Hayward <alan.hayward@arm.com>
3862 David Sherwood <david.sherwood@arm.com>
3864 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3865 * regcprop.c (maybe_mode_change): Likewise.
3866 * reload1.c (alter_reg): Likewise.
3868 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3870 * inchash.h (inchash::hash::add_wide_int): New function.
3871 * lto-streamer-out.c (hash_tree): Use it.
3873 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3875 * inchash.h (inchash::hash::add_wide_int): Rename to...
3876 (inchash::hash::add_hwi): ...this.
3877 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3878 (polymorphic_call_target_hasher::hash): Likewise.
3879 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3880 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3881 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3882 * lto-streamer-out.c (hash_tree): Likewise.
3883 * optc-save-gen.awk: Likewise.
3884 * tree.c (add_expr): Likewise.
3886 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3889 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
3890 for ordered inequality comparisons even with TARGET_IEEE_FP.
3892 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3895 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
3896 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3897 Expand with cmp<dwi>_doubleword.
3899 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3901 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3902 List CET intrinsics.
3903 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3904 specific to -fcf-protection option.
3906 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3908 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3909 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3910 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3911 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3912 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3913 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3914 (extra_objs): Add cet.o for Linux/x86 targets.
3915 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3916 * config/i386/cet.c: New file.
3917 * config/i386/cetintrin.h: Likewise.
3918 * config/i386/t-cet: Likewise.
3919 * config/i386/cpuid.h (bit_SHSTK): New.
3920 (bit_IBT): Likewise.
3921 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3922 pass IBT and SHSTK bits.
3923 * config/i386/i386-builtin-types.def
3924 (VOID_FTYPE_UNSIGNED_PVOID): New.
3925 (VOID_FTYPE_UINT64_PVOID): Likewise.
3926 * config/i386/i386-builtin.def: Add CET intrinsics.
3927 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3928 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3929 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3930 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3932 * config/i386/i386.c (rest_of_insert_endbranch): New.
3933 (pass_data_insert_endbranch): Likewise.
3934 (pass_insert_endbranch): Likewise.
3935 (make_pass_insert_endbranch): Likewise.
3936 (ix86_notrack_prefixed_insn_p): Likewise.
3937 (ix86_target_string): Add -mibt, -mshstk flags.
3938 (ix86_option_override_internal): Add flag_cf_protection
3940 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
3941 (ix86_print_operand): Add 'notrack' prefix output.
3942 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
3943 (ix86_expand_builtin): Expand CET intrinsics.
3944 (x86_output_mi_thunk): Add 'endbranch' instruction.
3945 * config/i386/i386.h (TARGET_IBT): New.
3946 (TARGET_IBT_P): Likewise.
3947 (TARGET_SHSTK): Likewise.
3948 (TARGET_SHSTK_P): Likewise.
3949 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
3950 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
3951 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
3952 (builtin_setjmp_setup): New pattern.
3953 (builtin_longjmp): Likewise.
3954 (rdssp<mode>): Likewise.
3955 (incssp<mode>): Likewise.
3956 (saveprevssp): Likewise.
3957 (rstorssp): Likewise.
3958 (wrss<mode>): Likewise.
3959 (wruss<mode>): Likewise.
3960 (setssbsy): Likewise.
3961 (clrssbsy): Likewise.
3962 (nop_endbr): Likewise.
3963 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
3965 * config/i386/immintrin.h: Include <cetintrin.h>.
3966 * config/i386/linux-common.h
3967 (file_end_indicate_exec_stack_and_cet): New prototype.
3968 (TARGET_ASM_FILE_END): New.
3970 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3972 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
3973 latencies instead of having separate table; make difference between
3974 integer and float costs.
3975 * i386.h (processor_costs): Remove scalar_stmt_cost,
3976 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
3977 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
3979 * x86-tune-costs.h: Remove entries which has been removed in
3980 procesor_costs from all tables; make cond_taken_branch_cost
3981 and cond_not_taken_branch_cost COST_N_INSNS based.
3983 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3985 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
3987 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3989 * config/i386/i386.md (isa): Remove fma_avx512f.
3990 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
3991 <avx512>_fmadd_<mode>_mask3<round_name>,
3992 <avx512>_fmsub_<mode>_mask<round_name>,
3993 <avx512>_fmsub_<mode>_mask3<round_name>,
3994 <avx512>_fnmadd_<mode>_mask<round_name>,
3995 <avx512>_fnmadd_<mode>_mask3<round_name>,
3996 <avx512>_fnmsub_<mode>_mask<round_name>,
3997 <avx512>_fnmsub_<mode>_mask3<round_name>,
3998 <avx512>_fmaddsub_<mode>_mask<round_name>,
3999 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4000 <avx512>_fmsubadd_<mode>_mask<round_name>,
4001 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4002 (*vec_widen_umult_even_v16si<mask_name>,
4003 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4004 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4006 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4008 * extend.texi: Add 'nocf_check' documentation.
4009 * gimple.texi: Add second parameter to
4010 gimple_build_call_from_tree.
4011 * invoke.texi: Add -fcf-protection documentation.
4012 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4014 2017-10-20 Richard Biener <rguenther@suse.de>
4016 PR tree-optimization/82473
4017 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4018 the largest input type.
4020 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4022 * c-attribs.c (handle_nocf_check_attribute): New function.
4023 (c_common_attribute_table): Add 'nocf_check' handling.
4024 * gimple-parser.c: Add second argument NULL to
4025 gimple_build_call_from_tree.
4026 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4027 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4029 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4030 * common.opt: Add fcf-protection flag.
4031 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4032 * flag-types.h: Add enum cf_protection_level.
4033 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4034 Add 'nocf_check' attribute propagation to gimple call.
4035 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4036 (gimple_build_call_from_tree): Update prototype.
4037 (gimple_call_nocf_check_p): New function.
4038 (gimple_call_set_nocf_check): Likewise.
4039 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4040 * ipa-icf.c: Add nocf_check attribute in statement hash.
4041 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4042 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4043 * toplev.c (process_options): Add flag_cf_protection handling.
4045 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4047 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4049 2017-10-20 Richard Biener <rguenther@suse.de>
4051 PR tree-optimization/82603
4052 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4053 remove false predicated stores.
4055 2017-10-20 Richard Biener <rguenther@suse.de>
4057 * graphite-isl-ast-to-gimple.c
4058 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4059 Remove return value and simplify, dump copied stmt after lhs
4061 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4062 Reduce dump verbosity.
4063 (gsi_insert_earliest): Likewise.
4064 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4065 * graphite.c (print_global_statistics): Adjust dumping.
4066 (print_graphite_scop_statistics): Likewise.
4067 (print_graphite_statistics): Do not dump loops here.
4068 (graphite_transform_loops): But here.
4070 2017-10-20 Nicolas Roche <roche@adacore.com>
4072 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4073 * configure: Regenerate.
4075 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4078 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4079 functions when optimizing replace GIMPLE_RETURN stmts with
4080 calls to __builtin_unreachable ().
4083 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4084 for -fsanitize=thread link of executables.
4085 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4086 link of executables.
4089 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4091 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4092 define_insns for logical vector shifts to use VI248_AVX512BW
4093 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4094 condition, useless isa and prefix attributes. Change the first
4095 2 of these define_insns to ...
4096 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4097 define_insn for avx512vl.
4098 (<shift_insn><mode>3): ... and this, new define_insn without
4099 masking for non-avx512vl.
4102 * config/i386/sse.md (*andnot<mode>3,
4103 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4104 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4105 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4106 not applied use empty suffix even for TARGET_AVX512VL.
4107 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4108 is applied, supply evex,evex or evex,evex,evex instead of just
4111 2017-10-20 Julia Koval <julia.koval@intel.com>
4113 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4114 (OPTION_MASK_ISA_GFNI_UNSET): New.
4115 (ix86_handle_option): Handle OPT_mgfni.
4116 * config/i386/cpuid.h (bit_GFNI): New.
4117 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4118 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4119 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4120 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4121 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4122 * config/i386/i386.opt: Add mgfni.
4124 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4126 * config/msp430/msp430.c (msp430_option_override): Disable
4127 -fdelete-null-pointer-checks.
4128 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4130 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4132 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4133 of x87 and SSE instructions.
4135 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4137 * asan.c (create_cond_insert_point): Do not update edge count.
4138 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4139 (afdo_propagate_circuit): Likewise.
4140 (afdo_calculate_branch_prob): Likewise.
4141 (afdo_annotate_cfg): Likewise.
4142 * basic-block.h (struct edge_def): Remove count.
4143 (edge_def::count): New accessor.
4144 * bb-reorder.c (rotate_loop): Update.
4145 (find_traces_1_round): Update.
4146 (connect_traces): Update.
4147 (sanitize_hot_paths): Update.
4148 * cfg.c (unchecked_make_edge): Update.
4149 (make_single_succ_edge): Update.
4150 (check_bb_profile): Update.
4151 (dump_edge_info): Update.
4152 (update_bb_profile_for_threading): Update.
4153 (scale_bbs_frequencies_int): Update.
4154 (scale_bbs_frequencies_gcov_type): Update.
4155 (scale_bbs_frequencies_profile_count): Update.
4156 (scale_bbs_frequencies): Update.
4157 * cfganal.c (connect_infinite_loops_to_exit): Update.
4158 * cfgbuild.c (compute_outgoing_frequencies): Update.
4159 (find_many_sub_basic_blocks): Update.
4160 * cfgcleanup.c (try_forward_edges): Update.
4161 (try_crossjump_to_edge): Update
4162 * cfgexpand.c (expand_gimple_cond): Update
4163 (expand_gimple_tailcall): Update
4164 (construct_exit_block): Update
4165 * cfghooks.c (verify_flow_info): Update
4166 (redirect_edge_succ_nodup): Update
4167 (split_edge): Update
4168 (make_forwarder_block): Update
4169 (duplicate_block): Update
4170 (account_profile_record): Update
4171 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4172 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4173 * cfgloopmanip.c (scale_loop_profile): Update.
4175 (lv_adjust_loop_entry_edge): Update.
4176 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4177 (force_nonfallthru_and_redirect): Update.
4178 (purge_dead_edges): Update.
4179 (rtl_flow_call_edges_add): Update.
4180 * cgraphunit.c (init_lowered_empty_function): Update.
4181 (cgraph_node::expand_thunk): Update.
4182 * gimple-pretty-print.c (dump_probability): Update.
4183 (dump_edge_probability): Update.
4184 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4185 * haifa-sched.c (sched_create_recovery_edges): Update.
4186 * hsa-gen.c (convert_switch_statements): Update.
4187 * ifcvt.c (dead_or_predicable): Update.
4188 * ipa-inline-transform.c (inline_transform): Update.
4189 * ipa-split.c (split_function): Update.
4190 * ipa-utils.c (ipa_merge_profiles): Update.
4191 * loop-doloop.c (add_test): Update.
4192 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4193 * lto-streamer-in.c (input_cfg): Update.
4194 (input_function): Update.
4195 * lto-streamer-out.c (output_cfg): Update.
4196 * modulo-sched.c (sms_schedule): Update.
4197 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4198 * predict.c (maybe_hot_edge_p): Update.
4199 (unlikely_executed_edge_p): Update.
4200 (probably_never_executed_edge_p): Update.
4201 (dump_prediction): Update.
4202 (drop_profile): Update.
4203 (propagate_unlikely_bbs_forward): Update.
4204 (determine_unlikely_bbs): Update.
4205 (force_edge_cold): Update.
4206 * profile.c (compute_branch_probabilities): Update.
4207 * reg-stack.c (better_edge): Update.
4208 * shrink-wrap.c (handle_simple_exit): Update.
4209 * tracer.c (better_p): Update.
4210 * trans-mem.c (expand_transaction): Update.
4211 (split_bb_make_tm_edge): Update.
4212 * tree-call-cdce.c: Update.
4213 * tree-cfg.c (gimple_find_sub_bbs): Update.
4214 (gimple_split_edge): Update.
4215 (gimple_duplicate_sese_region): Update.
4216 (gimple_duplicate_sese_tail): Update.
4217 (gimple_flow_call_edges_add): Update.
4218 (insert_cond_bb): Update.
4219 (execute_fixup_cfg): Update.
4220 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4221 * tree-complex.c (expand_complex_div_wide): Update.
4222 * tree-eh.c (lower_resx): Update.
4223 (unsplit_eh): Update.
4224 (cleanup_empty_eh_move_lp): Update.
4225 * tree-inline.c (copy_edges_for_bb): Update.
4226 (freqs_to_counts): Update.
4227 (copy_cfg_body): Update.
4228 * tree-ssa-dce.c (remove_dead_stmt): Update.
4229 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4230 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4231 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4232 (unloop_loops): Update.
4233 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4234 * tree-ssa-loop-split.c (connect_loops): Update.
4235 (split_loop): Update.
4236 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4237 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4238 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4239 * tree-ssa-reassoc.c (branch_fixup): Update.
4240 * tree-ssa-tail-merge.c (replace_block_by): Update.
4241 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4242 (compute_path_counts): Update.
4243 (update_profile): Update.
4244 (recompute_probabilities): Update.
4245 (update_joiner_offpath_counts): Update.
4246 (estimated_freqs_path): Update.
4247 (freqs_to_counts_path): Update.
4248 (clear_counts_path): Update.
4249 (ssa_fix_duplicate_block_edges): Update.
4250 (duplicate_thread_path): Update.
4251 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4252 (case_bit_test_cmp): Update.
4253 (collect_switch_conv_info): Update.
4254 (gen_inbound_check): Update.
4255 (do_jump_if_equal): Update.
4256 (emit_cmp_and_jump_insns): Update.
4257 * tree-tailcall.c (decrease_profile): Update.
4258 (eliminate_tail_call): Update.
4259 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4260 (vect_do_peeling): Update.
4261 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4262 * ubsan.c (ubsan_expand_null_ifn): Update.
4263 (ubsan_expand_ptr_ifn): Update.
4264 * value-prof.c (gimple_divmod_fixed_value): Update.
4265 (gimple_mod_pow2): Update.
4266 (gimple_mod_subtract): Update.
4267 (gimple_ic): Update.
4268 (gimple_stringop_fixed_value): Update.
4270 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4273 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4275 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4277 PR rtl-optimization/82395
4278 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4279 based on non_spilled_static_chain_regno_p.
4281 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4283 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4284 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4285 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4287 2017-10-19 Martin Sebor <msebor@redhat.com>
4289 PR tree-optimization/82596
4290 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4292 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4294 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4295 (get_mem_refs_of_builtin_call): Likewise.
4296 * builtins.c (expand_builtin_apply): Adjust call to
4297 allocate_dynamic_stack_space.
4298 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4299 the third argument to allocate_dynamic_stack_space, otherwise -1.
4300 (expand_builtin): Deal with all alloca variants.
4301 (is_inexpensive_builtin): Likewise.
4302 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4303 * calls.c (special_function_p): Deal with all alloca variants.
4304 (initialize_argument_information): Adjust call to
4305 allocate_dynamic_stack_space.
4306 (expand_call): Likewise.
4307 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4308 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4309 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4310 use it for the stack usage computation.
4311 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4312 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4313 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4314 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4315 (in_loop_p): Remove first argument and useless check.
4316 (pass_walloca::execute): Remove useless test and adjust call to above.
4317 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4318 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4319 (gimplify_call_expr): Deal with all alloca variants.
4320 * hsa-gen.c (gen_hsa_alloca): Likewise.
4321 (gen_hsa_insns_for_call): Likewise.
4322 * ipa-pure-const.c (special_builtin_state): Likewise.
4323 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4324 * tree-object-size.c (alloc_object_size): Likewise.
4325 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4326 (call_may_clobber_ref_p_1): Likewise.
4327 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4328 (ccp_fold_stmt): Likewise.
4329 (optimize_stack_restore): Likewise.
4330 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4331 (mark_all_reaching_defs_necessary_1): Likewise.
4332 (propagate_necessity): Likewise.
4333 (eliminate_unnecessary_stmts): Likewise.
4334 * tree.c (build_common_builtin_nodes): Build
4335 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4336 (build_alloca_call_expr): New function.
4337 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4338 (CASE_BUILT_IN_ALLOCA): Likewise.
4339 (build_alloca_call_expr): Declare.
4340 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4342 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4345 * dwarf2out.c (new_die_raw): New static inline function.
4346 (new_die): Use it to create the DIE.
4347 (add_AT_external_die_ref): Likewise.
4348 (clone_die): Likewise.
4349 (clone_as_declaration): Likewise.
4350 (dwarf2out_vms_debug_main_pointer): Likewise.
4351 (base_type_die): Likewise. Remove early return for corner cases.
4352 Do not call add_pubtype on the DIE here.
4353 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4354 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4355 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4356 native order exists for base types, attach the DIE manually and call
4357 add_pubtype on it. Do not equate a reverse order DIE to the type.
4359 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4361 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4362 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4364 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4366 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4368 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4369 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4370 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4371 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4372 * opts.c (sanitizer_opts): Add builtin.
4373 * ubsan.c (instrument_builtin): New function.
4374 (pass_ubsan::execute): Call it.
4375 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4376 * doc/invoke.texi: Document -fsanitize=builtin.
4378 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4379 builtins, store max (log2 (align), 0) into uchar field instead of
4380 align into uptr field.
4381 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4382 store uchar 0 field instead of uptr 0 field.
4383 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
4384 instead of passing one address of struct with 2 locations pass
4385 two addresses of structs with 1 location each.
4386 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
4387 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4388 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
4389 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
4390 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
4391 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
4392 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
4393 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
4395 2017-10-19 Martin Liska <mliska@suse.cz>
4398 * file-find.c (remove_prefix): Remove.
4399 * file-find.h (remove_prefix): Likewise.
4400 * gcc-ar.c: Remove smartness of lookup.
4402 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4404 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
4405 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
4406 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
4408 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4411 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
4412 (setcc + and to xor + setcc): New peephole2.
4414 2017-10-19 Tom de Vries <tom@codesourcery.com>
4416 * doc/sourcebuild.texi (Test Directives, Variants of
4417 dg-require-support): Add dg-require-stack-size.
4419 2017-10-19 Martin Liska <mliska@suse.cz>
4422 * gimplify.c (gimplify_decl_expr): Do not instrument variables
4423 that have a large alignment.
4424 (gimplify_target_expr): Likewise.
4426 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
4428 PR rtl-optimization/82602
4429 * ira.c (rtx_moveable_p): Return false for volatile asm.
4431 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
4434 * config/i386/i386-modes.def (CCGZ): New CC mode.
4435 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
4436 * config/i386/predicates.md (ix86_comparison_operator):
4438 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4439 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
4440 with double-word subtraction.
4441 (put_condition_code): Handle CCGZmode.
4443 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
4445 * wide-int.cc (debug (const wide_int &)): New.
4446 (debug (const wide_int *)): New.
4447 (debug (const widest_int &)): New.
4448 (debug (const widest_int *)): New.
4450 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
4453 * lra-constraints.c (curr_insn_transform): Use non-input operand
4454 instead of output one for matched reload.
4456 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4458 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
4459 (tree-ssa-loop-ivopts.h): New header file.
4460 (struct builtin_info): New fields.
4461 (classify_builtin_1): Compute and record base and offset parts for
4462 memset builtin partition by calling strip_offset.
4463 (offset_cmp, fuse_memset_builtins): New functions.
4464 (finalize_partitions): Fuse adjacent memset partitions by calling
4466 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
4467 Expose the interface.
4468 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
4470 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4472 PR tree-optimization/82574
4473 * tree-loop-distribution.c (find_single_drs): New parameter. Check
4474 that data reference must be executed exactly once per iteration
4475 against the outermost loop in nest.
4476 (classify_partition): Update call to above function.
4478 2017-10-18 Richard Biener <rguenther@suse.de>
4480 PR tree-optimization/82591
4481 * graphite.c (graphite_transform_loops): Move code gen message
4483 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4484 Here. Handle scop_to_isl_ast failing.
4485 (scop_to_isl_ast): Limit the number of ISL operations.
4487 2017-10-18 Richard Biener <rguenther@suse.de>
4489 * graphite-isl-ast-to-gimple.c
4490 (translate_isl_ast_to_gimple::set_rename): Simplify.
4491 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
4492 (graphite_copy_stmts_from_block): ... here.
4493 (copy_bb_and_scalar_dependences): Simplify.
4494 (add_parameters_to_ivs_params): Canonicalize.
4495 (generate_entry_out_of_ssa_copies): Simplify.
4496 * graphite-sese-to-poly.c (extract_affine_name): Simplify
4497 by passing in ISL dimension.
4498 (parameter_index_in_region_1): Rename to ...
4499 (parameter_index_in_region): ... this.
4500 (extract_affine): Adjust assert, pass down parameter index.
4501 (add_param_constraints): Use range-info when available.
4502 (build_scop_context): Adjust.
4503 * sese.c (new_sese_info): Adjust.
4504 (free_sese_info): Likewise.
4505 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
4506 Remove unused typedefs.
4507 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
4509 2017-10-18 Martin Liska <mliska@suse.cz>
4511 * combine.c (simplify_compare_const): Add gcc_fallthrough.
4513 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4515 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
4516 (s390_sched_init): Do not reset s390_sched_state if we entered the
4517 current basic block via a fallthru edge and all others are unlikely.
4519 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4521 * config/s390/s390.c (NUM_SIDES): New variable.
4522 (LONGRUNNING_THRESHOLD): New variable.
4523 (LATENCY_FACTOR): New variable.
4524 (s390_sched_score): Decrease score for long-running instructions on
4526 (s390_sched_variable_issue): Perform bookkeeping for long-running
4529 2017-10-18 Richard Biener <rguenther@suse.de>
4531 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
4532 Simplify with removal of the parameter rename map.
4533 (set_rename): Likewise.
4534 (should_copy_to_new_region): Likewise.
4535 (graphite_copy_stmts_from_block): Likewise.
4536 (copy_bb_and_scalar_dependences): Remove initialization of
4537 unused copied_bb_map.
4539 (copy_internal_parameters): Likewise.
4540 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
4541 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
4542 Use INTEGRAL_TYPE_P.
4543 (parameter_index_in_region_1): Rename to ...
4544 (assign_parameter_index_in_region): ... this. Assert we have
4545 a parameter we handle.
4546 (scan_tree_for_params): Adjust.
4547 * sese.h (parameter_rename_map_t): Remove.
4548 (struct sese_info_t): Remove unused parameter_rename_map and
4549 copied_bb_map members.
4550 * sese.c (new_sese_info): Adjust.
4551 (free_sese_info): Likewise.
4553 2017-10-18 Martin Liska <mliska@suse.cz>
4556 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
4557 on an abnormal edge.
4559 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
4561 * doc/invoke.texi (ffunction-sections and fdata-sections):
4564 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4566 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
4567 the use statement can throw internally.
4569 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4571 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
4572 any RTX present on the RHS of a SET.
4573 * compare-elim.c (try_eliminate_compare): Restore comment.
4575 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4577 * langhooks.h (struct lang_hooks): Document that tree_size langhook
4578 may be also called on tcc_type nodes.
4579 * langhooks.c (lhd_tree_size): Likewise.
4581 2017-10-17 David Malcolm <dmalcolm@redhat.com>
4583 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
4584 format_warning_at_substring.
4585 (maybe_warn): Convert source_range * param to a location_t. Pass
4586 UNKNOWN_LOCATION rather than NULL to fmtwarn.
4587 (format_directive): Remove code to extract source_ranges and
4588 source_range * in favor of just a location_t.
4589 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
4591 * substring-locations.c (format_warning_va): Convert
4592 source_range * param to a location_t.
4593 (format_warning_at_substring): Likewise.
4594 * substring-locations.h (format_warning_va): Likewise.
4595 (format_warning_at_substring): Likewise.
4597 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
4599 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
4601 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
4602 and scatter/gather ops.
4604 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
4605 vec_gather_load and vec_scatter_store.
4606 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
4607 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
4608 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
4609 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
4610 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
4611 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
4613 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
4615 * reg-stack.c (compare_for_stack_reg): Add bool argument.
4616 Detect FTST instruction and handle its register pops. Only pop
4617 second operand if can_pop_second_op is true.
4618 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
4619 set can_pop_second_op to false in the compare_for_stack_reg call.
4621 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
4622 output_fp_compare for stack register operands.
4623 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
4624 instructions here. Do not emit stack register pops here. Assert
4625 that FCOMPP pops next to top stack register. Rewrite function.
4627 2017-10-17 Nathan Sidwell <nathan@acm.org>
4630 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
4631 use DECL_ASSEMBLER_NAME_RAW.
4634 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
4637 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
4638 Wilco Dijkstra <wilco.dijkstra@arm.com>
4640 * builtins.c (expand_builtin_update_setjmp_buf): Add a
4641 converstion to Pmode from the buf_addr.
4643 2017-10-17 Richard Biener <rguenther@suse.de>
4645 * graphite-dependences.c (scop_get_reads_and_writes): Change
4646 output parameters to references.
4648 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
4650 PR 71026/tree-optimization
4651 * fold-const.c (distribute_real_division): Removed.
4652 (fold_binary_loc): Remove calls to distribute_real_divison.
4654 2017-10-17 Richard Biener <rguenther@suse.de>
4656 * graphite-scop-detection.c
4657 (scop_detection::stmt_has_simple_data_refs_p): Always use
4658 the full nest as region.
4659 (try_generate_gimple_bb): Likewise.
4660 * sese.c (scalar_evolution_in_region): Simplify now that
4661 SCEV can handle instantiation in regions.
4662 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
4663 in the non-loop part of a function if requested.
4665 2017-10-17 Richard Biener <rguenther@suse.de>
4667 PR tree-optimization/82563
4668 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
4670 (graphite_regenerate_ast_isl): Call it.
4671 * graphite-scop-detection.c (build_scops): Remove entry edge split.
4673 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4675 PR tree-optimization/82549
4676 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
4677 Formatting fixes. Instead of calling make_bit_field_ref with negative
4680 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
4682 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
4683 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
4684 _mm_maskz_reduce_ss): New.
4685 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
4686 __builtin_ia32_reducess_mask): Ditto..
4687 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
4688 * config/i386/sse.md (reduces<mode>): Renamed to ...
4689 (reduces<mode><mask_scalar_name>): ... this.
4690 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
4692 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
4693 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
4695 2017-10-16 David Malcolm <dmalcolm@redhat.com>
4697 * Makefile.in (OBJS): Add unique-ptr-tests.o.
4698 * selftest-run-tests.c (selftest::run_tests): Call
4699 selftest::unique_ptr_tests_cc_tests.
4700 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
4701 * unique-ptr-tests.cc: New file.
4703 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
4706 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4708 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4709 (make_hard_regno_born, make_hard_regno_dead): Update
4710 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
4712 2017-10-16 Jeff Law <law@redhat.com>
4714 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4716 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
4718 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
4720 2017-10-16 Olivier Hainque <hainque@adacore.com>
4722 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
4723 with_cpu if we were configured for an e500v2 target cpu name.
4725 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
4727 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
4728 * doc/invoke.texi: Document +nodsp as a valid extension for
4731 2017-10-16 Martin Liska <mliska@suse.cz>
4733 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
4734 (test_set_range): Likewise.
4735 (test_range_functions): Rename to ...
4736 (test_bit_in_range): ... this.
4737 (sbitmap_c_tests): Add new test.
4739 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4741 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
4743 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
4744 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
4746 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4748 * config/aarch64/aarch64-builtins.c
4749 (aarch64_types_quadopu_lane_qualifiers): New.
4750 (TYPES_QUADOPU_LANE): New.
4751 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
4752 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
4753 (aarch64_<sur>dot_laneq<vsi2qi>): New.
4754 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
4755 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
4756 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
4757 (Vdottype, DOTPROD): New.
4758 (sur): Add SDOT and UDOT.
4760 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4762 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
4763 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
4764 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
4766 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
4767 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
4768 Enable TARGET_DOTPROD.
4769 (cortex-a75.cortex-a55): Likewise.
4770 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
4772 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4774 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
4775 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
4777 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
4779 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
4780 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
4781 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
4782 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
4783 * config/arm/types.md (neon_dot, neon_dot_q): New.
4784 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
4786 2017-10-16 Tamar Christina <tamar.christina@arm.com>
4788 * config/arm/arm.h (TARGET_DOTPROD): New.
4789 * config/arm/arm.c (arm_arch_dotprod): New.
4790 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
4791 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
4792 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
4793 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
4794 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
4795 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
4796 * doc/invoke.texi (armv8.2-a): Document dotprod
4798 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4800 * i386.c (ix86_vec_cost): New function.
4801 (ix86_rtx_costs): Handle vector operations better.
4802 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
4803 * x86-tune-costs.h: Add new costs to all tables.
4805 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4807 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
4809 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
4810 divsd, sqrtss and sqrtsd
4811 * x86-tune-costs.h: Add new entries to all costs.
4812 (znver1_cost): Fix to match real instruction latencies.
4814 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4815 Michael Collison <michael.collison@arm.com>
4817 * compare-elim.c: Include emit-rtl.h.
4818 (can_merge_compare_into_arith): New function.
4819 (try_validate_parallel): Likewise.
4820 (try_merge_compare): Likewise.
4821 (try_eliminate_compare): Call the above when no previous clobber
4823 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
4826 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4830 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
4831 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4835 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4836 to be any operand_equal_p operands. For & (B - 1) require
4837 B to be power of 2. Recognize
4838 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4840 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4843 * optabs.c (expand_memory_blockage): Fix call of
4844 targetm.have_memory_blockage.
4846 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4849 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4850 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4851 objects to extra_objs instead of overwriting it.
4853 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4855 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4856 Use any_fp_register_operand as operand[3] predicate. Simplify
4857 equality test for operands[2] and operands[4] memory location.
4858 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4859 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4860 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4861 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4862 any_fp_register_operand as operand[1] predicate. Simplify
4863 equality test for operands[0] and operands[3] memory location.
4864 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4865 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4866 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4868 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4870 * target-insns.def: Add memory_blockage.
4871 * optabs.c (expand_memory_blockage): New function.
4872 (expand_asm_memory_barrier): Rename ...
4873 (expand_asm_memory_blockage): ... to this.
4874 (expand_mem_thread_fence): Call expand_memory_blockage
4875 instead of expand_asm_memory_barrier.
4876 (expand_mem_singnal_fence): Ditto.
4877 (expand_atomic_load): Ditto.
4878 (expand_atomic_store): Ditto.
4879 * doc/md.texi (Standard Pattern Names For Generation):
4880 Document memory_blockage instruction pattern.
4882 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4884 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4885 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4886 * config/rl78/rl78.md: New define_expand "adddi3".
4888 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4890 * cfghooks.c (verify_flow_info): Disable check that all probabilities
4893 2017-10-13 Jeff Law <law@redhat.com>
4895 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
4897 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4900 * internal-fn.c (expand_mul_overflow): If both operands have
4901 the same highpart of -1 or 0 and the topmost bit of lowpart
4902 is different, overflow is if res <= 0 rather than res < 0.
4904 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4906 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4907 TARGET_P9_VECTOR code for unaligned_load case.
4909 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4911 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4913 2017-10-13 Nathan Sidwell <nathan@acm.org>
4915 * tree-core.h (tree_contains_struct): Make bool.
4916 * tree.c (tree_contains_struct): Likewise.
4917 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4918 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4919 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4920 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4921 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4923 2017-10-13 Richard Biener <rguenther@suse.de>
4925 * graphite-isl-ast-to-gimple.c
4926 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4927 parameters and dominance check.
4928 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4929 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4930 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4931 Do not update SSA form here or do intermediate IL verification.
4932 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4933 (graphite_initialize): Remove check on the number of loops in
4934 the function and inline into graphite_transform_loops.
4935 (graphite_finalize): Inline into graphite_transform_loops.
4936 (graphite_transform_loops): Perform SSA update and IL verification
4938 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
4940 2017-10-13 Richard Biener <rguenther@suse.de>
4942 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
4943 graphite_expression_type_precision): Avoid global constructor
4945 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
4946 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
4947 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
4948 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
4949 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
4950 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
4952 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
4955 * config/i386/i386.h (ix86_red_zone_size): New.
4956 * config/i386/i386.md (push peephole2s): Replace
4957 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
4959 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4960 Alan Hayward <alan.hayward@arm.com>
4961 David Sherwood <david.sherwood@arm.com>
4963 * combine.c (can_change_dest_mode): Reject changes in
4964 REGMODE_NATURAL_SIZE.
4966 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4967 Alan Hayward <alan.hayward@arm.com>
4968 David Sherwood <david.sherwood@arm.com>
4970 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
4971 (expand_debug_source_expr): Likewise.
4972 * combine.c (combine_simplify_rtx): Likewise.
4973 * cse.c (fold_rtx): Likewise.
4974 * fwprop.c (canonicalize_address): Likewise.
4975 * targhooks.c (default_shift_truncation_mask): Likewise.
4977 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4978 Alan Hayward <alan.hayward@arm.com>
4979 David Sherwood <david.sherwood@arm.com>
4981 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4982 (widened_mode): Likewise.
4983 (expand_unop): Likewise.
4984 * ree.c (transform_ifelse): Likewise.
4985 (merge_def_and_ext): Likewise.
4986 (combine_reaching_defs): Likewise.
4987 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4989 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4990 Alan Hayward <alan.hayward@arm.com>
4991 David Sherwood <david.sherwood@arm.com>
4993 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
4994 * combine.c (gen_lowpart_for_combine): Likewise.
4995 * dwarf2out.c (rtl_for_decl_location): Likewise.
4996 * final.c (alter_subreg): Likewise.
4997 * rtlhooks.c (gen_lowpart_general): Likewise.
4998 (gen_lowpart_if_possible): Likewise.
5000 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5001 Alan Hayward <alan.hayward@arm.com>
5002 David Sherwood <david.sherwood@arm.com>
5004 * calls.c (expand_call): Use subreg_lowpart_offset.
5005 * cse.c (cse_insn): Likewise.
5006 * regcprop.c (copy_value): Likewise.
5007 (copyprop_hardreg_forward_1): Likewise.
5009 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5012 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5013 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5014 =Q constraints to +Q and into insn condition add check
5015 that operands[0] and operands[1] are equal.
5016 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5017 =Q constraints to +Q and into insn condition add check
5018 that operands[0] is equal to either operands[1] or operands[2].
5021 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5022 instead of handling MINUS_EXPR twice (once for each argument),
5023 canonicalize operand order and handle just once, use rtype where
5024 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5027 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5028 any values of __C while still being pattern recognizable as a simple
5031 2017-10-13 Richard Biener <rguenther@suse.de>
5033 PR tree-optimization/82451
5035 2017-10-02 Richard Biener <rguenther@suse.de>
5037 PR tree-optimization/82355
5038 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5039 a mapping for the enclosing loop but avoid generating one for
5041 (copy_bb_and_scalar_dependences): Remove premature codegen
5042 error on PHIs in blocks duplicated into multiple places.
5043 * graphite-scop-detection.c
5044 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5045 in the region use it as loop and nest to analyze the DR in.
5046 (try_generate_gimple_bb): Likewise.
5047 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5048 (add_loop_constraints): For blocks in a loop not in the region
5049 create a dimension with a single iteration.
5050 * sese.h (gbb_loop_at_index): Remove assert.
5052 * cfgloop.c (loop_preheader_edge): For the loop tree root
5053 return the single successor of the entry block.
5054 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5055 Reset the SCEV hashtable and niters.
5056 * graphite-scop-detection.c
5057 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5058 assert that we only have POLYNOMIAL_CHREC that vary in loops
5059 contained in the region.
5060 (scop_detection::graphite_can_represent_expr): Adjust.
5061 (scop_detection::stmt_has_simple_data_refs_p): For loops
5062 not in the region set loop to NULL. The nest is now the
5063 entry edge to the region.
5064 (try_generate_gimple_bb): Likewise.
5065 * sese.c (scalar_evolution_in_region): Adjust for
5066 instantiate_scev change.
5067 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5068 Make nest parameter the edge into the region.
5069 (create_data_ref): Likewise.
5070 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5071 entry edge into a region and adjust instantiate_scev calls.
5072 (create_data_ref): Likewise.
5073 (graphite_find_data_references_in_stmt): Likewise.
5074 (find_data_references_in_stmt): Pass the loop preheader edge
5075 from the nest argument.
5076 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5077 parameter the edge into the region.
5078 (instantiate_parameters): Use the loop preheader edge as entry.
5079 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5081 (get_instantiated_value_entry): Make instantiate_below parameter
5082 the edge into the region.
5083 (instantiate_scev_name): Likewise. Adjust dominance checks,
5084 when we cannot use loop-based instantiation instantiate by
5085 walking use-def chains.
5086 (instantiate_scev_poly): Adjust.
5087 (instantiate_scev_binary): Likewise.
5088 (instantiate_scev_convert): Likewise.
5089 (instantiate_scev_not): Likewise.
5090 (instantiate_array_ref): Remove.
5091 (instantiate_scev_3): Likewise.
5092 (instantiate_scev_2): Likewise.
5093 (instantiate_scev_1): Likewise.
5094 (instantiate_scev_r): Do not blindly handle N-operand trees.
5095 Do not instantiate array-refs. Handle all constants and invariants.
5096 (instantiate_scev): Make instantiate_below parameter
5097 the edge into the region.
5098 (resolve_mixers): Use the loop preheader edge for the region
5099 parameter to instantiate_scev_r.
5100 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5102 2017-10-13 Richard Biener <rguenther@suse.de>
5104 PR tree-optimization/82525
5105 * graphite-isl-ast-to-gimple.c
5106 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5108 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5109 Fail code generation when we cannot represent the isl integer.
5110 (binary_op_to_tree): Elide modulo operations that are no-ops
5111 in the type we code generate. Remove now superfluous code
5114 2017-10-13 Richard Biener <rguenther@suse.de>
5116 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5117 (scop_detection::harmful_loop_in_region): Remove premature
5118 IV type restriction.
5119 (scop_detection::graphite_can_represent_scev): We can handle
5120 pointer IVs just fine.
5122 2017-10-13 Alan Modra <amodra@gmail.com>
5124 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5125 "Clobbers and Scratch Registers". Add paragraph on
5126 alternative to clobbers for scratch registers and OpenBLAS
5129 2017-10-13 Alan Modra <amodra@gmail.com>
5131 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5132 example of a memory input for a string of known length. Move
5133 commentary out of table. Add a number of new examples
5134 covering array memory inputs.
5136 2017-10-12 Martin Liska <mliska@suse.cz>
5138 PR tree-optimization/82493
5139 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5140 (test_range_functions): New function.
5141 (sbitmap_c_tests): Likewise.
5142 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5143 * selftest.h (sbitmap_c_tests): New function.
5145 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5147 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5149 * config/rs6000/amo.h: Fix spacing issue.
5151 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5154 * config/i386/i386.md (*ashl<mode>3_mask_1,
5155 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5156 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5159 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5161 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5162 (profile_probability): Set max_probability
5163 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5165 * profile-count.c (profile_probability::differs_from_p): Do not
5166 rely on max_probaiblity == 10000
5168 2017-10-12 Jeff Law <law@redhat.com>
5170 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5173 2017-10-12 Martin Sebor <msebor@redhat.com>
5177 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5178 (handle_alias_pairs): Call it.
5179 * common.opt (-Wattribute-alias): New option.
5180 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5181 * doc/invoke.texi (-Wattribute-alias): Document.
5183 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5186 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5188 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5190 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5191 (make_hard_regno_born, make_hard_regno_dead): Update
5192 bb_killed_pseudos and bb_gen_pseudos.
5194 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5196 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5198 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5200 * config/alpha/alpha.c (alpha_split_conditional_move):
5201 Use std::swap instead of manually swapping.
5202 (alpha_stdarg_optimize_hook): Ditto.
5203 (alpha_canonicalize_comparison): Ditto.
5205 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5207 * tree-loop-distribution.c (struct builtin_info): New struct.
5208 (struct partition): Refactor fields into struct builtin_info.
5209 (partition_free): Free struct builtin_info.
5210 (build_size_arg_loc, build_addr_arg_loc): Delete.
5211 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5212 information from struct builtin_info.
5213 (find_single_drs): New function refactored from classify_partition.
5214 Also moved builtin validity checks to this function.
5215 (compute_access_range, alloc_builtin): New functions.
5216 (classify_builtin_st, classify_builtin_ldst): New functions.
5217 (classify_partition): Refactor code into functions find_single_drs,
5218 classify_builtin_st and classify_builtin_ldst.
5219 (distribute_loop): Don't do runtime alias check when distributing
5221 (find_seed_stmts_for_distribution): New function.
5222 (pass_loop_distribution::execute): Refactor code finding seed
5223 stmts into above function. Support distribution for the innermost
5224 two-level loop nest. Adjust dump information.
5226 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5228 * tree-loop-distribution.c: Adjust the general comment.
5229 (NUM_PARTITION_THRESHOLD): New macro.
5230 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5231 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5232 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5233 in call to build_partition_graph.
5234 (finalize_partitions): New parameter. Make loop distribution more
5235 conservative by fusing more partitions.
5236 (distribute_loop): Don't do runtime alias check in case of loop nest
5238 (find_seed_stmts_for_distribution): New function.
5239 (prepare_perfect_loop_nest): New function.
5240 (pass_loop_distribution::execute): Refactor code finding seed stmts
5241 and loop nest into above functions. Support loop nest distribution.
5242 Adjust dump information accordingly.
5244 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5246 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5247 and set PTYPE_SEQUENTIAL for merged partition.
5249 2017-10-12 Richard Biener <rguenther@suse.de>
5251 PR tree-optimization/69728
5253 2017-09-19 Richard Biener <rguenther@suse.de>
5255 PR tree-optimization/69728
5256 * graphite-sese-to-poly.c (schedule_error): New global.
5257 (add_loop_schedule): Handle empty domain by failing the
5259 (build_original_schedule): Handle schedule_error.
5261 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5262 domain by returning an unchanged schedule.
5264 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5266 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5267 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5269 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5271 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5274 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5277 * expr.c (store_field): Don't optimize away bitsize == 0 store
5278 from CALL_EXPR with addressable return type.
5280 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5282 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5283 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5284 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5285 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5286 TARGET_ISEL instead of TARGET_ISEL<sel>.
5288 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5290 * config/rs6000/rs6000.c
5291 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5293 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5295 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5297 (reg_or_cint_operand): Fix comment.
5298 (reg_or_zero_operand): New predicate.
5299 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5300 * config/rs6000/rs6000.c (output_isel): Delete.
5301 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5302 instead of reg_or_cint_operand. Output instruction directly (not via
5304 (isel_unsigned_<mode>): Ditto.
5305 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5306 gpc_reg_operand. Add an instruction alternative for this. Output
5307 instruction directly.
5308 (*isel_reversed_unsigned_<mode>): Ditto.
5310 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5312 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5313 (TARGET_CANONICALIZE_COMPARISON): Define.
5315 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5318 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5319 Check whether the dest is REG before adding REG_EQUIV note.
5321 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5324 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5326 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5327 (make_hard_regno_born, make_hard_regno_dead): Update
5328 bb_killed_pseudos and bb_gen_pseudos.
5330 2017-10-11 Nathan Sidwell <nathan@acm.org>
5332 * incpath.h (enum incpath_kind): Name enum, prefix values.
5333 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5334 * incpath.c (heads, tails): Use INC_MAX.
5335 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5336 (merge_include_chains, split_quote_chain,
5337 register_include_chains): Update incpath_kind names.
5338 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5339 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5341 (add_framework_path, darwin_register_objc_includes): Likewise.
5342 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5344 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5346 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5347 Do not use float_operator operator predicate.
5348 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5349 * config/i386/predicates.md (float_operator): Remove predicate.
5351 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5353 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5354 (*jccxf_i387): Ditto.
5355 (*jcc<mode>_i387): Ditto.
5356 (*jccu<mode>_i387): Ditto.
5357 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5358 (*jcc_*_i387 splitters): Remove.
5359 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5360 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5361 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5364 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5366 * profile-count.h (slow_safe_scale_64bit): New function.
5367 (safe_scale_64bit): New inline.
5368 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5369 * profile-count.c: Include wide-int.h
5370 (slow_safe_scale_64bit): New.
5372 2017-10-11 Nathan Sidwell <nathan@acm.org>
5374 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5375 HAS_DECL_ASSEMBLER_NAME_P.
5376 * gimple-expr.c (gimple_decl_printable_name: Check
5377 HAS_DECL_ASSEMBLER_NAME_P too.
5378 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5379 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5380 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5381 * passes.c (rest_of_decl_compilation): Check
5382 HAS_DECL_ASSEMBLER_NAME_P too.
5383 * recog.c (verify_changes): Likewise.
5384 * tree-pretty-print.c (dump_decl_name): Likewise.
5385 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
5387 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
5388 (SET_DECL_ASSEMBLER_NAME): Use it.
5389 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
5390 (COPY_DECL_ASSEMBLER_NAME): Likewise.
5391 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
5393 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5395 * config.gcc (i386, x86_64): Add extra objects.
5396 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
5397 (ix86_min_insn_size): Declare.
5398 (ix86_issue_rate): Declare.
5399 (ix86_adjust_cost): Declare.
5400 (ia32_multipass_dfa_lookahead): Declare.
5401 (ix86_macro_fusion_p): Declare.
5402 (ix86_macro_fusion_pair_p): Declare.
5403 (ix86_bd_has_dispatch): Declare.
5404 (ix86_bd_do_dispatch): Declare.
5405 (ix86_core2i7_init_hooks): Declare.
5406 (ix86_atom_sched_reorder): Declare.
5407 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
5408 (COSTS_N_BYTES): Move to x86-tune-costs.h.
5409 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
5410 (rip_relative_addr_p): Rename to ...
5411 (ix86_rip_relative_addr_p): ... this one; export.
5412 (memory_address_length): Update.
5413 (ix86_issue_rate): Move to x86-tune-sched.c.
5414 (ix86_flags_dependent): Move to x86-tune-sched.c.
5415 (ix86_agi_dependent): Move to x86-tune-sched.c.
5416 (exact_dependency_1): Move to x86-tune-sched.c.
5417 (exact_store_load_dependency): Move to x86-tune-sched.c.
5418 (ix86_adjust_cost): Move to x86-tune-sched.c.
5419 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
5420 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
5421 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
5422 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
5423 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
5424 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
5425 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
5426 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
5427 (min_insn_size): Rename to ...
5428 (ix86_min_insn_size): ... this one; export.
5429 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
5430 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
5431 (core2i7_first_cycle_multipass_backtrack): Move to
5432 x86-tune-sched-core.c.
5433 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
5434 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
5435 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
5436 (ix86_avoid_jump_mispredicts): Update.
5437 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
5438 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
5439 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
5440 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
5441 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
5442 (MAX_INSN): Move to ix86-tune-sched-bd.c.
5443 (MAX_IMM): Move to ix86-tune-sched-bd.c.
5444 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
5445 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
5446 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
5447 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
5448 (MAX_STORE): Move to ix86-tune-sched-bd.c.
5449 (BIG): Move to ix86-tune-sched-bd.c.
5450 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
5451 (enum insn_path): Move to ix86-tune-sched-bd.c.
5452 (get_mem_group): Move to ix86-tune-sched-bd.c.
5453 (is_cmp): Move to ix86-tune-sched-bd.c.
5454 (dispatch_violation): Move to ix86-tune-sched-bd.c.
5455 (is_branch): Move to ix86-tune-sched-bd.c.
5456 (is_prefetch): Move to ix86-tune-sched-bd.c.
5457 (init_window): Move to ix86-tune-sched-bd.c.
5458 (allocate_window): Move to ix86-tune-sched-bd.c.
5459 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
5460 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
5461 (process_end_window): Move to ix86-tune-sched-bd.c.
5462 (allocate_next_window): Move to ix86-tune-sched-bd.c.
5463 (find_constant): Move to ix86-tune-sched-bd.c.
5464 (get_num_immediates): Move to ix86-tune-sched-bd.c.
5465 (has_immediate): Move to ix86-tune-sched-bd.c.
5466 (get_insn_path): Move to ix86-tune-sched-bd.c.
5467 (get_insn_group): Move to ix86-tune-sched-bd.c.
5468 (count_num_restricted): Move to ix86-tune-sched-bd.c.
5469 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
5470 (add_insn_window): Move to ix86-tune-sched-bd.c.
5471 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
5472 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
5473 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
5474 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
5475 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
5476 (do_dispatch): Move to ix86-tune-sched-bd.c.
5477 (has_dispatch): Move to ix86-tune-sched-bd.c.
5478 * config/i386/t-i386: Add new object files.
5479 * config/i386/x86-tune-costs.h: New file.
5480 * config/i386/x86-tune-sched-atom.c: New file.
5481 * config/i386/x86-tune-sched-bd.c: New file.
5482 * config/i386/x86-tune-sched-core.c: New file.
5483 * config/i386/x86-tune-sched.c: New file.
5485 2017-10-11 Liu Hao <lh_mouse@126.com>
5487 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
5488 the generic version below instead.
5489 (should_colorize): Recognize Windows consoles as terminals
5491 * pretty-print.c [__MINGW32__] (write_all): New function.
5492 [__MINGW32__] (find_esc_head): Likewise.
5493 [__MINGW32__] (find_esc_terminator): Likewise.
5494 [__MINGW32__] (eat_esc_sequence): Likewise.
5495 [__MINGW32__] (mingw_ansi_fputs): New function that handles
5497 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
5500 2017-10-11 Richard Biener <rguenther@suse.de>
5502 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
5503 Properly call analyze_scalar_evolution with the loop of the stmt.
5505 2017-10-11 Richard Biener <rguenther@suse.de>
5507 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
5508 * tree-core.h (tree_base): Add chrec_var union member.
5509 * tree.h (CHREC_VAR): Remove.
5510 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
5511 * tree-chrec.h (build_polynomial_chrec): Adjust.
5512 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
5513 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
5515 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
5517 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
5518 * match.pd: ... here.
5519 ((T) X == (T) Y): Relax condition.
5521 2017-10-11 Bin Cheng <bin.cheng@arm.com>
5523 PR tree-optimization/82472
5524 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
5526 (break_alias_scc_partitions): Update postorder number.
5528 2017-10-11 Martin Liska <mliska@suse.cz>
5531 * opts.c (parse_no_sanitize_attribute): Do not use error_value
5533 * opts.h (parse_no_sanitize_attribute): Remove last argument.
5535 2017-10-11 Martin Liska <mliska@suse.cz>
5537 * print-rtl.c (print_insn): Move declaration of idbuf
5538 to same scope as name.
5540 2017-10-11 Martin Liska <mliska@suse.cz>
5545 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5548 2017-10-11 Martin Liska <mliska@suse.cz>
5551 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5554 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
5556 * config.gcc (default_gnu_indirect_function): Default to yes for
5557 arm*-*-linux* with glibc.
5559 2017-10-11 Richard Biener <rguenther@suse.de>
5561 * tree-scalar-evolution.c (get_scalar_evolution): Handle
5562 default-defs and types we do not want to analyze.
5563 (interpret_loop_phi): Replace unreachable code with an assert.
5564 (compute_scalar_evolution_in_loop): Remove and inline ...
5565 (analyze_scalar_evolution_1): ... here, replacing condition with
5566 what makes the intent clearer. Remove handling of cases
5567 get_scalar_evolution now handles.
5569 2017-10-10 Jim Wilson <wilson@tuliptree.org>
5571 PR rtl-optimization/81434
5572 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
5573 comment for main loop. In sched_group_found if, also add checks for
5574 pass and min_cost_group.
5576 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
5578 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
5579 (rs6000_insn_cost): New function.
5580 * config/rs6000/rs6000.md (cost): New attribute.
5582 2017-10-10 Jakub Jelinek <jakub@redhat.com>
5583 H.J. Lu <hongjiu.lu@intel.com>
5587 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
5588 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
5589 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
5590 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
5591 to OPTION_MASK_ISA_AVX512VL - builtins that have both
5592 OPTION_MASK_ISA_MMX and some other bit set require both
5593 mmx and the ISAs without the mmx bit.
5594 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
5595 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
5596 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
5597 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
5598 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
5599 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
5600 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
5601 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
5602 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
5603 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
5604 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
5605 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
5606 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
5607 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
5608 Add OPTION_MASK_ISA_MMX.
5610 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
5612 * config.gcc (armv7*-*-freebsd*): New target.
5613 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
5615 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
5617 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
5618 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
5621 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
5623 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
5624 a template parameter.
5625 (WIDE_INT_REF_FOR): Update accordingly.
5626 * tree.h (wi::int_traits <const_tree>): Delete.
5627 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
5628 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
5629 (wi::tree_to_wide_ref): New typedef.
5630 (wi::to_wide): New function.
5631 * calls.c (get_size_range): Use wi::to_wide when operating on
5633 * cgraph.c (cgraph_node::create_thunk): Likewise.
5634 * config/i386/i386.c (ix86_data_alignment): Likewise.
5635 (ix86_local_alignment): Likewise.
5636 * dbxout.c (stabstr_O): Likewise.
5637 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
5638 * expr.c (const_vector_from_tree): Likewise.
5639 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
5640 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
5641 (fold_negate_expr_1, int_const_binop_1, const_binop)
5642 (fold_convert_const_int_from_real, optimize_bit_field_compare)
5643 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
5644 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
5645 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
5646 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
5647 (fold_not_const, round_up_loc): Likewise.
5648 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
5649 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
5650 (alloca_call_type): Likewise.
5651 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
5652 * godump.c (go_output_typedef): Likewise.
5653 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
5654 * internal-fn.c (get_min_precision): Likewise.
5655 * ipa-cp.c (ipcp_store_vr_results): Likewise.
5656 * ipa-polymorphic-call.c
5657 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
5658 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
5659 (ipa_modify_call_arguments): Likewise.
5660 * match.pd: Likewise.
5661 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
5662 * print-tree.c (print_node_brief, print_node): Likewise.
5663 * stmt.c (expand_case): Likewise.
5664 * stor-layout.c (layout_type): Likewise.
5665 * tree-affine.c (tree_to_aff_combination): Likewise.
5666 * tree-cfg.c (group_case_labels_stmt): Likewise.
5667 * tree-data-ref.c (dr_analyze_indices): Likewise.
5668 (prune_runtime_alias_test_list): Likewise.
5669 * tree-dump.c (dequeue_and_dump): Likewise.
5670 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
5671 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
5672 * tree-pretty-print.c (dump_generic_node): Likewise.
5673 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
5674 (simple_iv_with_niters): Likewise.
5675 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
5676 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
5677 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
5678 * tree-ssa-loop-niter.c (split_to_var_and_offset)
5679 (refine_value_range_using_guard, number_of_iterations_ne_max)
5680 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
5681 (get_cst_init_from_scev, record_nonwrapping_iv)
5682 (scev_var_range_cant_overflow): Likewise.
5683 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
5684 * tree-ssa-pre.c (compute_avail): Likewise.
5685 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
5686 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
5687 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
5688 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
5689 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
5690 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
5691 (dump_case_nodes, try_switch_expansion): Likewise.
5692 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
5693 (vect_do_peeling): Likewise.
5694 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
5695 * tree-vect-stmts.c (vectorizable_load): Likewise.
5696 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
5697 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
5698 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
5699 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
5700 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
5701 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
5702 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
5703 (evrp_dom_walker::before_dom_children): Likewise.
5704 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
5705 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
5706 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
5707 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
5708 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
5709 (get_range_pos_neg): Likewise.
5710 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
5711 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
5712 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
5713 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
5714 * config/avr/avr.c (avr_fold_builtin): Likewise.
5715 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
5716 * config/msp430/msp430.c (msp430_attr): Likewise.
5717 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
5718 * config/powerpcspe/powerpcspe-c.c
5719 (altivec_resolve_overloaded_builtin): Likewise.
5720 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
5721 (rs6000_expand_ternop_builtin): Likewise.
5722 * config/rs6000/rs6000-c.c
5723 (altivec_resolve_overloaded_builtin): Likewise.
5724 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
5725 (rs6000_expand_ternop_builtin): Likewise.
5726 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
5728 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5730 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
5731 when copying loop nest with only one inner loop.
5733 2017-10-10 Richard Biener <rguenther@suse.de>
5735 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
5736 blocks if SCEV is active.
5737 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
5739 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
5740 (scev_initialize): Assert we are not yet initialized.
5742 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5744 * tree-loop-distribution.c (generate_loops_for_partition): Remove
5745 inner loop's exit stmt by making it always exit the loop, otherwise
5746 we would generate an infinite empty loop.
5748 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5750 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
5751 renaming variables in new preheader if it's deleted.
5753 2017-10-10 Bin Cheng <bin.cheng@arm.com>
5755 * tree-loop-distribution.c (struct partition): Remove unused field
5756 loops of the structure.
5757 (partition_alloc, partition_free): Ditto.
5758 (build_rdg_partition_for_vertex): Ditto.
5760 2017-10-09 Jeff Law <law@redhat.com>
5762 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
5763 return type to match prototype and documentation.
5765 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5767 * config/rs6000/rs6000.c (processor_costs): Move to ...
5768 * config/rs6000/rs6000.h: ... here.
5769 (rs6000_cost): Declare.
5771 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
5773 * except.c (setjmp_fn): New global variable.
5774 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
5775 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
5776 if DONT_USE_BUILTIN_SETJMP is defined.
5778 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5780 * target.def (insn_cost): New hook.
5781 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
5782 * doc/tm.texi: Regenerate.
5783 * rtlanal.c (insn_cost): Use the new hook.
5785 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5787 * combine.c (combine_validate_cost): Compute the new insn_cost,
5788 not just pattern_cost.
5789 (try_combine): Adjust comment.
5791 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
5793 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
5795 * combine.c (uid_insn_cost): Adjust comment.
5796 (combine_validate_cost): Adjust comment. Use pattern_cost instead
5798 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
5799 * dse.c (find_shift_sequence): Ditto.
5800 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
5801 (bb_valid_for_noce_process_p): Use pattern_cost.
5802 * rtl.h (insn_rtx_cost): Delete.
5803 (pattern_cost): New prototype.
5804 (insn_cost): New prototype.
5805 * rtlanal.c (insn_rtx_cost): Rename to...
5806 (pattern_cost): ... this.
5809 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
5811 * config/i386/i386.md (*jcc_2): Remove insn pattern.
5812 (*jcc<mode>_0_r_i387): Ditto.
5813 (*jccxf_r_i387): Ditto.
5814 (*jcc<mode>_r_i387): Ditto.
5815 (*jccu<mode>_r_i387): Ditto.
5816 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
5817 (*jcc): Rename from *jcc_1.
5819 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5821 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
5822 deferred rescans after the lvx/stvx recombination pre-pass.
5824 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
5826 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
5827 memory operation instruction support.
5828 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
5829 (rs6000-ibm-aix[789]*): Likewise.
5830 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
5831 Document new functions.
5833 2017-10-09 Richard Biener <rguenther@suse.de>
5835 PR tree-optimization/82397
5836 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5837 equality only for semantically equal trees.
5839 2017-10-09 Richard Biener <rguenther@suse.de>
5841 PR tree-optimization/82449
5842 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5843 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5844 allow constant addresses.
5845 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5848 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5850 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5853 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5856 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5859 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5862 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5864 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5867 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5868 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5870 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5872 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5873 (WI_BINARY_PREDICATE_RESULT): Likewise.
5874 (wi::binary_traits::operator_result): New type.
5875 (wi::binary_traits::predicate_result): Likewise.
5876 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5877 (generic_wide_int::operator==, generic_wide_int::operator!=)
5878 (generic_wide_int::operator&, generic_wide_int::and_not)
5879 (generic_wide_int::operator|, generic_wide_int::or_not)
5880 (generic_wide_int::operator^, generic_wide_int::operator+
5881 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5883 (operator~, unary operator-, operator==, operator!=, operator&)
5884 (operator|, operator^, operator+, binary operator-, operator*): New
5886 * expr.c (get_inner_reference): Use wi::bit_and_not.
5887 * fold-const.c (fold_binary_loc): Likewise.
5888 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
5889 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
5890 (bit_value_binop): Likewise.
5891 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
5892 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
5893 (extract_range_from_binary_expr_1): Likewise.
5894 (masked_increment): Likewise.
5895 (simplify_bit_ops_using_ranges): Likewise.
5897 2017-10-09 Martin Jambor <mjambor@suse.cz>
5900 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5901 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5902 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5904 (hsa_fixup_mov_insn_type): New function.
5905 (hsa_op_with_type::get_in_type): Use it.
5906 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5907 immediates in an assert.
5908 (hsa_op_with_type::extend_int_to_32bit): New method.
5909 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5910 types. Convert to dest type if necessary.
5911 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5912 (reg_for_gimple_ssa): Pass false as min32int to
5913 hsa_type_for_scalar_tree_type.
5914 (gen_hsa_addr): Fixup type when creating addresable temporary.
5915 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5916 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5917 necessary. Call hsa_fixup_mov_insn_type.
5918 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5919 extend operands and convert to dest type if necessary.
5920 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5921 to dest type if necessary.
5922 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5923 if conversion nt necessary and size matches.
5924 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5925 to dest type if necessary.
5926 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5927 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5929 (gen_hsa_clrsb): Likewise.
5930 (gen_hsa_ffs): Likewise.
5931 (gen_hsa_divmod): Extend operands and convert to dest type if
5933 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
5935 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5937 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
5938 Remove empty default arguments. Use a brace block as output
5940 (conditional return): Ditto.
5942 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
5943 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
5945 (group_ending_nop): Ditto.
5946 (doloop_end): Ditto.
5947 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
5948 (splitters for those): Ditto.
5950 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5952 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
5953 a conditional jump (and the compare for it) so that pc_rtx is the
5955 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
5956 for the deleted and renamed ctr<mode>_internal[234] patterns.
5957 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
5958 Delete second conditional return pattern.
5959 (ctr<mode>_internal2): Delete this second bdnz pattern.
5960 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
5961 (ctr<mode>_internal4): Delete this second bdz pattern.
5963 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5965 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
5966 (always_initialized_rtx_for_ssa_name_p): New predicate.
5967 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
5968 (finish_out_of_ssa): Free new field of SA.
5969 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
5970 * tree-ssa-coalesce.c: Include tree-ssa.h.
5971 (get_parm_default_def_partitions): Remove extern keyword.
5972 (get_undefined_value_partitions): New function.
5973 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
5974 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
5976 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
5978 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5980 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5982 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5984 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5985 for targets that preffer 128bit.
5987 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5989 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
5991 2017-10-08 Olivier Hainque <hainque@adacore.com>
5993 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
5994 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
5996 (thumb_set_return_address): Likewise.
5998 2017-10-08 Olivier Hainque <hainque@adacore.com>
6000 * common/config/arm/arm-common.c (arm_except_unwind_info):
6001 Handle DWARF2_UNWIND_INFO.
6003 2017-10-07 Michael Collison <michael.collison@arm.com>
6005 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6008 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6010 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6011 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6012 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6013 defined, force the creation of a new block for a dispatch label.
6015 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6017 * invoke.texi (Wsuggest-attribute=cold): Document.
6018 * common.opt (Wsuggest-attribute=cold): New
6019 * ipa-pure-const.c (warn_function_cold): New function.
6020 * predict.c (compute_function_frequency): Use it.
6021 * predict.h (warn_function_cold): Declare.
6023 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6025 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6028 2017-10-06 Martin Liska <mliska@suse.cz>
6030 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6031 keyword for member functions.
6032 (struct sanopt_tree_couple): New struct.
6033 (struct sanopt_tree_couple_hash): New function.
6034 (struct sanopt_ctx): Add new hash_map.
6035 (has_dominating_ubsan_ptr_check): New function.
6036 (record_ubsan_ptr_check_stmt): Likewise.
6037 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6038 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6039 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6041 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6044 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6045 aarch64_simd_valid_immediate on CONST_VECTORs.
6046 (aarch64_reg_or_bic_imm): Likewise.
6048 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6050 PR rtl-optimization/82396
6051 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6053 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6055 * graphite-dependences.c (scop_get_reads): Move code to...
6056 (scop_get_must_writes): Move code to...
6057 (scop_get_may_writes): Move code to...
6058 (scop_get_reads_and_writes): ... here.
6059 (scop_get_dependences): Call scop_get_reads_and_writes.
6061 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6063 PR tree-optimization/82434
6064 * fold-const.h (can_native_encode_type_p,
6065 can_native_encode_string_p): Remove.
6066 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6067 don't encode anything, just return what would be otherwise returned.
6068 (native_encode_fixed, native_encode_complex, native_encode_vector):
6070 (native_encode_string): Likewise. Inline by hand
6071 can_native_encode_string_p.
6072 (can_native_encode_type_p): Remove.
6073 (can_native_encode_string_p): Remove.
6074 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6075 STRING_CSTs using can_native_encode_string_p, test all
6076 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6077 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6078 argument from native_encode_expr.
6079 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6080 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6083 2017-10-06 Richard Biener <rguenther@suse.de>
6085 PR tree-optimization/82397
6086 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6087 operand_equal_p but rely on data_ref_compare_tree for detecting
6089 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6090 to match up with dr_group_sort_cmp.
6092 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6095 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6097 * config/s390/s390-builtin-types.def: Regenerate.
6099 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6102 * config/s390/s390-builtin-types.def: Regenerate.
6103 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6104 Change flag from B_VXE to B_VX.
6105 (s390_vec_min_dbl): Remove B_VXE flag.
6107 2017-10-06 Richard Biener <rguenther@suse.de>
6109 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6110 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6111 translate_isl_ast_to_gimple::is_valid_rename,
6112 translate_isl_ast_to_gimple::get_rename,
6113 translate_isl_ast_to_gimple::get_def_bb_for_const,
6114 translate_isl_ast_to_gimple::get_new_name,
6115 translate_isl_ast_to_gimple::collect_all_ssa_names,
6116 translate_isl_ast_to_gimple::copy_loop_phi_args,
6117 translate_isl_ast_to_gimple::collect_all_ssa_names,
6118 translate_isl_ast_to_gimple::copy_loop_phi_args,
6119 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6120 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6121 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6122 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6123 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6124 translate_isl_ast_to_gimple::copy_cond_phi_args,
6125 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6126 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6127 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6128 translate_isl_ast_to_gimple::rename_uses,
6129 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6130 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6131 (set_rename_for_each_def): Likewise.
6132 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6133 here. Handle rewriting SCEV analyzable uses here.
6134 (copy_bb_and_scalar_dependences): Generate code for PHI
6136 (graphite_regenerate_ast_isl): Adjust.
6137 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6138 (add_write, add_read): New functions.
6139 (build_cross_bb_scalars_def): Use it and simplify.
6140 (build_cross_bb_scalars_use): Likewise.
6141 (graphite_find_cross_bb_scalar_vars): Inline into...
6142 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6143 simulating out-of-SSA. Compute liveout and add dependencies.
6144 (build_scops): Force an empty entry block.
6145 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6147 (sese_build_liveouts): Declare.
6148 (sese_trivially_empty_bb_p): Likewise.
6149 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6150 compute liveout and debug_liveout.
6151 (sese_bad_liveouts_use): Remove.
6152 (sese_reset_debug_liveouts_bb): Likewise.
6153 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6154 (sese_build_liveouts): Build liveout and debug_liveout and store
6156 (new_sese_info): Adjust.
6157 (free_sese_info): Likewise.
6158 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6159 do not build liveout here.
6160 (move_sese_in_condition): Adjust region entry.
6161 (scev_analyzable_p): Match up with chrec_apply requirements.
6162 (sese_trivially_empty_bb_p): New.
6163 * tree-into-ssa.c (get_reaching_def): Properly support generating
6164 default-defs for incremental rewrite of anonymous names.
6166 2017-10-06 Richard Biener <rguenther@suse.de>
6168 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6169 precision do not perform modulo reduction.
6171 2017-10-06 Richard Biener <rguenther@suse.de>
6173 PR tree-optimization/82436
6174 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6175 conservatively choose the vectorization factor when checking
6176 whether we can perform the required load permutation.
6177 (vect_transform_slp_perm_load): Assert when we may not fail.
6179 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6181 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6182 message for incompatible -msdata=* and -mcall-* options.
6184 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6186 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6187 rate for post-reload scheduling.
6189 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6191 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6193 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6195 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6196 to improve monte carlo in scimark.
6198 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6200 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6201 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6202 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6203 pentium4_cost, nocona_cost): Set reassociation width to 1.
6204 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6205 width to 2 for fp operations and 1 otherwise.
6206 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6208 (atom_cost): Set reassociation width to 2.
6209 (slm_cost, generic_cost): Set fp reassociation width
6210 to 2 and 1 otherwise.
6211 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6212 (core_cost): Set fp reassociation width to 4 and vector to 2.
6213 (ix86_reassociation_width): Rewrite using cost table; special case
6214 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6215 and TARGET_AVX128_OPTIMAL.
6216 * config/i386/i386.h (processor_costs): Add
6217 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6218 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6219 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6220 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6221 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6222 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6224 2017-10-05 Nathan Sidwell <nathan@acm.org>
6226 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6228 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6230 * config/arm/arm.c (arm_test_fpu_data): New.
6231 (arm_run_selftests): Call arm_test_fpu_data.
6233 2017-10-04 Nathan Sidwell <nathan@acm.org>
6235 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6237 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6239 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6241 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6242 check type for aarch64_simd_valid_immediate.
6243 (aarch64_output_simd_mov_immediate): Update prototype.
6244 (aarch64_simd_valid_immediate): Update prototype.
6245 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6246 support for ORR-immediate.
6247 (and<mode>3): modified pattern to add support for BIC-immediate.
6248 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6249 now checks for valid immediate for BIC and ORR based on new enum
6251 (aarch64_output_simd_mov_immediate): Function now used to output
6252 BIC/ORR imm as well based on new enum argument.
6253 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6255 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6256 (aarch64_reg_or_bic_imm): Likewise.
6258 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6260 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6261 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6263 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6267 PR rtl-optimization/82396
6268 * haifa-sched.c (autopref_multipass_init): Simplify
6270 (autopref_rank_data): Simplify sort order.
6271 * sched-int.h (autopref_multipass_data_): Remove
6272 multi_mem_insn_p, min_offset and max_offset.
6274 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6276 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6278 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6280 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6283 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6285 * doc/sourcebuild.texi: Document vect_long_mult.
6287 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6289 PR tree-optimization/82413
6290 * fold-const.c (build_range_check): Use widest_int when comparing
6291 the maximum ETYPE value with HIGH.
6293 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6295 PR rtl-optimization/82396
6296 * haifa-sched.c (autopref_multipass_init): Simplify
6298 (autopref_rank_data): Simplify sort order.
6299 * sched-int.h (autopref_multipass_data_): Remove
6300 multi_mem_insn_p, min_offset and max_offset.
6302 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6304 PR tree-optimization/82381
6305 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6306 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6309 PR tree-optimization/82374
6310 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6311 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6312 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6313 current_function_decl to the new decl.
6315 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6317 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6318 helper macro for IEEE float128 hardware built-in functions.
6319 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6321 (TRUNCF128_ODD): Likewise.
6322 (ADDF128_ODD): Likewise.
6323 (SUBF128_ODD): Likewise.
6324 (MULF128_ODD): Likewise.
6325 (DIVF128_ODD): Likewise.
6326 (FMAF128_ODD): Likewise.
6327 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6328 UNSPEC_TRUNC_ROUND_TO_ODD.
6329 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6330 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6331 floating point round to odd instructions.
6332 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6333 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6334 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6335 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6336 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6337 (trunc<mode>sf2_hw): Change the truncate with round to odd
6338 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6339 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6340 to odd hardware instructions.
6341 (sub<mode>3_odd): Likewise.
6342 (mul<mode>3_odd): Likewise.
6343 (div<mode>3_odd): Likewise.
6344 (sqrt<mode>2_odd): Likewise.
6345 (fma<mode>4_odd): Likewise.
6346 (fms<mode>4_odd): Likewise.
6347 (nfma<mode>4_odd): Likewise.
6348 (nfms<mode>4_odd): Likewise.
6349 (trunc<mode>df2_odd): Change the truncate with round to odd
6350 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6352 * doc/extend.texi (PowerPC built-in functions): Update documentation
6353 for existing IEEE float128-bit built-in functions. Add built-in
6354 functions that generate the IEEE 128-bit floating point round to
6357 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6359 PR rtl-optimization/77729
6360 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6361 to (X&(C1&~C2))|C2 transformations.
6363 2017-10-03 Martin Jambor <mjambor@suse.cz>
6365 PR tree-optimization/82363
6366 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6367 mismatch, mark lacc written regardless of racc.
6369 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6371 PR tree-optimization/82381
6372 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6373 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6374 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6375 fallthrough into reassoc_stmt_dominates_stmt_p.
6378 * combine.c (combine_instructions): Don't combine in unreachable
6381 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6384 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
6385 function to not use the have_cpu variable. Do not set cpu_index,
6386 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
6388 (rs6000_valid_attribute_p): Remove duplicate initializations of
6389 old_optimize and func_optimize.
6390 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
6391 (rs6000_activate_target_options): Make global.
6392 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
6395 2017-10-02 Jakub Jelinek <jakub@redhat.com>
6397 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
6398 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
6399 Set *poffset to 0 and *psize and *pmax_size to -1 if
6400 *poffset + *psize overflows in HOST_WIDE_INT.
6402 PR tree-optimization/82387
6403 PR tree-optimization/82388
6404 PR tree-optimization/82389
6405 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
6406 instead of live_bytes non-NULL.
6408 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
6411 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
6414 2017-10-02 Richard Biener <rguenther@suse.de>
6416 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
6417 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
6418 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
6420 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6422 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
6423 requested precision matches the type's.
6424 * calls.c (alloc_max_size): Calculate the new candidate size as
6425 a widest_int and use wi::to_widest when comparing it with the
6426 current candidate size.
6427 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
6428 zero rather than integer_zero_node.
6429 * match.pd: Check for a no-op conversion before using wi::add
6430 rather than after. Use tree_to_uhwi when summing small shift
6431 counts into an unsigned int.
6433 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6434 Alan Hayward <alan.hayward@arm.com>
6435 David Sherwood <david.sherwood@arm.com>
6438 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
6439 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
6440 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
6441 POINTER_AND_FP_REGS.
6443 2017-10-02 Richard Biener <rguenther@suse.de>
6445 PR tree-optimization/82355
6446 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6447 a mapping for the enclosing loop but avoid generating one for
6449 (copy_bb_and_scalar_dependences): Remove premature codegen
6450 error on PHIs in blocks duplicated into multiple places.
6451 * graphite-scop-detection.c
6452 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6453 in the region use it as loop and nest to analyze the DR in.
6454 (try_generate_gimple_bb): Likewise.
6455 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6456 (add_loop_constraints): For blocks in a loop not in the region
6457 create a dimension with a single iteration.
6458 * sese.h (gbb_loop_at_index): Remove assert.
6460 2017-10-01 Kevin Buettner <kevinb@redhat.com>
6462 * omp-expand.c (adjust_context_scope): New function.
6463 (expand_parallel_call): Call adjust_context_scope.
6465 2017-10-01 Jeff Law <law@redhat.com>
6467 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
6468 dom_opt_dom_walker class with direct access to private members.
6469 Add comments. Call test_for_singularity.
6470 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
6471 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
6472 m_dummy_cond anymore.
6473 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
6475 (pass_dominator:execute): Build the dummy_cond here and pass it
6476 to the dom_opt_dom_walker ctor.
6477 (test_for_singularity): New function.
6479 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
6480 Maya Rashish <coypu@sdf.org>
6482 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
6483 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
6484 (alpha*-*-netbsd*) Use nbsd_tm_file.
6485 (arm*-*-netbsdelf*) Likewise.
6486 (i[34567]86-*-netbsdelf*) Likewise.
6487 (x86_64-*-netbsd*) Likewise.
6488 (mips*-*-netbsd*) Likewise.
6489 (powerpc-*-netbsd*) Likewise.
6490 (sh*-*-netbsd*) Likewise.
6491 (sparc-*-netbsdelf*) Likewise.
6492 (sparc64-*-netbsd*) Likewise.
6493 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
6495 (vax-*-netbsdelf*) Likewise.
6496 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
6497 (UINT_FAST8_TYPE) Likewise.
6498 (INT_FAST16_TYPE) Check CHAR_FAST16.
6499 (UINT_FAST16_TYPE) Likewise.
6501 2017-09-30 Jakub Jelinek <jakub@redhat.com>
6504 * config/i386/i386.md
6505 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
6506 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
6507 *divmodsi4_zext_2): New define_insn_and_split.
6508 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
6509 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
6510 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
6511 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
6512 New define_insn_and_split.
6513 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
6514 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
6515 operands[1] having DImode when mode is SImode.
6517 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
6518 always SImode for DIV and MOD in REG_EQUAL notes.
6520 2017-09-29 Yury Gribov <tetra2005@gmail.com>
6523 * match.pd: Fix handling of NaNs in pattern.
6525 2017-09-29 Jeff Law <law@redhat.com>
6527 * sbitmap.c (bitmap_bit_in_range_p): New function.
6528 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
6529 * tree-ssa-dse.c (live_bytes_read): New function.
6530 (dse_classify_store): Ignore reads of dead bytes.
6532 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
6533 typos and whitespace errors.
6534 * config/i386/predicates.md (address_no_seg_operand): Likewise.
6535 * config/s390/s390.c (s390_emit_prologue): Likewise.
6537 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6540 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
6541 with a symbol for LRA.
6543 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6545 PR rtl-optimization/82338
6546 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
6548 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
6550 * genmodes.c (calc_wider_mode): Suppress qsort macro.
6551 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
6552 (qsort_chk): Declare.
6553 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
6554 (qsort_chk): New function.
6556 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6558 PR tree-optimization/82337
6559 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
6560 phi definition if the PHI result appears in an abnormal PHI.
6561 (find_basis_for_base_expr): Don't record a basis if the LHS of the
6562 basis appears in an abnormal PHI.
6564 2017-09-29 Richard Biener <rguenther@suse.de>
6566 * graphite-isl-ast-to-gimple.c
6567 (translate_isl_ast_to_gimple::set_codegen_error): New function.
6568 (binary_op_to_tree): Use it.
6569 (get_rename_from_scev): Likewise.
6570 (copy_loop_phi_nodes): Likewise.
6571 (copy_bb_and_scalar_dependences): Likewise.
6572 (translate_pending_phi_nodes): Likewise.
6574 2017-09-29 Jakub Jelinek <jakub@redhat.com>
6577 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
6578 for movabsq $(i32 << shift), r64.
6580 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
6582 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
6583 index when encoding %esp as %rsp to avoid 0x67 prefix.
6585 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
6587 * config/i386/i386.md (*movsf_internal, *movdf_internal):
6588 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
6590 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
6592 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
6593 Extensions with more than 16 double VFP registers.
6594 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
6595 to_clear_mask and all code related to it. Replace the remaining
6596 entry by a sbitmap and adapt code accordingly.
6598 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
6600 * brig-builtins.def: Change pure attributes to const.
6602 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6604 * config.gcc (default_gnu_indirect_function): Default to yes for
6605 sparc*-*-linux* with glibc.
6607 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6609 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
6610 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
6611 when creating .init_array and .fini_array sections with priority
6614 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
6617 * config/aarch64/aarch64.c
6618 (aarch64_builtin_support_vector_misalignment): Always return false
6619 when misalignment is unknown.
6621 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
6623 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
6624 this function to return false if the definition used by the swap
6625 instruction is artificial, or if the memory address from which the
6626 constant value is loaded is not represented by a base address held
6627 in a register or if the base address register is a frame or stack
6628 pointer. Additionally, return false if the base address of the
6629 loaded constant is a SYMBOL_REF but is not considered to be a
6631 (replace_swapped_load_constant): New function.
6632 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
6633 loaded constant vector with a load of a swapped constant vector.
6635 2017-09-27 Carl Love <cel@us.ibm.com>
6637 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
6638 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
6639 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
6642 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
6644 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
6645 first, always call autopref_rank_data otherwise.
6647 2017-09-27 Richard Biener <rguenther@suse.de>
6649 * graphite-scop-detection.c (find_scop_parameters): Move
6650 loop bound handling ...
6651 (gather_bbs::before_dom_children): ... here, avoiding the need
6652 to build scop_info->loop_nest.
6653 (record_loop_in_sese): Remove.
6654 * sese.h (sese_info_t::loop_nest): Remove.
6655 * sese.c (new_sese_info): Do not allocate loop_nest.
6656 (free_sese_info): Do not free loop_nest.
6658 2017-09-27 Jakub Jelinek <jakub@redhat.com>
6661 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
6662 lhs from calls if the lhs has addressable type.
6664 2017-09-27 Richard Biener <rguenther@suse.de>
6666 * graphite.h (scop::max_alias_set): New member.
6667 * graphite-scop-detection.c: Remove references to non-existing
6668 --param in comments.
6669 (build_alias_sets): Record the maximum alias set used for drs.
6670 (build_scops): Support zero as unlimited for
6671 --param graphite-max-arrays-per-scop.
6672 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
6674 (build_poly_sr_1): ... here. Compute alias set based on the
6675 maximum alias set used for drs rather than
6676 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
6678 2017-09-27 Richard Biener <rguenther@suse.de>
6680 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
6681 --param loop-block-tile-size=0 to disable tiling.
6683 2017-09-27 Richard Biener <rguenther@suse.de>
6685 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
6686 (graphite-max-nb-scop-params): Document special value zero.
6687 * domwalk.h (dom_walker::STOP): New symbolical constant.
6688 (dom_walker::dom_walker): Add optional parameter for bb to
6690 (dom_walker::~dom_walker): Declare.
6691 (dom_walker::before_dom_children): Document STOP return value.
6692 (dom_walker::m_user_bb_to_rpo): New member.
6693 (dom_walker::m_bb_to_rpo): Likewise.
6694 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
6695 mapping here if not provided by the user.
6696 (dom_walker::~dom_walker): Free bb to RPO mapping if not
6697 provided by the user.
6698 (dom_walker::STOP): Define.
6699 (dom_walker::walk): Do not compute bb to RPO mapping here.
6700 Support STOP return value from before_dom_children to stop
6702 * graphite-optimize-isl.c (optimize_isl): If the schedule
6703 is the same still generate code if -fgraphite-identity
6704 or -floop-parallelize-all are given.
6705 * graphite-scop-detection.c: Include cfganal.h.
6706 (gather_bbs::gather_bbs): Get and pass through bb to RPO
6708 (gather_bbs::before_dom_children): Return STOP for BBs
6710 (build_scops): Compute bb to RPO mapping and pass it to
6711 the domwalk. Treat --param graphite-max-nb-scop-params=0
6712 as not limiting the number of params.
6713 * graphite.c (graphite_initialize): Remove limit on the
6714 number of basic-blocks in a function.
6715 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
6716 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
6717 default value of 10.
6719 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6721 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
6722 Adjust code to eliminate needing to do the shift right 32-bits
6723 operation after XSCVDPSPN.
6725 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6727 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
6728 ((X / Y) != 0 -> X >= Y): Likewise.
6730 2017-09-26 Carl Love <cel@us.ibm.com>
6732 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
6733 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
6734 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
6735 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
6736 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
6737 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
6738 definitions and overloading.
6739 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
6740 statement for P9V_BUILTIN_XST_LEN_R.
6741 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
6742 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
6743 define_expand and define_insn for the instructions and builtins.
6744 * doc/extend.texi: Update the built-in documentation file for the new
6746 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
6747 define_insn for the instructions
6749 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
6752 * gcc/config/netbsd-protos.h: New file.
6753 * gcc/config/netbsd.c: New file.
6754 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
6755 * gcc/config/t-netbsd: New file.
6756 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
6757 (tmake_file) Add t-netbsd.
6758 (extra_objs) Add netbsd.o.
6760 2017-09-26 Janus Weil <janus@gcc.gnu.org>
6764 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
6766 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6768 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
6769 sign extension from a vector register to a GPR by doing a 32-bit
6770 direct move and then an EXTSW.
6771 (extendsi<mode>2 splitter): Likewise.
6772 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
6773 right or vector extract after doing XSCVDPSPN. Use
6774 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
6776 (movdi_from_sf_zero_ext): Likewise.
6777 (reload_gpr_from_vsxsf): Likewise.
6778 (p8_mfvsrd_4_disf): Delete, no longer used.
6779 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
6780 then needing to move the SFmode to a GPR to use the XSCVDPSP
6781 instruction instead of FRSP and XSCVDPSPN.
6782 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
6783 it is adjacent to the other XSCVSPDP insns.
6784 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
6785 SFmode to be in traditional Altivec registers.
6786 (vsx_xscvdpspn): Eliminate useless alternative constraint.
6787 (vsx_xscvspdpn): Likewise.
6788 (vsx_xscvspdpn_scalar): Likewise.
6790 2017-09-26 Martin Jambor <mjambor@suse.cz>
6792 * tree-sra.c (compare_access_positions): Put integral types first,
6793 stabilize sorting of integral types, remove conditions putting
6794 non-full-precision integers last.
6795 (sort_and_splice_var_accesses): Disable scalarization if a
6796 non-integert would be represented by a non-full-precision integer.
6798 2017-09-26 Joseph Myers <joseph@codesourcery.com>
6800 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
6801 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
6802 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
6803 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
6804 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
6805 conditionals inside the function instead of around it. Call
6806 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
6807 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
6809 2017-09-26 Richard Biener <rguenther@suse.de>
6811 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
6813 (scop_detection::build_scop_breadth): ... this. Removed.
6814 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
6815 (scop_detection::harmful_stmt_in_bb): Likewise.
6816 (scop_detection::graphite_can_represent_stmt): Likewise.
6817 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
6818 (scop_detection::can_represent_loop): Remove recursion, fold in ...
6819 (scop_detection::can_represent_loop_1): ... this. Removed.
6820 (scop_detection::harmful_loop_in_region): Simplify after inlining
6821 the above and remove more quadraticness.
6822 (build_scops): Adjust.
6823 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
6826 2017-09-26 Jakub Jelinek <jakub@redhat.com>
6829 * config/i386/i386.c (ix86_print_operand_address_as): Only test
6830 REGNO (base) == SP_REG if base is a REG.
6833 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6834 if it is different SSA_NAME.
6835 (optimize_range_tests_cmp_bitwise): New function.
6836 (optimize_range_tests): Call it.
6838 2017-09-26 Richard Biener <rguenther@suse.de>
6840 PR tree-optimization/82321
6841 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6842 for the def being inside the loop.
6844 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6846 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6848 * config/s390/s390-builtins.def: Fix constraint on op4.
6850 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6852 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6853 independent expanders.
6854 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6855 ("vec_ordered", "vec_unordered"): New expanders.
6857 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6859 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6862 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6864 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6865 vec_unpacks_lo_v16qi.
6866 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6868 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6870 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6871 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6872 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6874 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6876 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6878 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6880 ("*vec_slb<mode>"): New insn pattern.
6881 ("vec_shr_<mode>"): New expander.
6882 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6883 and force the shift count operand to V16QImode.
6884 ("vec_srb<mode>"): Set shift count mode to V16QI.
6886 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6888 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
6889 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
6890 ("vec_widen_smult_hi_<mode>"): New expander definitions.
6892 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
6895 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
6897 2017-09-26 Richard Biener <rguenther@suse.de>
6899 PR tree-optimization/82320
6900 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6903 2017-09-25 Jeff Law <law@redhat.com>
6905 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6906 prototype for new argument.
6907 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6908 mostly extracted from rs6000_emit_allocate_stack.
6909 (rs6000_emit_probe_stack_range_stack_clash): New function.
6910 (rs6000_emit_allocate_stack): Call
6911 rs6000_emit_probe_stack_range_stack_clash as needed.
6912 (rs6000_emit_probe_stack_range): Add additional argument
6913 to call to gen_probe_stack_range{si,di}.
6914 (output_probe_stack_range): New.
6915 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6916 (output_probe_stack_range_stack_clash): New.
6917 (rs6000_emit_prologue): Emit notes into dump file as requested.
6918 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6919 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6920 Add additional operand and pass it to output_probe_stack_range.
6922 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6924 PR tree-optimization/82163
6925 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6926 (checking_verify_loop_closed_ssa): New parameter.
6927 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6928 (check_loop_closed_ssa_stmt): Delete.
6929 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6930 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6931 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6934 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
6936 * brig-builtins.def: Treat HSAIL barrier builtins as
6937 setjmp/longjump style functions.
6939 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6941 * target.def (constant_alignment): New hook.
6942 * defaults.h (CONSTANT_ALIGNMENT): Delete.
6943 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
6944 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
6945 * doc/tm.texi: Regenerate.
6946 * targhooks.h (default_constant_alignment): Declare.
6947 (constant_alignment_word_strings): Likewise.
6948 * targhooks.c (default_constant_alignment): New function.
6949 (constant_alignment_word_strings): Likewise.
6950 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
6951 instead of CONSTANT_ALIGNMENT.
6952 * varasm.c (align_variable, get_variable_align, build_constant_desc)
6953 (force_const_mem): Likewise.
6954 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
6955 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
6956 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
6957 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6958 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
6960 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
6961 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6962 constant_alignment_word_strings.
6963 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
6964 (CONSTANT_ALIGNMENT): Likewise.
6965 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6966 (arm_constant_alignment): New function.
6967 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
6968 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6969 constant_alignment_word_strings.
6970 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
6971 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6972 constant_alignment_word_strings.
6973 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
6974 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6975 (cris_constant_alignment): New function.
6976 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
6977 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6978 (epiphany_constant_alignment): New function.
6979 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
6980 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6981 constant_alignment_word_strings.
6982 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6983 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6984 constant_alignment_word_strings.
6985 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6986 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6987 constant_alignment_word_strings.
6988 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
6989 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
6990 * config/i386/i386.c (ix86_constant_alignment): Make static.
6991 Use the same interface as the target hook.
6992 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6993 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
6994 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6995 constant_alignment_word_strings.
6996 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
6997 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
6998 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6999 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7000 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7001 constant_alignment_word_strings.
7002 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7003 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7004 constant_alignment_word_strings.
7005 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7006 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7007 constant_alignment_word_strings.
7008 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7009 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7011 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7012 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7013 * config/mips/mips.c (mips_constant_alignment): New function.
7014 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7015 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7016 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7017 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7018 (mmix_constant_alignment): Make static. Use the same interface
7020 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7021 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7022 constant_alignment_word_strings.
7023 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7024 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7025 constant_alignment_word_strings.
7026 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7027 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7028 constant_alignment_word_strings.
7029 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7030 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7031 (rs6000_constant_alignment): New function.
7032 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7033 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7034 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7035 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7036 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7037 (rs6000_constant_alignment): New function.
7038 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7039 * config/s390/s390.c (s390_constant_alignment): New function.
7040 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7041 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7042 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7043 constant_alignment_word_strings.
7044 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7045 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7046 (sparc_constant_alignment): New function.
7047 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7048 * config/spu/spu.c (spu_constant_alignment): New function.
7049 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7050 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7051 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7052 constant_alignment_word_strings.
7053 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7054 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7055 constant_alignment_word_strings.
7056 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7057 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7058 constant_alignment_word_strings.
7059 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7060 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7061 (visium_constant_alignment): New function.
7062 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7063 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7064 (xtensa_constant_alignment): New function.
7065 * system.h (CONSTANT_ALIGNMENT): Poison.
7067 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7069 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7070 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7071 (rs6000_builtin_valid_without_lhs): New helper function.
7072 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7073 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7075 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7077 * target.h (vec_perm_indices): Use unsigned short rather than
7079 (auto_vec_perm_indices): Likewise.
7080 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7081 Use unsigned int rather than unsigned char.
7082 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7084 2017-09-25 Richard Biener <rguenther@suse.de>
7086 * cfgloop.h (sort_sibling_loops): Declare.
7087 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7088 (sort_sibling_loops): New function sorting the sibling loop list
7090 * graphite.c (graphite_transform_loops): Sort sibling loops.
7092 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7094 * target.def (vec_perm_const_ok): Change sel parameter to
7096 * optabs-query.c (can_vec_perm_p): Update accordingly.
7097 * doc/tm.texi: Regenerate.
7098 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7099 auto_vec_perm_indices and remove separate nelt field.
7100 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7101 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7102 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7103 (aarch64_expand_vec_perm_const): Update accordingly.
7104 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7105 to vec_perm_indices.
7106 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7107 auto_vec_perm_indices and remove separate nelt field.
7108 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7109 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7110 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7112 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7113 to vec_perm_indices.
7114 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7115 sel to vec_perm_indices.
7116 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7117 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7118 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7120 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7123 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7126 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7127 on the FUNCTION_DECL function context if it has a DIE that is a
7130 2017-09-25 Richard Biener <rguenther@suse.de>
7132 PR tree-optimization/82285
7133 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7136 2017-09-25 Tom de Vries <tom@codesourcery.com>
7140 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7143 2017-09-25 Richard Biener <rguenther@suse.de>
7145 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7146 ISL errors other than isl_error_quota happen. Dump if the
7147 schedule is the same.
7148 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7149 errors instead of aborting inside ISL.
7151 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7154 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7155 of libgcc_eh for m64.
7156 * config/i386/darwin64.h: Likewise.
7158 2017-09-25 Richard Biener <rguenther@suse.de>
7161 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7162 attribute for incomplete types nor twice for complete ones.
7164 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7167 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7168 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7171 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7174 * config/i386/i386.opt (mprefer-avx256): Use
7175 ix86_target_flags variable.
7176 * config/i386/i386.c (ix86_target_string): Move
7177 -mprefer-avx256 to flag2_opts.
7179 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7182 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7183 and x != -1 | y != -1 into (x & y) != -1.
7185 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7187 * config.gcc: Add new case statement to set
7188 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7189 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7190 s390x-*-linux* case statements. Added aarch64 to the list of
7191 supported architectures.
7193 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7195 PR tree-optimization/82289
7196 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7197 STMT_VINFO_RELEVANT_P.
7199 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7200 Alan Hayward <alan.hayward@arm.com>
7201 David Sherwood <david.sherwood@arm.com>
7203 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7204 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7205 (extract_range_from_binary_expr_1): Don't call
7206 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7208 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7209 Alan Hayward <alan.hayward@arm.com>
7210 David Sherwood <david.sherwood@arm.com>
7212 * target.def (preferred_vector_alignment): New hook.
7213 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7215 * doc/tm.texi: Regenerate.
7216 * targhooks.h (default_preferred_vector_alignment): Declare.
7217 * targhooks.c (default_preferred_vector_alignment): New function.
7218 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7220 (DR_TARGET_ALIGNMENT): New macro.
7221 (aligned_access_p): Update commentary.
7222 (vect_known_alignment_in_bytes): New function.
7223 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7225 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7226 Calculate the misalignment based on the target alignment rather than
7228 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7229 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7230 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7231 the target alignment, rather than masking the element misalignment
7232 with the number of elements in a vector. Also use the target
7233 alignment when calculating the maximum number of peels.
7234 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7235 instead of TYPE_ALIGN_UNIT.
7236 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7237 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7238 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7239 (vect_create_data_ref_ptr): Likewise.
7240 (vect_setup_realignment): Realign by ANDing with
7241 -DR_TARGET_MISALIGNMENT.
7242 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7243 the number of peels based on DR_TARGET_ALIGNMENT.
7244 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7245 with the guaranteed alignment boundary when deciding whether
7247 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7248 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7249 (ensure_base_align): Remove stmt_info parameter. Get the
7250 target base alignment from DR_TARGET_ALIGNMENT.
7251 (vectorizable_store): Update call accordingly. Interpret
7252 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7254 (vectorizable_load): Likewise.
7256 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7257 Alan Hayward <alan.hayward@arm.com>
7258 David Sherwood <david.sherwood@arm.com>
7260 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7261 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7262 (vect_enhance_data_refs_alignment): Likewise.
7264 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7266 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7267 error. Only quit immediately if parsing is complete.
7268 (BEGIN): Initialize fatal_err and parse_done.
7269 (begin fpu, end fpu): Check number of arguments.
7270 (begin arch, end arch): Likewise.
7271 (begin cpu, end cpu): Likewise.
7272 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7273 (optalias): Likewise.
7275 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7277 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7278 * config/arm/arm-isa.h: Delete. Move definitions to ...
7279 * arm-cpus.in: ... here. Use new feature and fgroup values.
7280 * config/arm/arm.c (arm_option_override): Use lower case for feature
7282 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7283 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7284 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7286 (print_isa_bits_for): New function.
7287 (gen_isa): New function.
7288 (gen_comm_data): Use print_isa_bits_for.
7289 (define feature): New keyword.
7290 (define fgroup): New keyword.
7291 * config/arm/t-arm (TM_H): Remove.
7292 (GTM_H): Add arm-isa.h.
7293 (arm-isa.h): Add rule to generate file.
7294 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7295 case for feature bit names.
7297 2017-09-22 Richard Biener <rguenther@suse.de>
7299 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7301 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7302 print of no dependency loops ...
7303 * graphite.c (graphite_transform_loops): ... here.
7304 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7306 (same_close_phi_node, remove_duplicate_close_phi,
7307 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7308 (canonicalize_loop_closed_ssa): ... here and simplify.
7309 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7310 (optimize_isl): Use dump_printf_loc to tell when we stopped
7311 optimizing because of an ISL timeout.
7313 2017-09-22 Richard Biener <rguenther@suse.de>
7315 PR tree-optimization/82291
7316 * tree-if-conv.c (predicate_mem_writes): Make sure to
7317 remove writes in blocks predicated with false.
7319 2017-09-22 Richard Biener <rguenther@suse.de>
7321 * sese.c: Include cfganal.h.
7322 (if_region_set_false_region): Remove.
7323 (create_if_region_on_edge): Likewise.
7324 (move_sese_in_condition): Re-implement without destroying
7327 2017-09-22 Richard Biener <rguenther@suse.de>
7329 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7330 Verify both BBs contain loop PHI nodes before dispatching to
7332 (graphite_regenerate_ast_isl): Do not recompute dominators,
7333 do not verify three times. Restructure for clarity.
7334 * graphite-scop-detection.c (same_close_phi_node,
7335 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7336 defined_in_loop_p, canonicalize_loop_closed_ssa,
7337 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7338 checking and SSA rewrite, move to ...
7339 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7340 (graphite_initialize): Do not pass in ctx, do not reset the
7341 SCEV cache, compute only dominators.
7342 (graphite_transform_loops): Allocate ISL ctx after
7343 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7344 Maintain post-dominators only around build_scops.
7345 * sese.c (if_region_set_false_region): Make static. Free
7346 and recompute dominators.
7347 (move_sese_in_condition): Assert we don't get called with
7348 post-dominators computed.
7349 * sese.h (if_region_set_false_region): Remove.
7351 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7353 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7354 mode attribute for TARGET_AVX512VL.
7356 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7358 * config/i386/i386.opt (mprefer-avx256): New option.
7359 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7361 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7362 for TARGET_PREFER_AVX256.
7363 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7365 2017-09-21 Jeff Law <law@redhat.com>
7367 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7368 Fix dump output if the only stack space is for pushed registers.
7370 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7372 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7375 2017-09-21 Martin Sebor <msebor@redhat.com>
7378 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7379 code (in C++) or code that triggers warnings.
7381 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7383 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
7385 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7387 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
7388 * sched-rgn.c: Ditto.
7389 * sel-sched-ir.c: Ditto.
7391 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
7393 * toplev.h (set_random_seed): Adjust return type.
7394 * toplev.c (init_local_tick): Move eager initialization of random_seed
7395 to get_random_seed. Adjust comment.
7396 (init_random_seed): Inline to get_random_seed, delete.
7397 (get_random_seed): Initialize random_seed lazily.
7398 (set_random_seed): Do not return previous value.
7399 (print_switch_value): Do not call get_random_seed.
7401 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
7403 * cgraph.c (delete_function_version): New, broken out from...
7404 (cgraph_node::delete_function_version): ...here. Rename to
7405 cgraph_node::delete_function_version_by_decl. Update all uses.
7406 (cgraph_node::remove): Call delete_function_version.
7408 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7411 * tree-inline.c (expand_call_inline): Emit clobber stmts for
7412 VAR_DECLs to which addressable non-volatile parameters are mapped
7413 and for id->retvar after the return value assignment. Clear
7414 id->retval and id->retbnd after inlining.
7416 2017-09-21 Richard Biener <rguenther@suse.de>
7418 PR tree-optimization/82276
7419 PR tree-optimization/82244
7420 * tree-vrp.c (build_assert_expr_for): Set
7421 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
7423 (remove_range_assertions): Revert earlier change.
7425 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
7428 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
7430 2017-09-21 Richard Biener <rguenther@suse.de>
7432 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7433 Restore valid IL after code generation errors.
7434 * graphite.c (graphite_transform_loops): Diagnose code
7435 generation issues as MSG_MISSED_OPTIMIZATION and continue
7436 with processing SCOPs.
7438 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7439 Alan Hayward <alan.hayward@arm.com>
7440 David Sherwood <david.sherwood@arm.com>
7442 * calls.c (compute_argument_addresses): Use simplify_gen_binary
7443 rather than choosing between plus_constant and gen_rtx_<CODE>.
7444 * expr.c (emit_push_insn): Likewise.
7445 (expand_expr_real_2): Likewise.
7447 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7448 Alan Hayward <alan.hayward@arm.com>
7449 David Sherwood <david.sherwood@arm.com>
7451 * loop-unroll.c (split_iv): Call copy_rtx on the step.
7453 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7454 Alan Hayward <alan.hayward@arm.com>
7455 David Sherwood <david.sherwood@arm.com>
7457 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
7458 calling tree_to_uhwi.
7460 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7461 Alan Hayward <alan.hayward@arm.com>
7462 David Sherwood <david.sherwood@arm.com>
7464 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
7465 INTEGER_CST rather than a negative test for ADDR_EXPR.
7467 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7468 Alan Hayward <alan.hayward@arm.com>
7469 David Sherwood <david.sherwood@arm.com>
7471 * tree-vrp.c (extract_range_from_binary_expr_1): Check
7472 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
7474 2017-09-21 Richard Biener <rguenther@suse.de>
7476 PR tree-optimization/71351
7477 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
7478 graphite_create_new_loop_guard): Remove, fold remaining parts
7480 (translate_isl_ast_node_for): ... here and simplify.
7482 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7485 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
7486 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
7487 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
7488 alternative always use QI mode, for -Os imov (=R,R) alternative
7489 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
7492 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7493 Jeff Law <law@redhat.com>
7495 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
7496 (allocate_stack_space): New function, partially extracted from
7498 (s390_emit_prologue): Track offset to most recent stack probe.
7499 Code to allocate space moved into allocate_stack_space.
7500 Dump actions when no stack is allocated.
7501 (s390_prologue_plus_offset): New function.
7502 (s390_emit_stack_probe): Likewise.
7504 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
7506 * common.opt (Wa, Wl, Wp, g, gz=): Add
7508 (gno-column-info): Remove.
7509 (gcolumn-info): Drop RejectNegative.
7511 (gno-record-gcc-switches): Remove.
7512 (grecord-gcc-switches): Drop RejectNegative.
7513 (gno-split-dwarf): Remove.
7514 (gsplit-dwarf): Drop RejectNegative.
7515 (gno-strict-dwarf): Remove.
7516 (gstrict-dwarf): Drop RejectNegative.
7517 * config/darwin.opt (gfull, gused): Add RejectNegative.
7518 * dwarf2out.c (gen_producer_string): Drop
7519 gno-record-gcc-switches handler.
7520 * optc-gen.awk: Add g to prefixes with negative forms.
7521 * opts-common.c (remapping_prefix_p): New.
7522 (find_opt): Check it.
7523 (generate_canonical_option): Test g prefix.
7524 (option_map): Add -gno- mapping.
7525 (add_misspelling_candidates): Check remapping_prefix_p.
7527 2017-09-20 Jeff Law <law@redhat.com>
7529 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
7530 thinko in stack clash protection support.
7532 * explow.c (compute_stack_clash_protection_loop_data): Use
7533 CONST_INT_P instead of explicit test. Verify object is a
7534 CONST_INT_P before looking at INTVAL.
7535 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
7536 instead of explicit test.
7538 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7541 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
7542 address instead of to r1 and r11.
7544 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
7546 * config.gcc: Support "knm".
7547 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
7548 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7550 * config/i386/i386.c (m_KNM): Define.
7551 (processor_target_table): Add "knm".
7553 (ix86_option_override_internal): Add "knm".
7554 (ix86_issue_rate): Add PROCESSOR_KNM.
7555 (ix86_adjust_cost): Ditto.
7556 (ia32_multipass_dfa_lookahead): Ditto.
7557 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
7558 (fold_builtin_cpu): Add M_INTEL_KNM.
7559 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
7560 (processor_type): Add PROCESSOR_KNM.
7561 * config/i386/x86-tune.def: Add m_KNM.
7562 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
7564 2017-09-20 Richard Biener <rguenther@suse.de>
7566 PR tree-optimization/80213
7567 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
7568 are allowed in empty BBs as well.
7569 (canonicalize_loop_closed_ssa): Also look for other complex
7571 (scop_detection::get_sese): Include the loop-closed PHI block
7573 (scop_detection::merge_sese): Remove code adding extra blocks.
7574 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
7575 (build_scops): Assert the final returned scop is invalid.
7577 2017-09-20 Richard Biener <rguenther@suse.de>
7579 PR tree-optimization/82264
7580 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
7582 (vn_phi_lookup): Likewise.
7583 (vn_phi_insert): Likewise.
7585 2017-09-20 Jakub Jelinek <jakub@redhat.com>
7587 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
7588 that fits into uhwi or shwi, add DW_AT_const_value regardless
7589 of early_dwarf without going through RTL, using add_AT_unsigned
7592 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
7593 (DEBUG_LTO_ABBREV_SECTION): Likewise.
7594 (DEBUG_LTO_MACINFO_SECTION): Likewise.
7595 (DEBUG_MACRO_SECTION): Likewise.
7596 (DEBUG_LTO_MACRO_SECTION): Likewise.
7597 (DEBUG_STR_DWO_SECTION): Likewise.
7598 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
7599 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
7600 (DEBUG_LTO_DWO_LINE_SECTION): Define.
7601 (DEBUG_LTO_LINE_STR_SECTION): Define.
7602 (init_sections_and_labels): Initialize debug_line_str_section
7603 variable. Initialize debug_loc_section for -gdwarf-5 to
7604 DEBUG_LOCLISTS_SECTION. Formatting fixes.
7606 2017-09-20 Richard Biener <rguenther@suse.de>
7608 * graphite-sese-to-poly.c (extract_affine): Properly handle
7609 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
7611 2017-09-20 Richard Biener <rguenther@suse.de>
7613 PR tree-optimization/81373
7614 * graphite-scop-detection.c (build_cross_bb_scalars_def):
7615 Force SESE live-out defs to be handled even if they are
7618 2017-09-19 Jeff Law <law@redhat.com>
7620 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
7621 nothing for stack adjustments with REG_STACK_CHECK.
7622 * sched-deps.c (parse_add_or_inc): Reject insns with
7623 REG_STACK_CHECK from dependency breaking.
7624 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
7625 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
7626 * reg-notes.def (STACK_CHECK): New note.
7628 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
7629 (ix86_expand_prologue): Dump stack clash info as needed.
7630 Call ix86_adjust_stack_and_probe_stack_clash as needed.
7632 * function.c (dump_stack_clash_frame_info): New function.
7633 * function.h (dump_stack_clash_frame_info): Prototype.
7634 (enum stack_clash_probes): New enum.
7636 * config/alpha/alpha.c (alpha_expand_prologue): Also check
7637 flag_stack_clash_protection.
7638 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
7639 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
7640 (arm_frame_pointer_required): Likewise.
7641 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
7642 (ia64_expand_prologue): Likewise.
7643 * config/mips/mips.c (mips_expand_prologue): Likewise.
7644 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
7645 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7646 (sparc_flat_expand_prologue): Likewise.
7647 * config/spu/spu.c (spu_expand_prologue): Likewise.
7649 * explow.c: Include "params.h".
7650 (anti_adjust_stack_and_probe_stack_clash): New function.
7651 (get_stack_check_protect): Likewise.
7652 (compute_stack_clash_protection_loop_data): Likewise.
7653 (emit_stack_clash_protection_loop_start): Likewise.
7654 (emit_stack_clash_protection_loop_end): Likewise.
7655 (allocate_dynamic_stack_space): Use get_stack_check_protect.
7656 Use anti_adjust_stack_and_probe_stack_clash.
7657 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
7658 (emit_stack_clash_protection_loop_start): Likewise.
7659 (emit_stack_clash_protection_loop_end): Likewise.
7660 * rtl.h (get_stack_check_protect): Prototype.
7661 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
7662 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
7663 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
7665 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
7667 * doc/tm.texi: Rebuilt.
7668 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
7669 get_stack_check_protect.
7670 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
7671 * config/arm/arm.c (arm_expand_prologue): Likewise.
7672 (arm_frame_pointer_required): Likewise.
7673 * config/i386/i386.c (ix86_expand_prologue): Likewise.
7674 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
7675 * config/mips/mips.c (mips_expand_prologue): Likewise.
7676 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
7677 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
7678 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7679 (sparc_flat_expand_prologue): Likewise.
7681 * common.opt (-fstack-clash-protection): New option.
7682 * flag-types.h (enum stack_check_type): Note difference between
7683 -fstack-check= and -fstack-clash-protection.
7684 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
7685 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
7686 * toplev.c (process_options): Issue warnings/errors for cases
7687 not handled with -fstack-clash-protection.
7688 * doc/invoke.texi (-fstack-clash-protection): Document new option.
7689 (-fstack-check): Note additional problem with -fstack-check=generic.
7690 Note that -fstack-check is primarily for Ada and refer users
7691 to -fstack-clash-protection for stack-clash-protection.
7692 Document new params for stack clash protection.
7694 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
7696 * config/i386/i386.c (ix86_split_long_move): Do not handle
7697 address used for LEA in a special way.
7699 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
7701 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
7703 2017-09-19 Martin Sebor <msebor@redhat.com>
7706 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
7707 of incompatible types.
7709 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
7711 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7712 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
7713 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7714 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
7716 2017-09-19 Richard Biener <rguenther@suse.de>
7718 PR tree-optimization/82244
7719 * tree-vrp.c (remove_range_assertions): Do not propagate
7720 a constant to abnormals but replace the assert with a copy.
7722 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
7724 PR rtl-optimization/57878
7725 PR rtl-optimization/68988
7726 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
7727 avoidance test involving non_reload_pseudos. Move frequency test
7728 below the general fragmentation avoidance test.
7730 2017-09-19 Richard Biener <rguenther@suse.de>
7732 PR tree-optimization/69728
7733 * graphite-sese-to-poly.c (schedule_error): New global.
7734 (add_loop_schedule): Handle empty domain by failing the
7736 (build_original_schedule): Handle schedule_error.
7738 2017-09-19 Richard Biener <rguenther@suse.de>
7740 * graphite-scop-detection.c (scop_detection::can_represent_loop):
7741 Do not iterate to sibling loops but only to siblings of inner
7744 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
7747 * config/m68k/m68k.md (moveq feeding equality comparison): Check
7748 that the registers are different.
7750 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
7752 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
7753 to processor_model and "amdfam17h" to arch_names_table.
7754 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
7756 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7759 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
7761 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7762 Alan Hayward <alan.hayward@arm.com>
7763 David Sherwood <david.sherwood@arm.com>
7765 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
7767 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
7769 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
7770 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
7771 vect_schedule_slp_instance.
7772 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
7773 Update call to vect_slp_analyze_node_operations. Simplify return
7775 (vect_slp_analyze_bb_1): Update call accordingly.
7776 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
7777 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
7778 (vect_schedule_slp): Update call accordingly.
7780 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7781 Alan Hayward <alan.hayward@arm.com>
7782 David Sherwood <david.sherwood@arm.com>
7784 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
7785 with types that aren't in fact scalar.
7787 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7789 * tree-vect-slp.c (vect_record_max_nunits): New function,
7791 (vect_build_slp_tree_1): ...here.
7792 (vect_build_slp_tree_2): Call it for phis too.
7794 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7796 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
7797 to vect_get_vec_def_for_operand when getting the mask operand.
7799 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7800 Alan Hayward <alan.hayward@arm.com>
7801 David Sherwood <david.sherwood@arm.com>
7803 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
7806 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7807 Alan Hayward <alan.hayward@arm.com>
7808 David Sherwood <david.sherwood@arm.com>
7810 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
7811 calculation for vector booleans.
7813 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7814 Alan Hayward <alan.hayward@arm.com>
7815 David Sherwood <david.sherwood@arm.com>
7817 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
7819 (vect_transform_stmt): ...here.
7820 (vect_analyze_stmt): Use it instead of calling
7821 vectorizable_live_operation directly.
7823 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
7825 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
7826 non-SIMT targets in acc vector loops.
7828 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
7830 * configure.ac: Add arc and check if assembler supports gdwarf2.
7831 * configure: Regenerate.
7833 2017-09-18 Richard Biener <rguenther@suse.de>
7835 PR tree-optimization/82220
7836 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7837 epilogue niters from the min_profitable_iters compute.
7839 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7842 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7843 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7844 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7846 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7849 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7850 switching to a new text section.
7852 2017-09-18 Richard Biener <rguenther@suse.de>
7854 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7856 (build_alias_set): Reject aliases with no access function.
7858 2017-09-18 Richard Biener <rguenther@suse.de>
7860 PR tree-optimization/79622
7861 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7863 (build_cross_bb_scalars_use): Likewise.
7865 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7867 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7869 2017-09-18 Alan Modra <amodra@gmail.com>
7872 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7873 stack_pointer_rtx for count 0. Update comments. Break up
7874 large rtl expression.
7876 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7878 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7879 Increase to 20 bytes.
7880 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7881 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7882 or avx version of the stub.
7884 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7887 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
7888 compute the minimum stack alignment. Also update preferred stack
7889 boundary for leaf functions.
7891 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
7893 PR tree-optimization/82228
7894 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
7897 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7899 * common/config/nds32/nds32-common.c
7900 (nds32_option_optimization_table): Refine formatting.
7901 (nds32_option_optimization_table): Use -fsched-pressure and
7902 -fomit-frame-pointer for specific optimization level.
7904 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7906 * config/nds32/nds32.c: Refine formatting and comments.
7907 * config/nds32/nds32.h: Likewise.
7908 * config/nds32/nds32.md: Likewise.
7909 * config/nds32/nds32-cost.c: Likewise.
7910 * config/nds32/nds32-isr.c: Likewise.
7911 * config/nds32/nds32-md-auxiliary.c: Likewise.
7912 * config/nds32/nds32-multiple.md: Likewise.
7913 * config/nds32/nds32-predicates.c: Likewise.
7915 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7916 Jakub Jelinek <jakub@redhat.com>
7918 Add support for -std=c++2a.
7919 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7921 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7923 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7926 * doc/extend.texi (Common Function Attributes): Add
7927 references to ARM, AArch64, and S/390 specific attributes.
7928 (Function Specific Option Pragmas): Add AArch64 and S/390
7929 to list of back ends that support the target pragma.
7931 2017-09-15 Nathan Sidwell <nathan@acm.org>
7933 * doc/standards.texi: Fix C++17 description. Update URLs for
7936 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7938 * common.opt (Wcast-align=strict): New warning option.
7939 * doc/invoke.texi: Document -Wcast-align=strict.
7941 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
7943 * cgraph.h (cgraph_thunk_info): Add comments.
7944 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
7945 assert for VIRTUAL_* arguments stricter.
7947 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
7949 PR tree-optimization/71026
7950 * match.pd: Move RDIV patterns from fold-const.c
7951 * fold-const.c (distribute_real_division): Removed.
7952 (fold_binary_loc): Remove calls to distribute_real_divison.
7954 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7956 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
7957 c++1z and gnu++1z as deprecated. Change other references to
7958 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
7959 Change -Wc++1z-compat to -Wc++17-compat.
7960 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
7961 * dwarf2out.c (highest_c_language): Handle C++17.
7962 (gen_compile_unit_die): Likewise.
7964 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7966 PR rtl-optimization/82192
7967 * combine.c (make_extraction): Don't look through non-paradoxical
7968 SUBREGs or TRUNCATE if pos + len is or might be bigger than
7971 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7972 Alan Hayward <alan.hayward@arm.com>
7973 David Sherwood <david.sherwood@arm.com>
7975 * target.def (function_arg_offset): New hook.
7976 * targhooks.h (default_function_arg_offset): Declare.
7977 * targhooks.c (default_function_arg_offset): New function.
7978 * function.c (locate_and_pad_parm): Use
7979 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
7980 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
7981 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7982 * doc/tm.texi: Regenerate.
7983 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7984 * config/spu/spu.c (spu_function_arg_offset): New function.
7985 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7986 * system.h (FUNCTION_ARG_OFFSET): Poison.
7988 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7989 Alan Hayard <alan.hayward@arm.com>
7990 David Sherwood <david.sherwood@arm.com>
7992 * target.def (truly_noop_truncation): New hook.
7993 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
7994 than TRULY_NOOP_TRUNCATION.
7995 * hooks.h (hook_bool_uint_uint_true): Declare.
7996 * hooks.c (hook_bool_uint_uint_true): New function.
7997 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
7998 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
7999 * doc/tm.texi: Regenerate.
8000 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8001 rather than TRULY_NOOP_TRUNCATION in comments.
8002 (simplify_comparison): Likewise.
8003 (record_truncated_value): Likewise.
8004 * expmed.c (extract_bit_field_1): Likewise.
8005 (extract_split_bit_field): Likewise.
8006 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8007 instead of TRULY_NOOP_TRUNCATION.
8008 * function.c (assign_parm_setup_block): Likewise.
8009 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8010 * rtlhooks.c: Include target.h.
8011 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8012 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8013 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8014 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8015 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8016 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8017 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8018 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8019 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8020 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8021 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8022 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8023 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8024 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8025 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8026 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8027 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8028 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8029 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8030 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8031 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8032 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8033 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8034 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8035 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8036 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8037 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8038 rather than TRULY_NOOP_TRUNCATION in comments.
8039 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8040 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8041 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8042 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8043 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8044 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8045 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8046 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8047 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8048 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8049 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8050 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8051 rather than TRULY_NOOP_TRUNCATION in comments.
8052 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8053 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8054 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8055 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8056 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8057 TRULY_NOOP_TRUNCATION condition.
8058 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8059 (TRULY_NOOP_TRUNCATION): Delete.
8060 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8061 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8062 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8063 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8064 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8065 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8066 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8067 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8068 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8069 rather than TRULY_NOOP_TRUNCATION in comments.
8070 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8071 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8072 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8073 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8074 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8075 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8077 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8080 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8081 (*cmp_ior): Likewise.
8082 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8083 (*ior_scc_scc_cmp): Likewise.
8084 (*and_scc_scc): Likewise.
8085 (*and_scc_scc_cmp): Likewise.
8087 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8088 Alan Hayard <alan.hayward@arm.com>
8089 David Sherwood <david.sherwood@arm.com>
8091 * target.def (can_change_mode_class): New hook.
8092 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8093 (hard_regno_nregs): Likewise.
8094 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8095 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8096 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8097 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8098 (LOAD_EXTEND_OP): Update accordingly.
8099 * doc/tm.texi: Regenerate.
8100 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8101 CANNOT_CHANGE_MODE_CLASS.
8102 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8103 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8104 * combine.c (simplify_set): Update accordingly.
8105 * emit-rtl.c (validate_subreg): Likewise.
8106 * recog.c (general_operand): Likewise.
8107 * regcprop.c (mode_change_ok): Likewise.
8108 * reload1.c (choose_reload_regs): Likewise.
8109 (inherit_piecemeal_p): Likewise.
8110 * rtlanal.c (simplify_subreg_regno): Likewise.
8111 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8112 instead of CANNOT_CHANGE_MODE_CLASS.
8113 (reload_cse_simplify_operands): Likewise.
8114 * reload.c (push_reload): Use targetm.can_change_mode_class
8115 instead of CANNOT_CHANGE_MODE_CLASS.
8116 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8117 REG_CANNOT_CHANGE_MODE_P.
8118 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8119 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8120 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8121 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8122 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8123 (arm_can_change_mode_class): New function.
8124 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8125 than CANNOT_CHANGE_MODE_CLASS in comments.
8126 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8127 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8128 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8129 (ix86_can_change_mode_class): ...this new function, inverting the
8130 sense of the return value.
8131 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8132 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8133 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8134 (ia64_can_change_mode_class): New function.
8135 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8136 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8137 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8138 (m32c_can_change_mode_class): ...this new function, inverting the
8139 sense of the return value.
8140 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8141 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8142 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8143 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8144 (mips_can_change_mode_class): ...this new function, inverting the
8145 sense of the return value.
8146 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8147 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8148 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8149 (msp430_can_change_mode_class): New function.
8150 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8151 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8152 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8153 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8154 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8155 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8156 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8157 (pa_cannot_change_mode_class): Replace with...
8158 (pa_can_change_mode_class): ...this new function, inverting the
8159 sense of the return value.
8160 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8161 than CANNOT_CHANGE_MODE_CLASS in comments.
8162 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8163 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8164 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8165 (pdp11_cannot_change_mode_class): Replace with...
8166 (pdp11_can_change_mode_class): ...this new function, inverting the
8167 sense of the return value.
8168 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8169 * config/powerpcspe/powerpcspe-protos.h
8170 (rs6000_cannot_change_mode_class_ptr): Delete.
8171 * config/powerpcspe/powerpcspe.c
8172 (rs6000_cannot_change_mode_class_ptr): Delete.
8173 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8174 (rs6000_option_override_internal): Assign to
8175 targetm.can_change_mode_class instead of
8176 rs6000_cannot_change_mode_class_ptr.
8177 (rs6000_cannot_change_mode_class): Replace with...
8178 (rs6000_can_change_mode_class): ...this new function, inverting the
8179 sense of the return value.
8180 (rs6000_debug_cannot_change_mode_class): Replace with...
8181 (rs6000_debug_can_change_mode_class): ...this new function.
8182 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8183 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8184 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8185 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8186 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8188 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8189 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8190 (rs6000_option_override_internal): Assign to
8191 targetm.can_change_mode_class instead of
8192 rs6000_cannot_change_mode_class_ptr.
8193 (rs6000_cannot_change_mode_class): Replace with...
8194 (rs6000_can_change_mode_class): ...this new function, inverting the
8195 sense of the return value.
8196 (rs6000_debug_cannot_change_mode_class): Replace with...
8197 (rs6000_debug_can_change_mode_class): ...this new function.
8198 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8199 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8200 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8201 (s390_can_change_mode_class): ...this new function, inverting the
8202 sense of the return value.
8203 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8204 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8205 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8206 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8207 (sh_cannot_change_mode_class): Replace with...
8208 (sh_can_change_mode_class): ...this new function, inverting the
8209 sense of the return value.
8210 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8211 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8212 (sparc_can_change_mode_class): New function.
8213 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8214 * config/spu/spu.c (spu_can_change_mode_class): New function.
8215 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8216 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8217 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8218 (visium_can_change_mode_class): New function.
8219 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8221 2017-09-15 Richard Biener <rguenther@suse.de>
8223 PR tree-optimization/82217
8224 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8225 but not undefined case.
8227 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8230 * postreload.c (reload_cse_simplify_operands): Skip
8231 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8233 2017-09-15 Richard Biener <rguenther@suse.de>
8235 PR tree-optimization/68823
8236 * graphite-scop-detection.c (build_alias_set): If we have a
8237 possible dependence check whether we can handle them by just
8238 looking at the DRs DR_ACCESS_FNs.
8239 (build_scops): If build_alias_set fails, fail the SCOP.
8241 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8243 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8244 to support float128 built-in functions that require the ISA 3.0
8246 (BU_FLOAT128_3_HW): Likewise.
8247 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8249 (FMAF128): Likewise.
8251 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8252 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8253 floating point instructions.
8254 (rs6000_invalid_builtin): Likewise.
8255 (rs6000_builtin_mask_names): Likewise.
8256 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8257 (RS6000_BTM_FLOAT128_HW): Likewise.
8258 (RS6000_BTM_COMMON): Likewise.
8259 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8261 * doc/extend.texi (RS/6000 built-in functions): Document the
8262 IEEE 128-bit floating point square root and fused multiply-add
8265 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8267 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8268 reg (r2) isn't in the set of registers defined in the prologue.
8270 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8271 Alan Hayward <alan.hayward@arm.com>
8272 David Sherwood <david.sherwood@arm.com>
8274 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8275 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8276 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8277 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8278 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8280 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8281 max_vectorization_factor.
8282 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8284 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8285 Alan Hayward <alan.hayward@arm.com>
8286 David Sherwood <david.sherwood@arm.com>
8288 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8289 (vect_worthwhile_without_simd_p): Declare.
8290 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8291 (vectorizable_reduction): Use it.
8292 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8293 (vectorizable_operation): Likewise.
8295 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8296 Alan Hayward <alan.hayward@arm.com>
8297 David Sherwood <david.sherwood@arm.com>
8299 * tree-vectorizer.h (vect_get_num_copies): New function.
8300 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8301 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8302 (vectorizable_induction): Likewise.
8303 (vectorizable_live_operation): Likewise.
8304 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8305 (vectorizable_bswap): Likewise.
8306 (vectorizable_call): Likewise.
8307 (vectorizable_conversion): Likewise.
8308 (vectorizable_assignment): Likewise.
8309 (vectorizable_shift): Likewise.
8310 (vectorizable_operation): Likewise.
8311 (vectorizable_store): Likewise.
8312 (vectorizable_load): Likewise.
8313 (vectorizable_condition): Likewise.
8314 (vectorizable_comparison): Likewise.
8315 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8317 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8318 Alan Hayward <alan.hayward@arm.com>
8319 David Sherwood <david.sherwood@arm.com>
8321 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8322 of vect_init_vector.
8324 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8325 Alan Hayward <alan.hayward@arm.com>
8326 David Sherwood <david.sherwood@arm.com>
8328 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8329 an inline wrapper that provides a location.
8330 (gimple_build_vector): Likewise.
8331 * gimple-fold.c (gimple_build_vector_from_val): New function.
8332 (gimple_build_vector): Likewise.
8333 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8334 functions to build the initial value. Always return a gimple value.
8335 (get_initial_defs_for_reduction): Likewise. Only compute
8337 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8338 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8339 (vectorizable_induction): Use gimple_build_vector rather than
8342 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8343 Alan Hayward <alan.hayward@arm.com>
8344 David Sherwood <david.sherwood@arm.com>
8346 * target.h (vec_perm_indices): New typedef.
8347 (auto_vec_perm_indices): Likewise.
8348 * optabs-query.h: Include target.h
8349 (can_vec_perm_p): Take a vec_perm_indices *.
8350 * optabs-query.c (can_vec_perm_p): Likewise.
8351 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8352 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8353 * tree-vect-generic.c (lower_vec_perm): Likewise.
8354 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8355 (vect_grouped_load_supported): Likewise.
8356 (vect_shift_permute_load_chain): Likewise.
8357 (vect_permute_store_chain): Use auto_vec_perm_indices.
8358 (vect_permute_load_chain): Likewise.
8359 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8360 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8361 Update uses of can_vec_perm_p.
8362 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8363 mode with a number of elements. Take a vec_perm_indices *.
8364 (vect_create_epilog_for_reduction): Update accordingly.
8365 Use auto_vec_perm_indices.
8366 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8367 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8368 (vect_transform_slp_perm_load): Likewise.
8369 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8370 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8371 (vect_gen_perm_mask_checked): Likewise.
8372 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8373 (vect_gen_perm_mask_checked): Likewise.
8374 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8375 (vectorizable_store): Likewise.
8376 (vectorizable_load): Likewise.
8377 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8378 (vectorizable_bswap): Likewise.
8380 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8381 Alan Hayward <alan.hayward@arm.com>
8382 David Sherwood <david.sherwood@arm.com>
8384 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
8385 * tree.c (build_vector): Likewise.
8386 (build_vector_from_ctor): Update accordingly.
8387 (build_vector_from_val): Likewise.
8388 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8389 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8390 * tree-vect-generic.c (add_rshift): Likewise.
8391 (expand_vector_divmod): Likewise.
8392 (optimize_vector_constructor): Likewise.
8393 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
8394 (vect_transform_slp_perm_load): Likewise.
8395 (vect_schedule_slp_instance): Likewise.
8396 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
8397 (vectorizable_call): Likewise.
8398 (vect_gen_perm_mask_any): Likewise. Add elements in order.
8399 * expmed.c (make_tree): Likewise.
8400 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
8401 a vector passed to build_vector.
8402 (fold_convert_const): Likewise.
8403 (exact_inverse): Likewise.
8404 (fold_ternary_loc): Likewise.
8405 (fold_relational_const): Likewise.
8406 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
8407 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
8408 (const_unop): Likewise. Store the reduction accumulator in a
8409 variable rather than an array.
8410 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
8411 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
8412 the new vector, rather than constructing it after the input arrays.
8413 (native_interpret_vector): Use auto_vec<tree> when building
8414 a vector passed to build_vector. Add elements in order.
8415 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
8416 auto_vec<tree> when building a vector passed to build_vector.
8417 (vect_create_epilog_for_reduction): Likewise.
8418 (vectorizable_induction): Likewise.
8419 (get_initial_def_for_reduction): Likewise. Fix indentation of
8421 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
8423 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
8424 passed to build_vector.
8426 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8427 Alan Hayward <alan.hayward@arm.com>
8428 David Sherwood <david.sherwood@arm.com>
8430 * tree-core.h (tree_base::u): Add an "nelts" field.
8431 (tree_vector): Use VECTOR_CST_NELTS as the length.
8432 * tree.c (tree_size): Likewise.
8433 (make_vector): Initialize VECTOR_CST_NELTS.
8434 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
8435 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
8436 TYPE_VECTOR_SUBPARTS.
8437 * expr.c (const_vector_mask_from_tree): Consistently use "units"
8438 as the number of units, setting it from VECTOR_CST_NELTS.
8439 (const_vector_from_tree): Likewise.
8440 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
8441 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
8442 (fold_negate_expr_1): Likewise.
8443 (fold_convert_const): Likewise.
8444 (const_binop): Likewise. Differentiate the number of output and
8446 (const_unop): Likewise.
8447 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
8448 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
8449 in cases that did the opposite.
8451 2017-09-14 Richard Biener <rguenther@suse.de>
8453 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
8456 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
8458 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
8460 2017-09-14 Jakub Jelinek <jakub@redhat.com>
8463 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
8464 if and where to split a bb, except for splitting before debug insn
8465 sequences followed by non-label real insn. Delete debug insns
8466 in between basic blocks.
8468 * combine.c (make_compound_operation_int): Formatting fixes.
8470 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
8471 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8472 * config/netbsd.h (LINK_EH_SPEC): Likewise.
8473 * config/sol2.h (LINK_EH_SPEC): Likewise.
8474 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8475 * config/s390/linux.h (LINK_SPEC): Likewise.
8476 * config/freebsd.h (LINK_EH_SPEC): Likewise.
8477 * config/openbsd.h (LINK_EH_SPEC): Likewise.
8478 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8479 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
8480 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
8481 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8482 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
8483 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
8485 2017-09-13 Jakub Jelinek <jakub@redhat.com>
8487 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
8489 (ENDFILE_LINUX_SPEC): Likewise.
8490 (LINK_EH_SPEC): Likewise.
8491 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
8492 (LINK_OS_LINUX_SPEC32): Likewise.
8493 (LINK_OS_LINUX_SPEC64): Likewise.
8494 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
8495 (LINK_OS_LINUX_SPEC): Likewise.
8497 2017-09-13 Martin Liska <mliska@suse.cz>
8500 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
8501 CASE_HIGH is NULL_TREE.
8503 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8504 Alan Hayward <alan.hayward@arm.com>
8505 David Sherwood <david.sherwood@arm.com>
8507 * target.def (secondary_memory_needed): New hook.
8508 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
8509 instead of SECONDARY_MEMORY_NEEDED.
8510 (secondary_memory_needed_mode): Likewise.
8511 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
8512 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
8513 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
8514 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
8515 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
8516 * doc/tm.texi: Regenerate.
8517 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
8518 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
8519 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8520 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
8521 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
8522 * config/i386/i386.c (inline_secondary_memory_needed): Put the
8523 mode argument first and change the reg_class arguments to reg_class_t.
8524 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
8525 Make static. Update the call to inline_secondary_memory_needed.
8526 (ix86_register_move_cost): Update the call to
8527 inline_secondary_memory_needed.
8528 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8529 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
8531 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
8532 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8534 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
8535 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
8536 * config/mips/mips.c (mips_secondary_memory_needed): Make static
8537 and match hook interface. Add comment from mips.h.
8538 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8539 * config/mmix/mmix.md (truncdfsf2): Refer to
8540 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8542 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
8543 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
8544 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8545 (pa_secondary_memory_needed): New function.
8546 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
8547 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
8548 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8549 (pdp11_secondary_memory_needed): Make static and match hook interface.
8550 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
8551 * config/powerpcspe/powerpcspe-protos.h
8552 (rs6000_secondary_memory_needed_ptr): Delete.
8553 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
8555 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8556 (rs6000_option_override_internal): Assign to
8557 targetm.secondary_memory_needed rather than
8558 rs6000_secondary_memory_needed_ptr.
8559 (rs6000_secondary_memory_needed): Match hook interface.
8560 (rs6000_debug_secondary_memory_needed): Likewise.
8561 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
8562 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
8563 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
8564 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8565 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
8566 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
8568 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
8569 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8570 (rs6000_option_override_internal): Assign to
8571 targetm.secondary_memory_needed rather than
8572 rs6000_secondary_memory_needed_ptr.
8573 (rs6000_secondary_memory_needed): Match hook interface.
8574 (rs6000_debug_secondary_memory_needed): Likewise.
8575 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
8576 * config/s390/s390.c (s390_secondary_memory_needed): New function.
8577 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8578 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
8579 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8580 (sparc_secondary_memory_needed): New function.
8581 * lra-constraints.c (check_and_process_move): Refer to
8582 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8584 (curr_insn_transform): Likewise.
8585 (process_alt_operands): Use targetm.secondary_memory_needed
8586 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8587 (check_secondary_memory_needed_p): Likewise.
8588 (choose_split_class): Likewise.
8589 * reload.c: Unconditionally include code that was previously
8590 conditional on SECONDARY_MEMORY_NEEDED.
8591 (push_secondary_reload): Use targetm.secondary_memory_needed
8592 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8593 (push_reload): Likewise.
8594 * reload1.c: Unconditionally include code that was previously
8595 conditional on SECONDARY_MEMORY_NEEDED.
8596 (choose_reload_regs): Use targetm.secondary_memory_needed
8597 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8598 (gen_reload): Likewise.
8599 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
8601 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8602 Alan Hayward <alan.hayward@arm.com>
8603 David Sherwood <david.sherwood@arm.com>
8605 * target.def (secondary_memory_needed_mode): New hook:
8606 * targhooks.c (default_secondary_memory_needed_mode): Declare.
8607 * targhooks.h (default_secondary_memory_needed_mode): New function.
8608 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
8609 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
8610 * doc/tm.texi: Regenerate.
8611 * lra-constraints.c (check_and_process_move): Use
8612 targetm.secondary_memory_needed_mode instead of
8613 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
8614 (curr_insn_transform): Likewise.
8615 * reload.c (get_secondary_mem): Likewise.
8616 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8617 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
8619 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8620 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8621 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
8622 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8623 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
8625 * config/powerpcspe/powerpcspe-protos.h
8626 (rs6000_secondary_memory_needed_mode): Delete.
8627 * config/powerpcspe/powerpcspe.c
8628 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8629 (rs6000_secondary_memory_needed_mode): Make static.
8630 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8631 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
8633 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8635 (rs6000_secondary_memory_needed_mode): Make static.
8636 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8637 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
8638 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8639 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8640 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8642 (sparc_secondary_memory_needed_mode): New function.
8643 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
8645 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
8647 * config/aarch64/constraints.md (Umq): New constraint.
8648 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
8650 (mov<mode>): Update condition.
8652 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8654 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
8655 when bitposition is the same.
8657 2017-09-13 Richard Biener <rguenther@suse.de>
8659 * dwarf2out.c (output_die_symbol): Remove.
8660 (output_die): Do not output a DIEs symbol.
8662 2017-09-13 Richard Biener <rguenther@suse.de>
8665 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
8666 default-def to avoid breaking iterator update with the weird
8667 interaction with cgraph_update_edges_for_call_stmt_node.
8669 2017-09-13 Richard Biener <rguenther@suse.de>
8671 * tree-cfg.c (verify_gimple_assign_binary): Add verification
8672 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
8673 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
8674 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
8676 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
8678 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
8679 Disable pc relative literal load irrespective of
8680 TARGET_FIX_ERR_A53_84341 for default.
8682 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
8684 * config/sparc/sparc.c (output_return): Output the source location of
8685 the insn in the delay slot, if any.
8686 (output_sibcall): Likewise.
8688 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
8691 * common.opt (-static-pie): New alias.
8692 (shared): Negate static-pie.
8693 (-no-pie): Update help text.
8695 (static-pie): New option.
8696 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
8697 -static-pie support.
8698 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
8699 (LINK_EH_SPEC): Likewise.
8700 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8701 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8702 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
8703 * gcc.c (LINK_COMMAND_SPEC): Likewise.
8704 (init_gcc_specs): Likewise.
8705 (init_spec): Likewise.
8706 (display_help): Update help message for -pie.
8707 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
8710 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
8712 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
8713 (movdi_aarch64): Likewise.
8714 (movti_aarch64): Likewise.
8716 2017-09-12 Simon Wright <simon@pushface.org>
8719 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
8720 calculation of the minor version, always output as 0.
8722 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8725 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
8726 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
8727 on it early, rather than manual conversion late. For
8728 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
8729 instead of performing manual conversion.
8731 2017-09-12 Carl Love <cel@us.ibm.com>
8733 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
8734 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
8736 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8737 VMULOSW): Add definitions.
8738 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8739 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8740 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8741 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8742 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8744 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8746 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
8748 (movti_aarch64): Likewise.
8749 (movdf_aarch64): Likewise.
8750 (movtf_aarch64): Likewise.
8751 (load_pairdi): Likewise.
8752 (store_pairdi): Likewise.
8753 (load_pairdf): Likewise.
8754 (store_pairdf): Likewise.
8755 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
8756 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
8757 (ldr_got_small_<mode>): Likewise.
8758 (ldr_got_small_28k_<mode>): Likewise.
8759 (ldr_got_tiny): Likewise.
8760 * config/aarch64/iterators.md (ldst_sz): New.
8761 (ldpstp_sz): Likewise.
8762 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
8764 (thunderx_load): Split load_8 to load_16.
8765 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
8767 (thunderx2t99_storepair_basic): Split store_8 to store_16.
8768 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
8769 (xgene1_store_pair): Split store_8 to store_16.
8770 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
8771 (falkor_st_0_st_sd): Split store_8 to store_16.
8773 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
8775 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
8776 and store1/2/3/4 to store_4/8/12/16.
8777 * config/aarch64/aarch64.md: Update for rename.
8778 * config/arm/arm.md: Likewise.: Likewise.
8779 * config/arm/arm.c: Likewise.
8780 * config/arm/thumb1.md: Likewise.
8781 * config/arm/thumb2.md: Likewise.
8782 * config/arm/vfp.md: Likewise.
8783 * config/arm/arm-generic.md: Likewise.
8784 * config/arm/arm1020e.md: Likewise.
8785 * config/arm/arm1026ejs.md: Likewise.
8786 * config/arm/arm1136jfs.md: Likewise.
8787 * config/arm/arm926ejs.md: Likewise.
8788 * config/arm/cortex-a15.md: Likewise.
8789 * config/arm/cortex-a17.md: Likewise.
8790 * config/arm/cortex-a5.md: Likewise.
8791 * config/arm/cortex-a53.md: Likewise.
8792 * config/arm/cortex-a57.md: Likewise.
8793 * config/arm/cortex-a7.md: Likewise.
8794 * config/arm/cortex-a8.md: Likewise.
8795 * config/arm/cortex-a9.md: Likewise.
8796 * config/arm/cortex-m4.md: Likewise.
8797 * config/arm/cortex-m7.md: Likewise.
8798 * config/arm/cortex-r4.md: Likewise.
8799 * config/arm/exynos-m1.md: Likewise.
8800 * config/arm/fa526.md: Likewise.
8801 * config/arm/fa606te.md: Likewise.
8802 * config/arm/fa626te.md: Likewise.
8803 * config/arm/fa726te.md: Likewise.
8804 * config/arm/fmp626.md: Likewise.
8805 * config/arm/iwmmxt.md: Likewise.
8806 * config/arm/ldmstm.md: Likewise.
8807 * config/arm/marvell-pj4.md: Likewise.
8808 * config/arm/xgene1.md: Likewise.
8809 * config/aarch64/thunderx.md: Likewise.
8810 * config/aarch64/thunderx2t99.md: Likewise.
8811 * config/aarch64/falkor.md: Likewise.
8813 2017-09-12 Martin Liska <mliska@suse.cz>
8815 * attribs.c (private_lookup_attribute): New function.
8816 * attribs.h (private_lookup_attribute): Declared here.
8817 (lookup_attribute): Called from this place.
8819 2017-09-12 Richard Biener <rguenther@suse.de>
8821 PR tree-optimization/82157
8822 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
8823 stmts with side-effects.
8825 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8826 Alan Hayward <alan.hayward@arm.com>
8827 David Sherwood <david.sherwood@arm.com>
8829 * target.def (hard_regno_nregs): New hook.
8830 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
8831 * targhooks.h (default_hard_regno_nregs): Declare.
8832 * targhooks.c (default_hard_regno_nregs): New function.
8833 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8834 (TARGET_HARD_REGNO_NREGS): ...this hook.
8835 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8836 (CLASS_MAX_NREGS): Likewise.
8837 * doc/tm.texi: Regenerate.
8838 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8839 instead of HARD_REGNO_NREGS.
8840 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8841 HARD_REGNO_NREGS in the comment.
8842 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8843 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8844 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8845 Return an unsigned int.
8846 (TARGET_HARD_REGNO_NREGS): Redefine.
8847 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8848 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8849 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8850 (arc_hard_regno_nregs): New function.
8851 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8852 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8853 (arm_hard_regno_nregs): New function.
8854 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8855 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8856 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8857 (TARGET_HARD_REGNO_NREGS): Redefine.
8858 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8859 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8860 (HARD_REGNO_NREGS): Delete.
8861 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8862 (cr16_hard_regno_nregs): New function.
8863 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8864 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8865 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8866 (cris_hard_regno_nregs): New function.
8867 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8868 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8869 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8870 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8871 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8872 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8873 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8874 (frv_hard_regno_nregs): Make static. Take and return an
8876 (frv_class_max_nregs): Remove outdated copy of documentation.
8877 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8878 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8879 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8880 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8881 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8882 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8883 (TARGET_HARD_REGNO_NREGS): Redefine.
8884 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8885 (CLASS_MAX_NREGS): Update comment.
8886 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8887 (ia64_hard_regno_nregs): New function.
8888 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
8889 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
8890 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
8891 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
8892 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
8894 (m32c_hard_regno_nregs): Likewise. Make static.
8895 (TARGET_HARD_REGNO_NREGS): Redefine.
8896 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
8897 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
8898 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
8899 (m68k_hard_regno_nregs): New function.
8900 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8901 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8902 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8903 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8904 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8905 Take and return an unsigned int.
8906 (TARGET_HARD_REGNO_NREGS): Redefine.
8907 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8908 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8909 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8910 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8911 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8912 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8913 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8914 (msp430_hard_regno_nregs): Make static. Take and return an
8916 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8917 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8918 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8919 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8920 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8921 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8922 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8923 (TARGET_HARD_REGNO_NREGS): Redefine.
8924 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8925 (PA_HARD_REGNO_NREGS): ...this.
8926 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8927 (PA_HARD_REGNO_NREGS): ...this.
8928 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8929 (pa_hard_regno_nregs): New function.
8930 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8931 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8932 (pdp11_hard_regno_nregs): New function.
8933 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8934 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
8935 (rs6000_hard_regno_nregs_hook): New function.
8936 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
8937 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8938 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
8939 Take and return an unsigned int. Move earlier in file.
8940 (TARGET_HARD_REGNO_NREGS): Redefine.
8941 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
8942 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
8943 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
8944 (rl78_hard_regno_nregs): Make static. Take and return an
8946 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
8947 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
8948 (rs6000_hard_regno_nregs_hook): New function.
8949 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
8950 * config/rx/rx.c (rx_hard_regno_nregs): New function.
8951 (TARGET_HARD_REGNO_NREGS): Redefine.
8952 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
8953 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
8954 instead of HARD_REGNO_NREGS.
8955 (s390_hard_regno_nregs): New function.
8956 (s390_hard_regno_mode_ok): Add comment from s390.h.
8957 (TARGET_HARD_REGNO_NREGS): Redefine.
8958 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
8959 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
8960 (sh_hard_regno_nregs): New function.
8961 (sh_pass_in_reg_p): Use it.
8962 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
8963 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8964 (sparc_hard_regno_nregs): New function.
8965 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
8966 * config/spu/spu.c (spu_hard_regno_nregs): New function.
8967 (spu_function_arg_advance): Use it, supplying a valid register number.
8968 (TARGET_HARD_REGNO_NREGS): Redefine.
8969 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
8970 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
8971 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
8972 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
8973 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
8974 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
8975 (CLASS_MAX_NREGS): Remove copy of old documentation.
8976 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
8977 (visium_hard_regno_nregs): New function.
8978 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
8979 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
8980 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8981 xtensa_hard_regno_nregs): New function.
8982 * system.h (HARD_REGNO_NREGS): Poison.
8984 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8986 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
8987 hard_regno_nregs instead of HARD_REGNO_NREGS.
8988 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
8989 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
8990 (c6x_expand_epilogue): Likewise.
8991 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
8992 (frv_read_iacc_argument): Likewise.
8993 * config/sh/sh.c: Include regs.h.
8994 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8995 (regs_used): Likewise.
8996 (output_stack_adjust): Likewise.
8997 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
8998 * expmed.c: Include regs.h.
8999 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9000 * ree.c: Include regs.h.
9001 (combine_reaching_defs): Use hard_regno_nregs instead of
9003 (add_removable_extension): Likewise.
9005 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9007 * regs.h (hard_regno_nregs): Turn into a function.
9008 (end_hard_regno): Update accordingly.
9009 * caller-save.c (setup_save_areas): Likewise.
9010 (save_call_clobbered_regs): Likewise.
9011 (replace_reg_with_saved_mem): Likewise.
9012 (insert_restore): Likewise.
9013 (insert_save): Likewise.
9014 * combine.c (can_change_dest_mode): Likewise.
9015 (move_deaths): Likewise.
9016 (distribute_notes): Likewise.
9017 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9018 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9019 (rs6000_split_multireg_move): Likewise.
9020 (rs6000_register_move_cost): Likewise.
9021 (rs6000_memory_move_cost): Likewise.
9022 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9023 (rs6000_split_multireg_move): Likewise.
9024 (rs6000_register_move_cost): Likewise.
9025 (rs6000_memory_move_cost): Likewise.
9026 * cselib.c (cselib_reset_table): Likewise.
9027 (cselib_lookup_1): Likewise.
9028 * emit-rtl.c (set_mode_and_regno): Likewise.
9029 * function.c (aggregate_value_p): Likewise.
9030 * ira-color.c (setup_profitable_hard_regs): Likewise.
9031 (check_hard_reg_p): Likewise.
9032 (calculate_saved_nregs): Likewise.
9033 (assign_hard_reg): Likewise.
9034 (improve_allocation): Likewise.
9035 (calculate_spill_cost): Likewise.
9036 * ira-emit.c (modify_move_list): Likewise.
9037 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9038 (ira_hard_reg_in_set_p): Likewise.
9039 * ira.c (setup_reg_mode_hard_regset): Likewise.
9040 (clarify_prohibited_class_mode_regs): Likewise.
9041 (check_allocation): Likewise.
9042 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9043 (lra_setup_reg_renumber): Likewise.
9044 (setup_try_hard_regno_pseudos): Likewise.
9045 (spill_for): Likewise.
9046 (assign_hard_regno): Likewise.
9047 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9048 * lra-constraints.c (in_class_p): Likewise.
9049 (lra_constraint_offset): Likewise.
9050 (simplify_operand_subreg): Likewise.
9051 (lra_constraints): Likewise.
9052 (split_reg): Likewise.
9053 (split_if_necessary): Likewise.
9054 (invariant_p): Likewise.
9055 (inherit_in_ebb): Likewise.
9056 * lra-lives.c (process_bb_lives): Likewise.
9057 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9058 (get_hard_regs): Likewise.
9059 (do_remat): Likewise.
9060 * lra-spills.c (assign_spill_hard_regs): Likewise.
9061 * mode-switching.c (create_pre_exit): Likewise.
9062 * postreload.c (reload_combine_recognize_pattern): Likewise.
9063 * recog.c (peep2_find_free_register): Likewise.
9064 * regcprop.c (kill_value_regno): Likewise.
9065 (set_value_regno): Likewise.
9066 (copy_value): Likewise.
9067 (maybe_mode_change): Likewise.
9068 (find_oldest_value_reg): Likewise.
9069 (copyprop_hardreg_forward_1): Likewise.
9070 * regrename.c (check_new_reg_p): Likewise.
9071 (regrename_do_replace): Likewise.
9072 * reload.c (push_reload): Likewise.
9073 (combine_reloads): Likewise.
9074 (find_dummy_reload): Likewise.
9075 (operands_match_p): Likewise.
9076 (find_reloads): Likewise.
9077 (find_equiv_reg): Likewise.
9078 (reload_adjust_reg_for_mode): Likewise.
9079 * reload1.c (count_pseudo): Likewise.
9080 (count_spilled_pseudo): Likewise.
9081 (find_reg): Likewise.
9082 (clear_reload_reg_in_use): Likewise.
9083 (free_for_value_p): Likewise.
9084 (allocate_reload_reg): Likewise.
9085 (choose_reload_regs): Likewise.
9086 (reload_adjust_reg_for_temp): Likewise.
9087 (emit_reload_insns): Likewise.
9088 (delete_output_reload): Likewise.
9089 * rtlanal.c (subreg_get_info): Likewise.
9090 * sched-deps.c (sched_analyze_reg): Likewise.
9091 * sel-sched.c (init_regs_for_mode): Likewise.
9092 (mark_unavailable_hard_regs): Likewise.
9093 (choose_best_reg_1): Likewise.
9094 (verify_target_availability): Likewise.
9095 * valtrack.c (dead_debug_insert_temp): Likewise.
9096 * var-tracking.c (track_loc_p): Likewise.
9097 (emit_note_insn_var_location): Likewise.
9098 * varasm.c (make_decl_rtl): Likewise.
9099 * reginfo.c (choose_hard_reg_mode): Likewise.
9100 (init_reg_modes_target): Refer directly to
9101 this_target_regs->x_hard_regno_nregs.
9103 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9105 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9106 instead of hard_regno_nregs.
9108 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9110 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9111 end_hard_regno instead of hard_regno_nregs.
9112 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9113 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9114 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9115 * ira-color.c (improve_allocation): Likewise.
9116 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9117 * lra-lives.c (mark_regno_live): Likewise.
9118 (mark_regno_dead): Likewise.
9119 * lra-remat.c (operand_to_remat): Likewise.
9120 * lra.c (collect_non_operand_hard_regs): Likewise.
9121 * postreload.c (reload_combine_note_store): Likewise.
9122 (move2add_valid_value_p): Likewise.
9123 * reload.c (regno_clobbered_p): Likewise.
9125 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9127 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9129 * config/v850/v850.c (v850_reorg): Likewise.
9130 * reload.c (refers_to_regno_for_reload_p): Likewise.
9131 (find_equiv_reg): Likewise.
9132 * reload1.c (reload_reg_reaches_end_p): Likewise.
9134 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9136 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9138 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9139 * config/arm/arm.c (output_move_neon): Likewise.
9140 (arm_attr_length_move_neon): Likewise.
9141 (neon_split_vcombine): Likewise.
9142 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9143 (c6x_mark_reg_written): Likewise.
9144 (c6x_dwarf_register_span): Likewise.
9145 * config/i386/i386.c (ix86_save_reg): Likewise.
9146 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9147 (rws_access_reg): Likewise.
9148 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9149 * mode-switching.c (create_pre_exit): Likewise.
9150 * ree.c (combine_reaching_defs): Likewise.
9151 (add_removable_extension): Likewise.
9152 * regcprop.c (find_oldest_value_reg): Likewise.
9153 (copyprop_hardreg_forward_1): Likewise.
9154 * reload.c (reload_inner_reg_of_subreg): Likewise.
9155 (push_reload): Likewise.
9156 (combine_reloads): Likewise.
9157 (find_dummy_reload): Likewise.
9158 (reload_adjust_reg_for_mode): Likewise.
9159 * reload1.c (find_reload_regs): Likewise.
9160 (forget_old_reloads_1): Likewise.
9161 (reload_reg_free_for_value_p): Likewise.
9162 (reload_adjust_reg_for_temp): Likewise.
9163 (emit_reload_insns): Likewise.
9164 (delete_output_reload): Likewise.
9165 * sel-sched.c (choose_best_reg_1): Likewise.
9166 (choose_best_pseudo_reg): Likewise.
9168 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9169 Alan Hayward <alan.hayward@arm.com>
9170 David Sherwood <david.sherwood@arm.com>
9172 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9173 * target.def (slow_unaligned_access): New hook.
9174 * targhooks.h (default_slow_unaligned_access): Declare.
9175 * targhooks.c (default_slow_unaligned_access): New function.
9176 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9177 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9178 * doc/tm.texi: Regenerate.
9179 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9180 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9181 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9183 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9184 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9186 (rs6000_slow_unaligned_access): New function.
9187 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9188 (expand_block_compare): Likewise.
9189 (expand_strn_compare): Likewise.
9190 (rs6000_rtx_costs): Likewise.
9191 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9192 (riscv_slow_unaligned_access): Likewise.
9193 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9194 (riscv_slow_unaligned_access_p): ...this and make static.
9195 (riscv_option_override): Update accordingly.
9196 (riscv_slow_unaligned_access): New function.
9197 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9198 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9199 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9200 (rs6000_slow_unaligned_access): New function.
9201 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9202 (rs6000_rtx_costs): Likewise.
9203 * config/rs6000/rs6000-string.c (expand_block_compare)
9204 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9205 of SLOW_UNALIGNED_ACCESS.
9206 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9207 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9208 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9209 of SLOW_UNALIGNED_ACCESS.
9210 * expmed.c (simple_mem_bitfield_p): Likewise.
9211 * expr.c (alignment_for_piecewise_move): Likewise.
9212 (emit_group_load_1): Likewise.
9213 (emit_group_store): Likewise.
9214 (copy_blkmode_from_reg): Likewise.
9215 (emit_push_insn): Likewise.
9216 (expand_assignment): Likewise.
9217 (store_field): Likewise.
9218 (expand_expr_real_1): Likewise.
9219 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9220 * lra-constraints.c (simplify_operand_subreg): Likewise.
9221 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9222 * gimple-ssa-store-merging.c: Likewise in block comment at start
9224 * tree-ssa-strlen.c: Include target.h.
9225 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9226 of SLOW_UNALIGNED_ACCESS.
9227 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9229 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9231 PR rtl-optimization/82185
9232 * expmed.c (emit_store_flag_int): Only test tem if it has been
9235 2017-09-12 Richard Biener <rguenther@suse.de>
9238 * match.pd ((FTYPE) N CMP CST): Fix typo.
9240 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9242 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9244 (mips_near_type_p): Add 'short_call' attribute as a synonym
9246 * doc/extend.texi (short_call): Document new function attribute.
9248 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9251 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9252 assertion check that in the condition.
9253 (get_atomic_generic_size): Likewise. Before testing if parameter
9254 has pointer type, if it has array type, call for C++
9255 default_conversion to perform array-to-pointer conversion.
9257 2017-09-12 Richard Biener <rguenther@suse.de>
9259 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9260 for operations we cannot scalarize.
9262 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9264 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9265 vectors heap vectors. Clean up comments.
9266 Make visited_bbs a reference.
9267 (profitable_jump_thread_path): Make GC
9268 vectors heap vectors. Clean up comments.
9270 (convert_and_register_jump_thread_path): Make GC vectors heap
9272 (check_subpath_and_update_thread_path): Same. Clean up comments.
9273 Make visited_bbs a reference.
9274 (handle_phi): Abstract common code to to
9275 register_jump_thread_path_if_profitable.
9276 Rename VAR_BB to DEF_BB.
9278 Make GC vectors heap vectors.
9279 Make visited_bbs a reference.
9280 (handle_assignment): Same.
9281 (register_jump_thread_path_if_profitable): New.
9282 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9284 Make GC vectors heap vectors. Clean up comments.
9285 Make visited_bbs a reference.
9286 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9287 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9290 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9293 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9294 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9296 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9298 Revert r251800 and r251799.
9300 2017-09-11 Martin Jambor <mjambor@suse.cz>
9303 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9304 arguments in advance.
9305 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9306 use it to find predecessor edges.
9307 (naive_outof_ssa): Collect vector of predecessors.
9309 2017-09-08 Jason Merrill <jason@redhat.com>
9311 PR c++/70029 - ICE with ref-qualifier and -flto
9312 * langhooks.h (struct lang_hooks_for_types): Add
9313 copy_lang_qualifiers.
9314 * attribs.c (build_type_attribute_qual_variant): Use it.
9315 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9317 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9318 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9320 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9323 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9324 (*mulsi3_sp64): New instruction.
9325 (mulsi3): New expander.
9327 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9329 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9331 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9333 * sancov.c: Include memmodel.h.
9335 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9338 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9341 2017-09-07 Carl Love <cel@us.ibm.com>
9343 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9344 the sldi instruction.
9346 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9348 * sancov.c: Include tm_p.h.
9350 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9353 * output.h (switch_to_other_text_partition): New declaration.
9354 * varasm.c (switch_to_other_text_partition): New function.
9355 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9356 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9357 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9358 to the other text partition before emitting LCL label and switch back
9359 after emitting the word after it.
9361 2017-09-07 Richard Biener <rguenther@suse.de>
9363 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9364 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9365 critical edges here, after loop init.
9366 (pass_data_pre): Remove PROP_no_crit_edges flags.
9367 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9368 for valueization of call args to avoid leaking VN_TOP.
9369 (visit_use): Assert we do not visit default defs.
9370 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9371 Use error_mark_node to more easily detect leaking VN_TOP.
9372 All default-defs are varying, not VN_TOP. Mark them visited.
9373 (run_scc_vn): Make code match comment.
9375 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9377 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9378 OPTION_MASK_FLOAT128_KEYWORD.
9379 (POWERPC_MASKS): Likewise.
9380 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9381 support for the -mfloat128-type option, and make -mfloat128
9382 default on PowerPC Linux systems. Define or undefine
9383 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
9384 Define __float128 to be __ieee128 if IEEE 128-bit support is
9385 enabled, or undefine it.
9386 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
9387 Delete defining __FLOAT128_TYPE__.
9388 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
9389 -mfloat128-type option and make -mfloat128 default on PowerPC
9391 (TARGET_FLOAT128_TYPE): Likewise.
9392 (-mfloat128-type): Likewise.
9393 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9394 Delete the -mfloat128-type option and make -mfloat128 default on
9395 PowerPC Linux systems. Always use __ieee128 to be the keyword for
9396 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
9397 128-bit floating point is enabled. Change tests from using
9398 -mfloat128-type to -mfloat128.
9399 (rs6000_mangle_type): Use the correct mangling for the __float128
9400 type even if normal long double is restricted to 64-bits.
9401 (floatn_mode): Enable the _Float128 type by default on VSX Linux
9403 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
9404 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
9405 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
9407 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
9408 documentation for -mfloat128.
9410 2017-09-06 Olivier Hainque <hainque@adacore.com>
9412 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
9414 2017-09-06 Wish Wu <wishwu007@gmail.com>
9415 Jakub Jelinek <jakub@redhat.com>
9417 * asan.c (initialize_sanitizer_builtins): Add
9418 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
9419 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
9420 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
9421 BT_FN_VOID_UINT64_PTR variables.
9422 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
9423 (BT_FN_VOID_UINT16_UINT16): Likewise.
9424 (BT_FN_VOID_UINT32_UINT32): Likewise.
9425 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
9426 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
9427 (BT_FN_VOID_UINT64_PTR): Likewise.
9428 * common.opt (flag_sanitize_coverage): New variable.
9429 (fsanitize-coverage=trace-pc): Remove.
9430 (fsanitize-coverage=): Add.
9431 * flag-types.h (enum sanitize_coverage_code): New enum.
9432 * fold-const.c (fold_range_test): Disable non-short-circuit
9433 optimization if flag_sanitize_coverage.
9434 (fold_truth_andor): Likewise.
9435 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
9436 * opts.c (COVERAGE_SANITIZER_OPT): Define.
9437 (coverage_sanitizer_opts): New array.
9438 (get_closest_sanitizer_option): Add OPTS argument, handle also
9439 OPT_fsanitize_coverage_.
9440 (parse_sanitizer_options): Adjusted to also handle
9441 OPT_fsanitize_coverage_.
9442 (common_handle_option): Add OPT_fsanitize_coverage_.
9443 * sancov.c (instrument_comparison, instrument_switch): New function.
9444 (sancov_pass): Add trace-cmp support.
9445 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
9446 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
9447 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
9448 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
9449 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
9450 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
9451 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
9452 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
9453 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
9454 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
9456 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9458 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9459 error. Only quit immediately if parsing is complete.
9460 (BEGIN): Initialize fatal_err and parse_done.
9461 (begin fpu, end fpu): Check number of arguments.
9462 (begin arch, end arch): Likewise.
9463 (begin cpu, end cpu): Likewise.
9464 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9465 (optalias): Likewise.
9467 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9469 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9470 * config/arm/arm-isa.h: Delete. Move definitions to ...
9471 * arm-cpus.in: ... here. Use new feature and fgroup values.
9472 * config/arm/arm.c (arm_option_override): Use lower case for feature
9474 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9475 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9476 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9478 (print_isa_bits_for): New function.
9479 (gen_isa): New function.
9480 (gen_comm_data): Use print_isa_bits_for.
9481 (define feature): New keyword.
9482 (define fgroup): New keyword.
9483 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
9484 (arm-isa.h): Add rule to generate file.
9485 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9486 case for feature bit names.
9488 2017-09-06 Richard Biener <rguenther@suse.de>
9490 * tree-ssa-pre.c (NECESSARY): Remove.
9491 (create_expression_by_pieces): Do not touch pass-local flags.
9492 (insert_into_preds_of_block): Likewise.
9493 (do_pre_regular_insertion): Likewise.
9494 (eliminate_insert): Likewise.
9495 (eliminate_dom_walker::before_dom_children): Likewise.
9496 (fini_eliminate): Do not look at inserted_exprs.
9497 (mark_operand_necessary): Remove.
9498 (remove_dead_inserted_code): Replace with simple work-list
9499 algorithm based on inserted_exprs and SSA uses.
9500 (pass_pre::execute): Re-order fini_eliminate and
9501 remove_dead_inserted_code.
9503 2017-09-06 Olivier Hainque <hainque@adacore.com>
9505 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
9506 for VxWorks 7. Adjust surrounding comments.
9508 2017-09-06 Richard Biener <rguenther@suse.de>
9510 * gimple-ssa-strength-reduction.c
9511 (find_candidates_dom_walker::before_dom_children): Also allow
9514 2017-09-06 Richard Biener <rguenther@suse.de>
9516 PR tree-optimization/82108
9517 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
9518 for gap in the non-permutation SLP case.
9520 2017-09-06 Martin Jambor <mjambor@suse.cz>
9522 PR tree-optimization/82078
9523 * tree-sra.c (sort_and_splice_var_accesses): Move call to
9524 add_access_to_work_queue...
9525 (build_accesses_from_assign): ...here.
9526 (propagate_all_subaccesses): Make sure racc is the group
9527 representative, if there is one.
9529 2017-09-06 Jakub Jelinek <jakub@redhat.com>
9532 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
9535 2017-09-06 Richard Biener <rguenther@suse.de>
9537 * gimple-ssa-strength-reduction.c
9538 (find_candidates_dom_walker::before_doom_children): Use a
9539 type and not a mode check.
9541 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9544 * config/arm/predicates.md (arm_general_adddi_operand): Create new
9546 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
9548 2017-09-05 Jeff Law <law@redhat.com>
9550 PR tree-optimization/64910
9551 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
9552 cases where we have 3 or more operands.
9554 2017-09-05 Jakub Jelinek <jakub@redhat.com>
9557 * omp-low.c (lower_omp_for): Recompute tree invariant if
9558 gimple_omp_for_initial/final is ADDR_EXPR.
9561 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
9562 into gimple val before gimplification fo the COND_EXPR.
9564 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
9566 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
9567 REGION_COPY argument.
9568 (thread_through_all_blocks): Remove unused argument to
9569 duplicate_thread_path.
9571 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9572 Alan Hayward <alan.hayward@arm.com>
9573 David Sherwood <david.sherwood@arm.com>
9575 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
9576 Take a scalar_mode rather than a machine_mode.
9577 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9578 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
9579 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9580 (aarch64_gen_adjusted_ldpstp): Likewise.
9581 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
9583 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9584 Alan Hayward <alan.hayward@arm.com>
9585 David Sherwood <david.sherwood@arm.com>
9587 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
9588 Take a scalar_int_mode instead of a machine_mode.
9589 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9590 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9591 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9592 (aarch64_simd_attr_length_rglist): Delete.
9593 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
9594 a scalar_int_mode instead of a machine_mode.
9595 (aarch64_add_offset): Likewise.
9596 (aarch64_internal_mov_immediate): Likewise
9597 (aarch64_add_constant_internal): Likewise.
9598 (aarch64_add_constant): Likewise.
9599 (aarch64_movw_imm): Likewise.
9600 (aarch64_rtx_arith_op_extract_p): Likewise.
9601 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9602 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9603 Remove assert that the mode isn't a vector.
9604 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9605 (aarch64_expand_mov_immediate): Update calls after above changes.
9606 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
9607 (aarch64_and_bitmask_imm): Check for scalar integer modes.
9608 (aarch64_move_imm): Likewise.
9609 (aarch64_can_const_movi_rtx_p): Likewise.
9610 (aarch64_strip_extend): Likewise.
9611 (aarch64_extr_rtx_p): Likewise.
9612 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
9613 a CONST_INT when the mode parameter is VOIDmode.
9614 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
9616 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9618 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
9619 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9620 (bitwise_type_for_mode): Update accordingly.
9622 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9624 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
9625 * stor-layout.c (mode_for_size_tree): Likewise.
9626 (mode_for_array): Update accordingly.
9627 (layout_decl): Likewise.
9628 (compute_record_mode): Likewise. Only set the mode once.
9630 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9632 * target.def (get_mask_mode): Change return type to opt_mode.
9634 * doc/tm.texi: Regenerate.
9635 * targhooks.h (default_get_mask_mode): Return an opt_mode.
9636 * targhooks.c (default_get_mask_mode): Likewise.
9637 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9638 * optabs-query.c (can_vec_mask_load_store_p): Update use of
9639 targetm.get_mask_mode.
9640 * tree.c (build_truth_vector_type): Likewise.
9642 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9644 * machmode.h (mode_for_vector): Return an opt_mode.
9645 * stor-layout.c (mode_for_vector): Likewise.
9646 (mode_for_int_vector): Update accordingly.
9647 (layout_type): Likewise.
9648 * config/i386/i386.c (emit_memmov): Likewise.
9649 (ix86_expand_set_or_movmem): Likewise.
9650 (ix86_expand_vector_init): Likewise.
9651 (ix86_get_mask_mode): Likewise.
9652 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
9654 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
9655 * expmed.c (extract_bit_field_1): Likewise.
9656 * expr.c (expand_expr_real_2): Likewise.
9657 * optabs-query.c (can_vec_perm_p): Likewise.
9658 (can_vec_mask_load_store_p): Likewise.
9659 * optabs.c (expand_vec_perm): Likewise.
9660 * targhooks.c (default_get_mask_mode): Likewise.
9661 * tree-vect-stmts.c (vectorizable_store): Likewise.
9662 (vectorizable_load): Likewise.
9663 (get_vectype_for_scalar_type_and_size): Likewise.
9665 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9667 * machmode.h (mode_for_int_vector): New function.
9668 * stor-layout.c (mode_for_int_vector): Likewise.
9669 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
9670 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
9671 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
9672 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9673 (s390_expand_vcond): Likewise.
9675 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9677 * machmode.h (opt_machine_mode): New type.
9678 (opt_mode<T>): Allow construction from anything that can be
9680 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
9681 (mode_for_size): Return an opt_machine_mode.
9682 * stor-layout.c (mode_for_size): Likewise.
9683 (mode_for_size_tree): Update call accordingly.
9684 (bitwise_mode_for_mode): Likewise.
9685 (make_fract_type): Likewise.
9686 (make_accum_type): Likewise.
9687 * caller-save.c (replace_reg_with_saved_mem): Update call
9689 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9690 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9691 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9692 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
9693 * expmed.c (extract_bit_field_1): Likewise.
9694 * reload.c (get_secondary_mem): Likewise.
9695 * varasm.c (assemble_integer): Likewise.
9696 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
9699 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9701 * machmode.h (decimal_float_mode_for_size): New function.
9702 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
9703 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
9704 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
9705 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
9706 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
9707 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
9709 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9711 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
9712 (get_builtin_sync_mode): Likewise.
9713 (expand_ifn_atomic_compare_exchange): Likewise.
9714 (expand_builtin_atomic_clear): Likewise.
9715 (expand_builtin_atomic_test_and_set): Likewise.
9716 (fold_builtin_atomic_always_lock_free): Likewise.
9717 * calls.c (compute_argument_addresses): Likewise.
9718 (emit_library_call_value_1): Likewise.
9719 (store_one_arg): Likewise.
9720 * combine.c (combine_instructions): Likewise.
9721 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
9722 * config/arm/arm.c (arm_function_value): Likewise.
9723 (aapcs_allocate_return_reg): Likewise.
9724 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
9725 * config/i386/i386.c (construct_container): Likewise.
9726 (ix86_gimplify_va_arg): Likewise.
9727 (ix86_expand_sse_cmp): Likewise.
9728 (emit_memmov): Likewise.
9729 (emit_memset): Likewise.
9730 (expand_small_movmem_or_setmem): Likewise.
9731 (ix86_expand_pextr): Likewise.
9732 (ix86_expand_pinsr): Likewise.
9733 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
9734 * config/microblaze/microblaze.c (microblaze_block_move_straight):
9736 * config/mips/mips.c (mips_function_value_1) Likewise.
9737 (mips_block_move_straight): Likewise.
9738 (mips_expand_ins_as_unaligned_store): Likewise.
9739 * config/powerpcspe/powerpcspe.c
9740 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9741 (rs6000_darwin64_record_arg_flush): Likewise.
9742 * config/rs6000/rs6000.c
9743 (rs6000_darwin64_record_arg_advance_flush): Likewise.
9744 (rs6000_darwin64_record_arg_flush): Likewise.
9745 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
9746 (sparc_function_value_1): Likewise.
9747 * config/spu/spu.c (adjust_operand): Likewise.
9748 (spu_emit_branch_or_set): Likewise.
9749 (arith_immediate_p): Likewise.
9750 * emit-rtl.c (gen_lowpart_common): Likewise.
9751 * expr.c (expand_expr_real_1): Likewise.
9752 * function.c (assign_parm_setup_block): Likewise.
9753 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9754 * reload1.c (alter_reg): Likewise.
9755 * stor-layout.c (mode_for_vector): Likewise.
9756 (layout_type): Likewise.
9758 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9760 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
9761 (spu_convert_move): Likewise.
9762 * lower-subreg.c (resolve_simple_move): Likewise.
9764 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9767 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
9768 define_insn to a define_expand.
9769 (altivec_vsum2sws_direct): New define_insn.
9770 (altivec_vsumsws): Convert from a define_insn to a define_expand.
9772 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9774 * config/arm/arm.c (arm_option_params_internal): Improve setting of
9777 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
9782 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
9783 realign stack if stack alignment needed is less than incoming
9786 2017-09-05 Marek Polacek <polacek@redhat.com>
9789 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
9792 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
9794 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
9796 2017-09-05 Richard Biener <rguenther@suse.de>
9798 PR tree-optimization/82084
9799 * fold-const.c (can_native_encode_string_p): Handle wide characters.
9801 2017-09-05 Richard Biener <rguenther@suse.de>
9803 PR tree-optimization/82102
9804 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
9806 2017-09-05 Martin Liska <mliska@suse.cz>
9808 PR tree-optimization/82032
9809 * tree-cfg.c (generate_range_test): New function.
9810 * tree-cfg.h (generate_range_test): Declared here.
9811 * tree-cfgcleanup.c (convert_single_case_switch): New function.
9812 (cleanup_control_expr_graph): Use it.
9813 * tree-switch-conversion.c (try_switch_expansion): Remove
9815 (emit_case_nodes): Use generate_range_test.
9817 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9820 * config/i386/i386.md (*<btsc><mode>_mask): Add
9821 TARGET_USE_BT to insn constraint.
9822 (*btr<mode>_mask): Ditto.
9824 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
9826 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
9827 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
9829 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9832 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9833 TARGET_NEON and TARGET_IWMMXT.
9834 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9835 TARGET_NEON and TARGET_IWMMXT.
9836 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9838 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9840 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9841 (ix86_rewrite_tls_address): Ditto.
9842 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9843 (ix86_rewrite_tls_address_1): Ditto.
9844 (ix86_rewrite_tls_address): Ditto.
9845 * config/i386/predicates.md (tls_address_pattern): New predicate.
9846 * config/i386/i386.md (TLS address splitter): New splitter.
9848 2017-09-04 Richard Biener <rguenther@suse.de>
9850 PR tree-optimization/82084
9851 * fold-const.h (can_native_encode_string_p): Declare.
9852 * fold-const.c (can_native_encode_string_p): Factor out from ...
9853 (native_encode_string): ... here.
9854 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9855 vectorizing stores from constants we later cannot handle.
9857 2017-09-04 Marek Polacek <polacek@redhat.com>
9860 * doc/invoke.texi: Update -Wtautological-compare documentation.
9862 2017-09-04 Jeff Law <law@redhat.com>
9864 PR tree-optimization/64910
9865 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9866 swap the first and last operand if the last is a constant.
9868 2017-09-04 Marek Polacek <polacek@redhat.com>
9871 * convert.c (do_narrow): When sanitizing signed integer overflows,
9872 bail out for signed types.
9873 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9875 2017-09-04 Richard Biener <rguenther@suse.de>
9877 PR tree-optimization/82060
9878 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9879 Move devirtualization after stmt folding and before EH/AB/noreturn
9880 cleanup to get the stmt refs canonicalized. Use a bool instead
9881 of gimple_modified_p since that doesn't work for NOPs. Schedule
9882 NOPs generated by folding for removal.
9884 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9885 Alan Hayward <alan.hayward@arm.com>
9886 David Sherwood <david.sherwood@arm.com>
9888 * coretypes.h (pad_direction): New enum.
9889 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
9890 (FUNCTION_ARG_PADDING): Likewise.
9891 * target.def (function_arg_padding): New hook.
9892 * targhooks.h (default_function_arg_padding): Declare.
9893 * targhooks.c (default_function_arg_padding): New function.
9894 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
9895 (TARGET_FUNCTION_ARG_PADDING): ...this.
9896 * doc/tm.texi: Regenerate.
9897 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
9898 instead of direction.
9899 (compute_argument_addresses): Likewise.
9900 (load_register_parameters): Likewise.
9901 (emit_library_call_value_1): Likewise.
9902 (store_one_arg): Use targetm.calls.function_arg_padding instead
9903 of FUNCTION_ARG_PADDING.
9904 (must_pass_in_stack_var_size_or_pad): Likewise.
9905 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9906 (emit_group_store): Likewise.
9907 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9908 instead of FUNCTION_ARG_PADDING.
9909 (emit_push_insn): Likewise, and propagate enum change throughout
9911 * function.h (direction): Delete.
9912 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9914 * function.c (assign_parm_find_stack_rtl): Likewise.
9915 (assign_parm_setup_block_p): Likewise.
9916 (assign_parm_setup_block): Likewise.
9917 (gimplify_parameters): Likewise.
9918 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9919 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9921 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9922 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9923 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9924 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9925 (aarch64_function_arg_padding): ...this new function.
9926 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9927 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9928 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9929 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9930 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9931 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9932 (arm_pad_arg_upward): Replace with...
9933 (arm_function_arg_padding): ...this new function.
9934 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
9936 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
9937 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
9938 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9939 (ia64_hpux_function_arg_padding): Replace with...
9940 (ia64_function_arg_padding): ...this new function. Use pad_direction
9941 instead of direction. Check for TARGET_HPUX.
9942 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
9943 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9944 (iq2000_function_arg_padding): New function.
9945 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
9946 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
9947 (mips_function_arg_padding): ...this new function.
9948 (mips_pad_reg_upward): Update accordingly.
9949 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9950 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
9951 targetm.calls.function_arg_padding.
9952 (FUNCTION_ARG_PADDING): Delete.
9953 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9954 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
9955 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9956 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
9957 (nios2_block_reg_padding): Return pad_direction instead of direction.
9958 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
9959 instead of direction.
9960 (nios2_function_arg_padding): Likewise. Make static.
9961 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9962 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
9963 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
9964 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
9965 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9966 (pa_function_arg_padding): Make static. Return pad_direction instead
9968 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
9969 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9970 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
9971 instead of direction. Use targetm.calls.function_arg_padding.
9972 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
9973 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9974 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
9975 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
9976 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
9978 (function_arg_padding): Rename to...
9979 (rs6000_function_arg_padding): ...this. Make static. Return
9980 pad_direction instead of direction.
9981 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9982 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9983 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9984 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9985 instead of direction. Use targetm.calls.function_arg_padding.
9986 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
9987 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9988 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
9989 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
9990 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9991 (function_arg_padding): Rename to...
9992 (rs6000_function_arg_padding): ...this. Make static. Return
9993 pad_direction instead of direction.
9994 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9995 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
9996 * config/s390/s390.c (s390_function_arg_padding): New function.
9997 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9998 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
9999 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10000 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10001 (function_arg_padding): Rename to...
10002 (sparc_function_arg_padding): ...this. Make static. Return
10003 pad_direction instead of direction.
10004 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10005 * config/spu/spu.c (spu_function_arg_padding): New function.
10006 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10007 * system.h (FUNCTION_ARG_PADDING): Poison.
10009 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10010 Alan Hayward <alan.hayward@arm.com>
10011 David Sherwood <david.sherwood@arm.com>
10013 * target.def (modes_tieable_p): New hook.
10014 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10015 (TARGET_MODES_TIEABLE_P): ...this.
10016 * doc/tm.texi.in: Regenerate.
10017 * hooks.h (hook_bool_mode_mode_true): Declare.
10018 * hooks.c (hook_bool_mode_mode_true): New function.
10019 * combine.c (subst): Use targetm.modes_tieable_p instead of
10021 * dse.c (find_shift_sequence): Likewise.
10022 * expmed.c (extract_low_bits): Likewise.
10023 * lower-subreg.c: Include target.h.
10024 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10026 * rtlanal.c (rtx_cost): Likewise.
10027 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10028 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10029 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10030 (TARGET_MODES_TIEABLE_P): Redefine.
10031 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10032 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10033 (TARGET_MODES_TIEABLE_P): Redefine.
10034 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10035 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10036 (arc_modes_tieable_p): New function.
10037 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10038 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10039 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10040 (arm_modes_tieable_p): Make static.
10041 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10042 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10043 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10044 (TARGET_MODES_TIEABLE_P): Redefine.
10045 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10046 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10047 (TARGET_MODES_TIEABLE_P): Redefine.
10048 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10049 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10050 (cr16_modes_tieable_p): New function.
10051 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10052 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10053 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10054 (TRULY_NOOP_TRUNCATION): Update comment.
10055 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10056 (TRULY_NOOP_TRUNCATION): Update comment.
10057 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10058 (frv_modes_tieable_p): New function.
10059 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10060 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10061 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10062 (TARGET_MODES_TIEABLE_P): Redefine.
10063 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10064 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10065 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10066 (TARGET_MODES_TIEABLE_P): Redefine.
10067 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10068 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10069 (ia64_modes_tieable_p): New function.
10070 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10071 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10072 (iq2000_modes_tieable_p): New function.
10073 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10074 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10075 (lm32_modes_tieable_p): New function.
10076 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10077 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10078 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10079 (TARGET_MODES_TIEABLE_P): Redefine.
10080 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10081 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10082 (m32r_modes_tieable_p): New function.
10083 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10084 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10085 (m68k_modes_tieable_p): New function.
10086 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10087 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10088 (mcore_modes_tieable_p): New function.
10089 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10090 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10092 (TARGET_MODES_TIEABLE_P): Redefine.
10093 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10094 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10095 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10096 (TARGET_MODES_TIEABLE_P): Redefine.
10097 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10098 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10099 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10100 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10101 (mn10300_modes_tieable_p): ...this and make static.
10102 (TARGET_MODES_TIEABLE_P): Redefine.
10103 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10104 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10105 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10106 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10107 (msp430_modes_tieable_p): Make static.
10108 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10109 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10110 (TARGET_MODES_TIEABLE_P): Redefine.
10111 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10112 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10113 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10114 (TARGET_MODES_TIEABLE_P): Redefine.
10115 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10116 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10117 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10118 (TARGET_MODES_TIEABLE_P): Redefine.
10119 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10120 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10121 (pdp11_modes_tieable_p): New function.
10122 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10123 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10124 (rs6000_modes_tieable_p): New function.
10125 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10126 * config/powerpcspe/powerpcspe.md: Update comment.
10127 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10128 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10129 (TARGET_MODES_TIEABLE_P): Redefine.
10130 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10131 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10132 (rl78_modes_tieable_p): New function.
10133 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10134 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10135 (rs6000_modes_tieable_p): New function.
10136 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10137 * config/rs6000/rs6000.md: Update comment.
10138 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10139 * config/rx/rx.c (rx_modes_tieable_p): New function.
10140 (TARGET_MODES_TIEABLE_P): Redefine.
10141 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10142 * config/s390/s390.c (s390_modes_tieable_p): New function.
10143 (TARGET_MODES_TIEABLE_P): Redefine.
10144 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10145 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10146 (sh_modes_tieable_p): New function.
10147 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10148 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10149 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10150 (sparc_modes_tieable_p): Make static.
10151 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10152 * config/spu/spu.c (spu_modes_tieable_p): New function.
10153 (TARGET_MODES_TIEABLE_P): Redefine.
10154 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10155 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10156 (TARGET_MODES_TIEABLE_P): Redefine.
10157 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10158 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10159 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10160 * config/v850/v850.c (v850_modes_tieable_p): New function.
10161 (TARGET_MODES_TIEABLE_P): Redefine.
10162 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10163 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10164 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10165 (visium_modes_tieable_p): New function.
10166 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10167 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10168 (xtensa_modes_tieable_p): New function.
10169 * system.h (MODES_TIEABLE_P): Poison.
10171 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10172 Alan Hayward <alan.hayward@arm.com>
10173 David Sherwood <david.sherwood@arm.com>
10175 * target.def (hard_regno_mode_ok): New hook.
10176 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10177 (TARGET_HARD_REGNO_MODE_OK): ...this.
10178 * doc/tm.texi.in: Regenerate.
10179 * hooks.h (hook_bool_uint_mode_true): Declare.
10180 * hooks.c (hook_bool_uint_mode_true): New function.
10181 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10182 HARD_REGNO_MODE_OK.
10183 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10184 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10185 instead of HARD_REGNO_MODE_OK.
10186 * caller-save.c: Include target.h.
10187 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10188 HARD_REGNO_MODE_OK.
10189 * combine.c (can_combine_p): Likewise.
10190 (combinable_i3pat): Likewise.
10191 (can_change_dest_mode): Likewise.
10192 * expr.c (init_expr_target): Likewise.
10193 (convert_move): Likewise.
10194 (convert_modes): Likewise.
10195 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10196 (setup_prohibited_mode_move_regs): Likewise.
10197 * ira.h (target_ira): Likewise.
10198 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10199 * lra-constraints.c (process_alt_operands): Likewise.
10200 (split_reg): Likewise.
10201 * recog.c (peep2_find_free_register): Likewise.
10202 * ree.c (combine_reaching_defs): Likewise.
10203 * regcprop.c (maybe_mode_change): Likewise.
10204 * reginfo.c (init_reg_sets_1): Likewise.
10205 (choose_hard_reg_mode): Likewise.
10206 (simplifiable_subregs): Likewise.
10207 * regrename.c (check_new_reg_p): Likewise.
10208 * reload.c (find_valid_class): Likewise.
10209 (find_valid_class_1): Likewise.
10210 (reload_inner_reg_of_subreg): Likewise.
10211 (push_reload): Likewise.
10212 (combine_reloads): Likewise.
10213 (find_dummy_reload): Likewise.
10214 (find_reloads): Likewise.
10215 * reload1.c (find_reg): Likewise.
10216 (set_reload_reg): Likewise.
10217 (allocate_reload_reg): Likewise.
10218 (choose_reload_regs): Likewise.
10219 (reload_adjust_reg_for_temp): Likewise.
10220 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10221 (simplify_subreg_regno): Likewise.
10222 * sel-sched.c (init_regs_for_mode): Likewise.
10223 * varasm.c (make_decl_rtl): Likewise.
10224 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10225 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10226 HARD_REGNO_MODE_OK.
10227 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10228 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10229 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10230 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10231 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10232 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10233 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10234 (arc_mode_class): Delete.
10235 (HARD_REGNO_MODE_OK): Delete.
10236 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10237 (arc_hard_regno_mode_ok): Rename old array to...
10238 (arc_hard_regno_mode_ok_modes): ...this.
10239 (arc_conditional_register_usage): Update accordingly.
10240 (arc_mode_class): Make static.
10241 (arc_hard_regno_mode_ok): New function.
10242 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10243 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10244 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10245 (arm_hard_regno_mode_ok): Make static.
10246 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10247 HARD_REGNO_MODE_OK.
10248 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10249 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10250 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10252 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10253 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10254 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10255 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10256 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10257 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10258 * config/bfin/predicates.md (valid_reg_operand): Use
10259 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10260 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10261 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10262 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10263 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10264 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10265 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10266 (cr16_hard_regno_mode_ok): Make static and return a bool.
10267 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10268 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10269 (cris_hard_regno_mode_ok): New function.
10270 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10271 (epiphany_mode_class): Delete.
10272 (HARD_REGNO_MODE_OK): Delete.
10273 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10274 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10275 (hard_regno_mode_ok): Rename to...
10276 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10277 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10278 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10279 HARD_REGNO_MODE_OK.
10280 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10281 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10282 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10283 (frv_hard_regno_mode_ok): Make static and return a bool.
10284 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10285 HARD_REGNO_MODE_OK.
10286 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10287 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10288 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10289 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10291 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10292 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10293 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10294 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10296 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10297 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10298 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10299 (ia64_hard_regno_mode_ok): New function.
10300 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10301 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10302 (iq2000_hard_regno_mode_ok): New function.
10303 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10304 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10305 (lm32_hard_regno_mode_ok): New function.
10306 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10307 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10308 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10309 instead of HARD_REGNO_MODE_OK.
10310 (m32c_hard_regno_ok): Rename to...
10311 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10312 (m32c_cannot_change_mode_class): Update accordingly.
10313 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10314 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10315 (m32r_mode_class): Delete.
10316 (HARD_REGNO_MODE_OK): Delete.
10317 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10318 (m32r_hard_regno_mode_ok): Rename to...
10319 (m32r_hard_regno_modes): ...this.
10320 (m32r_mode_class): Make static.
10321 (m32r_hard_regno_mode_ok): New function.
10322 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10323 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10324 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10325 (m68k_hard_regno_mode_ok): Make static.
10326 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10327 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10328 (mcore_hard_regno_mode_ok): New function.
10329 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10330 (HARD_REGNO_MODE_OK): Delete.
10331 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10333 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10334 (microblaze_hard_regno_mode_ok): New function.
10335 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10336 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10337 (mips_hard_regno_mode_ok): Delete.
10338 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10339 (mips_hard_regno_mode_ok_p): ...this and make static.
10340 (mips_hard_regno_mode_ok_p): Rename to...
10341 (mips_hard_regno_mode_ok_uncached): ...this.
10342 (mips_hard_regno_mode_ok): New function.
10343 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10344 of HARD_REGNO_MODE_OK.
10345 (mips_option_override): Update after above name changes.
10346 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10347 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10348 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10349 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10350 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10351 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10352 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10353 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10354 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10355 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10356 (msp430_hard_regno_mode_ok): Make static and return a bool.
10357 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10358 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10359 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10361 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10362 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10363 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10364 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10365 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10366 (PA_HARD_REGNO_MODE_OK): ...this
10367 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10368 (PA_HARD_REGNO_MODE_OK): ...this.
10369 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10370 (pa_hard_regno_mode_ok): New function.
10371 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10372 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10373 (pdp11_hard_regno_mode_ok): New function.
10374 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10375 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10377 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10379 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10380 (rs6000_hard_regno_mode_ok): Rename to...
10381 (rs6000_hard_regno_mode_ok_uncached): ...this.
10382 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10383 (rs6000_hard_regno_mode_ok): New function.
10384 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
10385 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
10386 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
10387 (riscv_hard_regno_mode_ok): ...this and make static.
10388 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10389 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
10390 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
10391 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10392 (rl78_hard_regno_mode_ok): Make static and return bool.
10393 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
10394 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
10396 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
10397 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10398 (rs6000_hard_regno_mode_ok): Rename to...
10399 (rs6000_hard_regno_mode_ok_uncached): ...this.
10400 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10401 (rs6000_hard_regno_mode_ok): New function.
10402 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
10403 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
10404 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10405 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
10406 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
10407 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
10408 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10409 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
10410 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
10411 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10412 (sh_hard_regno_mode_ok): Make static.
10413 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
10414 instead of HARD_REGNO_MODE_OK.
10415 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
10416 (sparc_mode_class): Delete.
10417 (HARD_REGNO_MODE_OK): Delete.
10418 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10419 (hard_regno_mode_classes): Make static.
10420 (sparc_mode_class): Likewise.
10421 (sparc_hard_regno_mode_ok): New function.
10422 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
10423 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
10424 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
10426 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10427 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
10428 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
10429 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
10430 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
10431 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10432 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
10433 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
10434 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10435 (visium_hard_regno_mode_ok): New function.
10436 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
10437 instead of HARD_REGNO_MODE_OK.
10438 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
10439 (HARD_REGNO_MODE_OK): Delete.
10440 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
10441 (xtensa_hard_regno_mode_ok_p): ...this and make static.
10442 (xtensa_option_override): Update accordingly.
10443 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10444 (xtensa_hard_regno_mode_ok): New function.
10445 * system.h (HARD_REGNO_MODE_OK): Poison.
10447 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10448 Alan Hayward <alan.hayward@arm.com>
10449 David Sherwood <david.sherwood@arm.com>
10451 * target.def (hard_regno_call_part_clobbered): New hook.
10452 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
10453 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
10454 * doc/tm.texi: Regenerate.
10455 * hooks.h (hook_bool_uint_mode_false): Declare.
10456 * hooks.c (hook_bool_uint_mode_false): New function.
10457 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10458 * cselib.c (cselib_process_insn): Use
10459 targetm.hard_regno_call_part_clobbered instead of
10460 HARD_REGNO_CALL_PART_CLOBBERED.
10461 * ira-conflicts.c (ira_build_conflicts): Likewise.
10462 * ira-costs.c (ira_tune_allocno_costs): Likewise.
10463 * lra-constraints.c (need_for_call_save_p): Likewise.
10464 * lra-lives.c: Include target.h.
10465 (check_pseudos_live_through_calls): Use
10466 targetm.hard_regno_call_part_clobbered instead of
10467 HARD_REGNO_CALL_PART_CLOBBERED.
10468 * regcprop.c: Include target.h.
10469 (copyprop_hardreg_forward_1): Use
10470 targetm.hard_regno_call_part_clobbered instead of
10471 HARD_REGNO_CALL_PART_CLOBBERED.
10472 * reginfo.c (choose_hard_reg_mode): Likewise.
10473 * regrename.c (check_new_reg_p): Likewise.
10474 * reload.c (find_equiv_reg): Likewise.
10475 * reload1.c (emit_reload_insns): Likewise.
10476 * sched-deps.c (deps_analyze_insn): Likewise.
10477 * sel-sched.c (init_regs_for_mode): Likewise.
10478 (mark_unavailable_hard_regs): Likewise.
10479 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
10480 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10481 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
10483 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10484 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10485 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
10487 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
10489 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10490 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10491 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
10493 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10494 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10495 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
10497 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10498 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
10500 * config/powerpcspe/powerpcspe.c
10501 (rs6000_hard_regno_call_part_clobbered): New function.
10502 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10503 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10504 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
10506 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10507 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10508 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
10510 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10511 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10512 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
10514 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10515 Alan Hayward <alan.hayward@arm.com>
10516 David Sherwood <david.sherwood@arm.com>
10518 * rtl.h (subreg_memory_offset): Declare.
10519 * emit-rtl.c (subreg_memory_offset): New function.
10520 * expmed.c (store_bit_field_1): Use it.
10521 * expr.c (undefined_operand_subword_p): Likewise.
10522 * simplify-rtx.c (simplify_subreg): Likewise.
10524 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
10526 PR rtl-optimization/57448
10529 * optabs.c (expand_atomic_load): Place compiler memory barriers if
10530 using atomic_load pattern.
10531 (expand_atomic_store): Likewise.
10533 2017-09-04 Jakub Jelinek <jakub@redhat.com>
10536 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
10537 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
10538 handling. Use replace_call_with_value with NULL instead of
10539 gsi_replace, unlink_stmt_vdef and release_defs.
10541 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
10544 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
10546 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10549 * rtl.h (emit_library_call_value_1): Declare.
10550 (emit_library_call): Replace declaration with a series of overloads.
10551 Remove the parameter count argument.
10552 (emit_library_call_value): Likewise.
10553 * calls.c (emit_library_call_value_1): Make global. Replace varargs
10554 with an "rtx_mode_t *".
10555 (emit_library_call_value): Delete.
10556 (emit_library_call): Likewise.
10557 * asan.c (asan_emit_stack_protection): Update calls accordingly.
10558 (asan_emit_allocas_unpoison): Likewise.
10559 * builtins.c (expand_builtin_powi): Likewise.
10560 (expand_asan_emit_allocas_unpoison): Likewise.
10561 * cfgexpand.c (expand_main_function): Likewise.
10562 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
10563 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
10564 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
10565 * config/arm/arm.c (arm_trampoline_init): Likewise.
10566 (arm_call_tls_get_addr): Likewise.
10567 (arm_expand_divmod_libfunc): Likewise.
10568 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
10569 (smulsi3_highpart): Likewise.
10570 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
10571 (c6x_expand_compare): Likewise.
10572 (c6x_expand_movmem): Likewise.
10573 * config/frv/frv.c (frv_trampoline_init): Likewise.
10574 * config/i386/i386.c (ix86_trampoline_init): Likewise.
10575 (ix86_expand_divmod_libfunc): Likewise.
10576 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
10577 (ia64_expand_compare): Likewise.
10578 (ia64_profile_hook): Likewise.
10579 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
10580 (nonlocal_goto): Likewise.
10581 (restore_stack_nonlocal): Likewise.
10582 * config/m32r/m32r.c (block_move_call): Likewise.
10583 (m32r_trampoline_init): Likewise.
10584 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
10585 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
10586 (m68k_call_m68k_read_tp): Likewise.
10587 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
10588 (microblaze_expand_divide): Likewise.
10589 * config/mips/mips.h (mips_args): Likewise.
10590 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
10591 (MIPS_ICACHE_SYNC): Likewise.
10592 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
10593 (nios2_trampoline_init): Likewise.
10594 * config/pa/pa.c (hppa_tls_call): Likewise.
10595 (pa_trampoline_init): Likewise.
10596 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
10597 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
10598 (expand_strn_compare): Likewise.
10599 (rs6000_generate_compare): Likewise.
10600 (rs6000_expand_float128_convert): Likewise.
10601 (output_profile_hook): Likewise.
10602 (rs6000_trampoline_init): Likewise.
10603 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
10604 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
10605 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
10606 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
10607 (rs6000_generate_compare): Likewise.
10608 (rs6000_expand_float128_convert): Likewise.
10609 (output_profile_hook): Likewise.
10610 (rs6000_trampoline_init): Likewise.
10611 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
10612 * config/sh/sh.c (sh_trampoline_init): Likewise.
10613 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
10614 (sparc_emit_float_lib_cmp): Likewise.
10615 (sparc32_initialize_trampoline): Likewise.
10616 (sparc64_initialize_trampoline): Likewise.
10617 (sparc_profile_hook): Likewise.
10618 * config/spu/spu.c (ea_load_store): Likewise.
10619 * config/spu/spu.md (floatunssidf2): Likewise.
10620 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
10621 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
10622 * config/visium/visium.c (expand_block_move_4): Likewise.
10623 (expand_block_move_2): Likewise.
10624 (expand_block_move_1): Likewise.
10625 (expand_block_set_4): Likewise.
10626 (expand_block_set_2): Likewise.
10627 (expand_block_set_1): Likewise.
10628 (visium_trampoline_init): Likewise.
10629 (visium_profile_hook): Likewise.
10630 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
10631 (xtensa_setup_frame_addresses): Likewise.
10632 (xtensa_trampoline_init): Likewise.
10633 * except.c (sjlj_emit_function_enter): Likewise.
10634 (sjlj_emit_function_exit): Likewise.
10635 * explow.c (allocate_dynamic_stack_space): Likewise.
10636 (probe_stack_range): Likewise.
10637 * expr.c (convert_mode_scalar): Likewise.
10638 * optabs.c (expand_binop): Likewise.
10639 (expand_twoval_binop_libfunc): Likewise.
10640 (expand_unop): Likewise.
10641 (prepare_cmp_insn): Likewise.
10642 (prepare_float_lib_cmp): Likewise.
10643 (expand_float): Likewise.
10644 (expand_fix): Likewise.
10645 (expand_fixed_convert): Likewise.
10646 (maybe_emit_sync_lock_test_and_set): Likewise.
10647 (expand_atomic_compare_and_swap): Likewise.
10648 (expand_mem_thread_fence): Likewise.
10649 (expand_atomic_fetch_op): Likewise.
10651 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
10653 * doc/generic.texi (OpenACC): Adjust URL.
10654 * doc/invoke.texi (C Dialect Options): Ditto.
10656 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
10658 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
10659 predicate for operand 1. Add (m,<S>) constraint.
10660 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
10661 Prevent memory operand 1 with register operand 2.
10663 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
10665 PR rtl-optimization/82024
10666 * combine.c (try_combine): If the combination result is a PARALLEL,
10667 and we only need to retain the SET in there that would be placed
10668 at I2, check that we can place that at I3 instead, before doing so.
10670 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10673 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
10675 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
10676 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
10679 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
10680 Jeff Law <law@redhat.com>
10682 * varasm.c (bss_initializer_p): Do not put constants into .bss
10683 (categorize_decl_for_section): Handle bss_initializer_p returning
10684 false when DECL_INITIAL is NULL.
10686 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10689 * config/s390/s390.c (s390_can_inline_p): New function.
10691 2017-09-01 Jeff Law <law@redhat.com>
10693 PR tree-optimization/82052
10694 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
10695 Always initialize the returned slot after a hash table miss
10696 when INSERT is true.
10698 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
10700 * config/s390/s390.md (mem_signal_fence): Remove.
10701 * doc/md.texi (mem_signal_fence): Remove.
10702 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
10704 * target-insns.def (mem_signal_fence): Remove.
10706 2017-09-01 Jakub Jelinek <jakub@redhat.com>
10709 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
10712 * asan.c (create_odr_indicator): Strip name encoding from assembler
10713 name before appending it after __odr_asan_.
10715 2017-09-01 Martin Liska <mliska@suse.cz>
10717 PR tree-optimization/82059
10718 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
10719 frequency only when an edge is redirected.
10721 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10723 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
10724 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
10725 (arc_conditional_register_usage): Remove ARC600 lp_count
10727 (arc_file_start): Emit Tag_ARC_CPU_variation.
10728 (arc_can_use_doloop_p): New conditions to use ZOLs.
10729 (hwloop_fail): New function.
10730 (hwloop_optimize): Likewise.
10731 (hwloop_pattern_reg): Likewise.
10732 (arc_doloop_hooks): New struct, to be used with reorg_loops.
10733 (arc_reorg_loops): New function, calls reorg_loops.
10734 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
10735 (arc600_corereg_hazard): Remove ZOL checking, case handled by
10737 (arc_loop_hazard): Remove function, functionality moved into
10739 (arc_hazard): Remove arc_loop_hazard call.
10740 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
10741 into hwloop_optimize.
10742 (arc_label_align): Remove ZOL handling.
10743 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
10744 * config/arc/arc.md (doloop_begin): Remove pattern.
10745 (doloop_begin_i): Likewise.
10746 (doloop_end_i): Likewise.
10747 (doloop_fallback): Likewise.
10748 (doloop_fallback_m): Likewise.
10749 (doloop_end): Reimplement expand.
10750 (arc_lp): New pattern for LP instruction.
10751 (loop_end): New pattern.
10752 (loop_fail): Likewise.
10753 (decrement_and_branch_until_zero): Likewise.
10754 * config/arc/arc.opt (mlpc-width): New option.
10755 * doc/invoke.texi (mlpc-width): Document option.
10757 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10759 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
10760 (arc_ccfsm_advance): Fix checking for delay slots.
10761 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
10763 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10765 * config/arc/arc.md (movqi_insn): Add stores to save constant long
10767 (movhi_insn): Update store instruction constraint which are saving
10768 6-bit short immediates.
10769 (movsi_insn): Consider also short scaled load operations.
10770 (zero_extendhisi2_i): Use Usd constraint instead of T.
10771 (extendhisi2_i): Add q constraint.
10772 (arc_clzsi2): Add type and length attributes.
10773 (arc_ctzsi2): Likewise.
10774 * config/arc/constraints.md (Usc): Update constraint, the
10775 assembler can parse two relocations for a single instruction.
10777 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
10779 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
10780 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
10782 2017-08-31 Olivier Hainque <hainque@adacore.com>
10784 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
10785 match as powerpc-wrs-vxworks*.
10787 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
10789 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
10790 register constraint for by-element operand.
10791 (aarch64_mls_elt_merge<mode>): Likewise.
10793 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10795 * config/arc/arc.c (arc_can_follow_jump): Check for short
10798 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10800 * config.gcc: Use g.opt for arc.
10801 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
10802 functionality moved to ...
10803 (legitimate_scaled_address_p): New function, ...here.
10804 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
10805 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
10806 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
10808 (arc_override_options): Handle G option.
10809 (arc_output_pic_addr_const): Correct function definition.
10810 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
10811 (arc_decl_anon_ns_mem_p): Delete.
10812 (arc_in_small_data_p): Overhaul this function to take into
10813 consideration the value given via G option.
10814 (arc_rewrite_small_data_1): Renamed and corrected old
10815 arc_rewrite_small_data function.
10816 (arc_rewrite_small_data): New function.
10817 (small_data_pattern): Don't use pic_offset_table_rtx.
10818 * config/arc/arc.h (CC1_SPEC): Recognize G option.
10819 * config/arc/simdext.md (movmisalignv2hi): Use
10820 prepare_move_operands function.
10822 (movmisalign*): Likewise.
10823 * doc/invoke.texi (ARC options): Document -G option.
10825 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10827 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
10829 * config/arc/arc.c (arc_print_operand): Output scalled address for
10830 sdata whenever is possible.
10831 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10832 load/stores are available.
10833 (compact_sda_memory_operand): Check for the alignment required by
10834 code density instructions.
10835 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10837 * config/arc/constraints.md (Usd): Update constraint.
10838 (Us0): New constraint.
10839 (Usc): Update constraint.
10841 2017-08-31 Richard Biener <rguenther@suse.de>
10843 PR middle-end/82054
10844 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10845 function only once.
10847 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10849 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10850 Resize type_signature.
10852 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10853 Alan Hayward <alan.hayward@arm.com>
10854 David Sherwood <david.sherwood@arm.com>
10856 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10857 subregs whose inner modes can be stored in GPRs.
10858 (aarch64_classify_index): Likewise.
10860 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10861 Alan Hayward <alan.hayward@arm.com>
10862 David Sherwood <david.sherwood@arm.com>
10864 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10865 (V_INT_EQUIV): ...this.
10866 (v_cmp_result): Rename to...
10867 (v_int_equiv): ...this.
10868 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10869 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10870 (copysign<mode>3): Likewise.
10871 (aarch64_simd_bsl<mode>_internal): Likewise.
10872 (aarch64_simd_bsl<mode>): Likewise.
10873 (vec_cmp<mode><mode>): Likewise.
10874 (vcond<mode><mode>): Likewise.
10875 (vcond<v_cmp_mixed><mode>): Likewise.
10876 (vcondu<mode><v_cmp_mixed>): Likewise.
10877 (aarch64_cm<optab><mode>): Likewise.
10878 (aarch64_cmtst<mode>): Likewise.
10879 (aarch64_fac<optab><mode>): Likewise.
10880 (vec_perm_const<mode>): Likewise.
10881 (vcond_mask_<mode><v_cmp_result>): Rename to...
10882 (vcond_mask_<mode><v_int_equiv>): ...this.
10883 (vec_cmp<mode><v_cmp_result>): Rename to...
10884 (vec_cmp<mode><v_int_equiv>): ...this.
10886 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10887 Alan Hayward <alan.hayward@arm.com>
10888 David Sherwood <david.sherwood@arm.com>
10890 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
10892 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
10893 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
10894 (UNSPEC_LD4_DREG): New unspecs.
10895 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
10896 (aarch64_ld2<mode>_dreg_be): Replace with...
10897 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
10899 (aarch64_ld3<mode>_dreg_le)
10900 (aarch64_ld3<mode>_dreg_be): Replace with...
10901 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10903 (aarch64_ld4<mode>_dreg_le)
10904 (aarch64_ld4<mode>_dreg_be): Replace with...
10905 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10908 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10910 PR tree-optimization/81987
10911 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10912 insert an initializer in a location not dominated by the stride
10915 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10917 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10918 on the entire header of the finally block in the fallthru case.
10920 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10922 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10924 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10926 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10927 rs6000_emit_move_from_cr and call renamed function.
10928 (rs6000_emit_prologue): Call renamed functions.
10929 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10930 movesi_from_cr, remove volatile CRs.
10932 2017-08-30 Jon Beniston <jon@beniston.com>
10933 Richard Biener <rguenther@suse.de>
10935 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
10936 of VECTOR_MODE_P check.
10937 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
10938 element vector types.
10940 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10942 * df.h (df_read_modify_subreg_p): Remove in favor of...
10943 * rtl.h (read_modify_subreg_p): ...this new function. Take a
10944 const_rtx instead of an rtx.
10945 * cprop.c (local_cprop_find_used_regs): Update accordingly.
10946 * df-problems.c (df_word_lr_mark_ref): Likewise.
10947 * ira-lives.c (mark_pseudo_reg_live): Likewise.
10948 (mark_pseudo_reg_dead): Likewise.
10949 (mark_ref_dead): Likewise.
10950 * reginfo.c (init_subregs_of_mode): Likewise.
10951 * sched-deps.c (sched_analyze_1): Likewise.
10952 * df-scan.c (df_def_record_1): Likewise.
10953 (df_uses_record): Likewise.
10954 (df_read_modify_subreg_p): Remove in favor of...
10955 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
10956 const_rtx instead of an rtx.
10958 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10959 Alan Hayward <alan.hayward@arm.com>
10960 David Sherwood <david.sherwood@arm.com>
10962 * rtl.h (partial_subreg_p): New function.
10963 * caller-save.c (save_call_clobbered_regs): Use it.
10964 * calls.c (expand_call): Likewise.
10965 * combine.c (combinable_i3pat): Likewise.
10966 (simplify_set): Likewise.
10967 (make_extraction): Likewise.
10968 (make_compound_operation_int): Likewise.
10969 (gen_lowpart_or_truncate): Likewise.
10970 (force_to_mode): Likewise.
10971 (make_field_assignment): Likewise.
10972 (reg_truncated_to_mode): Likewise.
10973 (record_truncated_value): Likewise.
10974 (move_deaths): Likewise.
10975 * cse.c (record_jump_cond): Likewise.
10976 (cse_insn): Likewise.
10977 * cselib.c (cselib_lookup_1): Likewise.
10978 * expmed.c (extract_bit_field_using_extv): Likewise.
10979 * function.c (assign_parm_setup_reg): Likewise.
10980 * ifcvt.c (noce_convert_multiple_sets): Likewise.
10981 * ira-build.c (create_insn_allocnos): Likewise.
10982 * lra-coalesce.c (merge_pseudos): Likewise.
10983 * lra-constraints.c (match_reload): Likewise.
10984 (simplify_operand_subreg): Likewise.
10985 (curr_insn_transform): Likewise.
10986 * lra-lives.c (process_bb_lives): Likewise.
10987 * lra.c (new_insn_reg): Likewise.
10988 (lra_substitute_pseudo): Likewise.
10989 * regcprop.c (mode_change_ok): Likewise.
10990 (maybe_mode_change): Likewise.
10991 (copyprop_hardreg_forward_1): Likewise.
10992 * reload.c (push_reload): Likewise.
10993 (find_reloads): Likewise.
10994 (find_reloads_subreg_address): Likewise.
10995 * reload1.c (alter_reg): Likewise.
10996 (eliminate_regs_1): Likewise.
10997 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10999 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11001 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11002 back to if statements, including unpack.
11004 2017-08-30 Martin Liska <mliska@suse.cz>
11006 PR inline-asm/82001
11007 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11009 (func_checker::compare_asm_inputs_outputs): ... this function.
11010 (func_checker::compare_gimple_asm): Use the function to compare
11011 also ASM constrains.
11012 * ipa-icf-gimple.h: Rename the function.
11014 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11015 Alan Hayward <alan.hayward@arm.com>
11016 David Sherwood <david.sherwood@arm.com>
11018 * coretypes.h (complex_mode): New type.
11019 * gdbhooks.py (build_pretty_printer): Handle it.
11020 * machmode.h (complex_mode): New class.
11021 (complex_mode::includes_p): New function.
11022 (is_complex_int_mode): Likewise.
11023 (is_complex_float_mode): Likewise.
11024 * genmodes.c (get_mode_class): Handle complex mode classes.
11025 * function.c (expand_function_end): Use is_complex_int_mode.
11027 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11028 Alan Hayward <alan.hayward@arm.com>
11029 David Sherwood <david.sherwood@arm.com>
11031 * coretypes.h (scalar_mode_pod): New typedef.
11032 * gdbhooks.py (build_pretty_printer): Handle it.
11033 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11034 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11035 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11036 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11037 as_a <scalar_mode>.
11039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11040 Alan Hayward <alan.hayward@arm.com>
11041 David Sherwood <david.sherwood@arm.com>
11043 * machmode.h (mode_for_vector): Take a scalar_mode instead
11045 * stor-layout.c (mode_for_vector): Likewise.
11046 * explow.c (promote_mode): Use as_a <scalar_mode>.
11047 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11049 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11050 Alan Hayward <alan.hayward@arm.com>
11051 David Sherwood <david.sherwood@arm.com>
11053 * target.def (preferred_simd_mode): Take a scalar_mode
11054 instead of a machine_mode.
11055 * targhooks.h (default_preferred_simd_mode): Likewise.
11056 * targhooks.c (default_preferred_simd_mode): Likewise.
11057 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11058 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11059 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11060 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11061 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11062 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11063 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11064 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11066 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11067 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11068 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11069 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11070 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11071 * doc/tm.texi: Regenerate.
11072 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11075 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11076 Alan Hayward <alan.hayward@arm.com>
11077 David Sherwood <david.sherwood@arm.com>
11079 * target.def (scalar_mode_supported_p): Take a scalar_mode
11080 instead of a machine_mode.
11081 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11082 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11083 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11084 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11085 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11086 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11087 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11088 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11089 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11090 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11091 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11092 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11093 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11094 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11096 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11097 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11098 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11099 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11100 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11102 * doc/tm.texi: Regenerate.
11104 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11105 Alan Hayward <alan.hayward@arm.com>
11106 David Sherwood <david.sherwood@arm.com>
11108 * coretypes.h (opt_scalar_mode): New typedef.
11109 * gdbhooks.py (build_pretty_printers): Handle it.
11110 * machmode.h (mode_iterator::get_2xwider): Add overload for
11112 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11114 * expr.c (convert_mode_scalar): Likewise.
11115 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11116 * optabs.c (expand_float): Likewise.
11117 (expand_fix): Likewise.
11118 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11120 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11121 Alan Hayward <alan.hayward@arm.com>
11122 David Sherwood <david.sherwood@arm.com>
11124 * optabs.c (expand_float): Explicitly check for scalars before
11125 using a branching expansion.
11126 (expand_fix): Likewise.
11128 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11129 Alan Hayward <alan.hayward@arm.com>
11130 David Sherwood <david.sherwood@arm.com>
11132 * expr.c (convert_mode): Split scalar handling out into...
11133 (convert_mode_scalar): ...this new function. Treat the modes
11136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11137 Alan Hayward <alan.hayward@arm.com>
11138 David Sherwood <david.sherwood@arm.com>
11140 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11142 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11144 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11145 Alan Hayward <alan.hayward@arm.com>
11146 David Sherwood <david.sherwood@arm.com>
11148 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11149 rather than a machine_mode.
11150 (fixed_from_string): Likewise.
11151 (fixed_convert): Likewise.
11152 (fixed_convert_from_int): Likewise.
11153 (fixed_convert_from_real): Likewise.
11154 (real_convert_from_fixed): Likewise.
11155 * fixed-value.c (fixed_from_double_int): Likewise.
11156 (fixed_from_string): Likewise.
11157 (fixed_convert): Likewise.
11158 (fixed_convert_from_int): Likewise.
11159 (fixed_convert_from_real): Likewise.
11160 (real_convert_from_fixed): Likewise.
11161 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11163 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11164 Alan Hayward <alan.hayward@arm.com>
11165 David Sherwood <david.sherwood@arm.com>
11167 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11168 of separate mode class checks. Do not allow vector modes here.
11169 (immed_wide_int_const): Use as_a <scalar_mode>.
11170 * explow.c (trunc_int_for_mode): Likewise.
11171 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11172 (wi::shwi): Likewise.
11173 (wi::min_value): Likewise.
11174 (wi::max_value): Likewise.
11175 * dwarf2out.c (loc_descriptor): Likewise.
11176 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11177 for CONST_WIDE_INT.
11179 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11180 Alan Hayward <alan.hayward@arm.com>
11181 David Sherwood <david.sherwood@arm.com>
11183 * tree.h (SCALAR_TYPE_MODE): New macro.
11184 * expr.c (expand_expr_addr_expr_1): Use it.
11185 (expand_expr_real_2): Likewise.
11186 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11187 (fold_convert_const_fixed_from_int): Likewise.
11188 (fold_convert_const_fixed_from_real): Likewise.
11189 (native_encode_fixed): Likewise
11190 (native_encode_complex): Likewise
11191 (native_encode_vector): Likewise.
11192 (native_interpret_fixed): Likewise.
11193 (native_interpret_real): Likewise.
11194 (native_interpret_complex): Likewise.
11195 (native_interpret_vector): Likewise.
11196 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11197 (simd_clone_adjust_argument_types): Likewise.
11198 (simd_clone_init_simd_arrays): Likewise.
11199 (simd_clone_adjust): Likewise.
11200 * stor-layout.c (layout_type): Likewise.
11201 * tree.c (build_minus_one_cst): Likewise.
11202 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11203 * tree-inline.c (estimate_move_cost): Likewise.
11204 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11205 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11206 (vectorizable_reduction): Likewise.
11207 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11208 (vect_recog_mixed_size_cond_pattern): Likewise.
11209 (check_bool_pattern): Likewise.
11210 (adjust_bool_pattern): Likewise.
11211 (search_type_for_mask_1): Likewise.
11212 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11213 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11214 (vectorizable_load): Likewise.
11215 (vectorizable_store): Likewise.
11216 * ubsan.c (ubsan_encode_value): Likewise.
11217 * varasm.c (output_constant): Likewise.
11219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11220 Alan Hayward <alan.hayward@arm.com>
11221 David Sherwood <david.sherwood@arm.com>
11223 * coretypes.h (scalar_mode): New class.
11224 * machmode.h (scalar_mode): Likewise.
11225 (scalar_mode::includes_p): New function.
11226 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11227 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11228 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11229 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11230 * expmed.c (store_bit_field_1): Likewise.
11231 (extract_bit_field_1): Likewise.
11232 * expr.c (write_complex_part): Likewise.
11233 (read_complex_part): Likewise.
11234 (emit_move_complex_push): Likewise.
11235 (expand_expr_real_2): Likewise.
11236 * function.c (assign_parm_setup_reg): Likewise.
11237 (assign_parms_unsplit_complex): Likewise.
11238 * optabs.c (expand_binop): Likewise.
11239 * rtlanal.c (subreg_get_info): Likewise.
11240 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11241 * varasm.c (output_constant_pool_2): Likewise.
11243 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11244 Alan Hayward <alan.hayward@arm.com>
11245 David Sherwood <david.sherwood@arm.com>
11247 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11249 (expmed_mult_highpart_optab): Likewise.
11250 (expmed_mult_highpart): Likewise.
11252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11253 Alan Hayward <alan.hayward@arm.com>
11254 David Sherwood <david.sherwood@arm.com>
11256 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11257 instead of a machine_mode.
11258 (builtin_memset_read_str): Likewise.
11259 * builtins.c (c_readstr): Likewise.
11260 (builtin_memcpy_read_str): Likewise.
11261 (builtin_strncpy_read_str): Likewise.
11262 (builtin_memset_read_str): Likewise.
11263 (builtin_memset_gen_str): Likewise.
11264 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11265 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11266 instead of a machine_mode.
11267 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11269 (make_extraction): Likewise.
11270 (try_widen_shift_mode): Take and return scalar_int_modes instead
11272 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11273 a scalar_int_mode instead of a machine_mode.
11274 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11275 (avr_addr_space_pointer_mode): Likewise.
11276 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11277 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11278 (msp430_unwind_word_mode): Likewise.
11279 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11280 (spu_addr_space_pointer_mode): Likewise.
11281 (spu_addr_space_address_mode): Likewise.
11282 (spu_libgcc_cmp_return_mode): Likewise.
11283 (spu_libgcc_shift_count_mode): Likewise.
11284 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11285 (rl78_addr_space_pointer_mode): Likewise.
11286 (fl78_unwind_word_mode): Likewise.
11287 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11289 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11290 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11291 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11292 (mips_valid_pointer_mode): Likewise.
11293 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11294 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11295 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11297 (ft32_addr_space_address_mode): Likewise.
11298 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11299 scalar_int_mode instead of a machine_mode.
11300 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11302 (m32c_addr_space_address_mode): Likewise.
11303 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11304 (rs6000_eh_return_filter_mode): Likewise.
11305 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11306 (rs6000_eh_return_filter_mode): Likewise.
11307 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11308 (s390_libgcc_shift_count_mode): Likewise.
11309 (s390_unwind_word_mode): Likewise.
11310 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11312 * target.def (mode_rep_extended): Likewise.
11313 (valid_pointer_mode): Likewise.
11314 (addr_space.valid_pointer_mode): Likewise.
11315 (eh_return_filter_mode): Return a scalar_int_mode rather than
11317 (libgcc_cmp_return_mode): Likewise.
11318 (libgcc_shift_count_mode): Likewise.
11319 (unwind_word_mode): Likewise.
11320 (addr_space.pointer_mode): Likewise.
11321 (addr_space.address_mode): Likewise.
11322 * doc/tm.texi: Regenerate.
11323 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11325 (do_jump): Use scalar_int_mode for local variables.
11326 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11327 rather than a machine_mode.
11328 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11329 (scompare_loc_descriptor_wide): Likewise.
11330 (scompare_loc_descriptor_narrow): Likewise.
11331 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11333 * except.c (sjlj_emit_dispatch_table): Likewise.
11334 (expand_builtin_eh_copy_values): Likewise.
11335 * explow.c (convert_memory_address_addr_space_1): Likewise.
11336 Take a scalar_int_mode rather than a machine_mode.
11337 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11338 than a machine_mode.
11339 (memory_address_addr_space): Use scalar_int_mode for local variables.
11340 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11341 rather than a machine_mode.
11342 * expmed.c (mask_rtx): Likewise.
11343 (init_expmed_one_conv): Likewise.
11344 (expand_mult_highpart_adjust): Likewise.
11345 (extract_high_half): Likewise.
11346 (expmed_mult_highpart_optab): Likewise.
11347 (expmed_mult_highpart): Likewise.
11348 (expand_smod_pow2): Likewise.
11349 (expand_sdiv_pow2): Likewise.
11350 (emit_store_flag_int): Likewise.
11351 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11353 (extract_low_bits): Likewise.
11354 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11356 * expr.c (pieces_addr::adjust): Likewise.
11357 (can_store_by_pieces): Likewise.
11358 (store_by_pieces): Likewise.
11359 (clear_by_pieces_1): Likewise.
11360 (expand_expr_addr_expr_1): Likewise.
11361 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11362 (expand_expr_real_1): Likewise.
11363 (try_casesi): Likewise.
11364 * final.c (shorten_branches): Likewise.
11365 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11366 type of "mode" to machine_mode.
11367 * internal-fn.c (expand_arith_overflow_result_store): Take a
11368 scalar_int_mode rather than a machine_mode.
11369 (expand_mul_overflow): Use scalar_int_mode for local variables.
11370 * loop-doloop.c (doloop_modify): Likewise.
11371 (doloop_optimize): Likewise.
11372 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11373 than a machine_mode.
11374 (expand_doubleword_shift_condmove): Likewise.
11375 (expand_doubleword_shift): Likewise.
11376 (expand_doubleword_clz): Likewise.
11377 (expand_doubleword_popcount): Likewise.
11378 (expand_doubleword_parity): Likewise.
11379 (expand_absneg_bit): Use scalar_int_mode for local variables.
11380 (prepare_float_lib_cmp): Likewise.
11381 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11382 rather than a machine_mode.
11383 (convert_memory_address_addr_space): Likewise.
11384 (get_mode_bounds): Likewise.
11385 (get_address_mode): Return a scalar_int_mode rather than a
11387 * rtlanal.c (get_address_mode): Likewise.
11388 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
11389 than a machine_mode.
11390 * targhooks.c (default_mode_rep_extended): Likewise.
11391 (default_valid_pointer_mode): Likewise.
11392 (default_addr_space_valid_pointer_mode): Likewise.
11393 (default_eh_return_filter_mode): Return a scalar_int_mode rather
11394 than a machine_mode.
11395 (default_libgcc_cmp_return_mode): Likewise.
11396 (default_libgcc_shift_count_mode): Likewise.
11397 (default_unwind_word_mode): Likewise.
11398 (default_addr_space_pointer_mode): Likewise.
11399 (default_addr_space_address_mode): Likewise.
11400 * targhooks.h (default_eh_return_filter_mode): Likewise.
11401 (default_libgcc_cmp_return_mode): Likewise.
11402 (default_libgcc_shift_count_mode): Likewise.
11403 (default_unwind_word_mode): Likewise.
11404 (default_addr_space_pointer_mode): Likewise.
11405 (default_addr_space_address_mode): Likewise.
11406 (default_mode_rep_extended): Take a scalar_int_mode rather than
11408 (default_valid_pointer_mode): Likewise.
11409 (default_addr_space_valid_pointer_mode): Likewise.
11410 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
11412 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
11413 rather than a machine_mode.
11414 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
11415 for local variables.
11416 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11417 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
11418 than a machine_mode.
11420 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11421 Alan Hayward <alan.hayward@arm.com>
11422 David Sherwood <david.sherwood@arm.com>
11424 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
11425 the mode argument to scalar_int_mode.
11426 (do_jump_by_parts_zero_rtx): Likewise.
11427 (do_jump_by_parts_equality_rtx): Likewise.
11428 (do_jump_by_parts_greater): Take a mode argument.
11429 (do_jump_by_parts_equality): Likewise.
11430 (do_jump_1): Update calls accordingly.
11432 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11433 Alan Hayward <alan.hayward@arm.com>
11434 David Sherwood <david.sherwood@arm.com>
11436 * is-a.h (safe_dyn_cast): New function.
11437 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
11438 (jump_table_for_label): Likewise.
11439 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
11440 instead of an rtx_insn *.
11441 (shorten_branches): Use dyn_cast instead of LABEL_P and
11442 JUMP_TABLE_DATA_P. Use jump_table_for_label and
11443 rtx_jump_table_data::get_data_mode.
11444 (final_scan_insn): Likewise.
11446 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11447 Alan Hayward <alan.hayward@arm.com>
11448 David Sherwood <david.sherwood@arm.com>
11450 * combine.c (try_combine): Use is_a <scalar_int_mode> when
11451 trying to combine a full-register integer set with a subreg
11454 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11455 Alan Hayward <alan.hayward@arm.com>
11456 David Sherwood <david.sherwood@arm.com>
11458 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
11459 that is always either address_mode or pointer_mode.
11461 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11462 Alan Hayward <alan.hayward@arm.com>
11463 David Sherwood <david.sherwood@arm.com>
11465 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
11466 when the two are known to be equal.
11468 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11469 Alan Hayward <alan.hayward@arm.com>
11470 David Sherwood <david.sherwood@arm.com>
11472 * simplify-rtx.c (simplify_const_unary_operation): Use
11473 is_a <scalar_int_mode> instead of checking for a nonzero
11474 precision. Forcibly convert op_mode to a scalar_int_mode
11475 in that case. More clearly differentiate the operand and
11476 result modes and use the former when deciding what the value
11477 of a count-bits operation should be. Use is_int_mode instead
11478 of checking for a MODE_INT. Remove redundant check for whether
11479 this mode has a zero precision.
11481 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11482 Alan Hayward <alan.hayward@arm.com>
11483 David Sherwood <david.sherwood@arm.com>
11485 * optabs.c (widen_leading): Change the type of the mode argument
11486 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
11487 (widen_bswap): Likewise.
11488 (expand_parity): Likewise.
11489 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
11490 (expand_ffs): Likewise.
11491 (epand_unop): Check for scalar integer modes before calling the
11494 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11495 Alan Hayward <alan.hayward@arm.com>
11496 David Sherwood <david.sherwood@arm.com>
11498 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
11500 (expand_expr_real_1): Update call accordingly.
11502 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11503 Alan Hayward <alan.hayward@arm.com>
11504 David Sherwood <david.sherwood@arm.com>
11506 * expmed.c (store_bit_field_using_insv): Add op0_mode and
11507 value_mode arguments. Use scalar_int_mode internally.
11508 (store_bit_field_1): Rename the new integer mode from imode
11509 to op0_mode and use it instead of GET_MODE (op0). Update calls
11510 to store_split_bit_field, store_bit_field_using_insv and
11511 store_fixed_bit_field.
11512 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
11513 Use scalar_int_mode internally. Use a bit count rather than a mode
11514 when calculating the largest bit size for get_best_mode.
11515 Update calls to store_split_bit_field and store_fixed_bit_field_1.
11516 (store_fixed_bit_field_1): Add mode and value_mode arguments.
11517 Remove assertion that OP0 has a scalar integer mode.
11518 (store_split_bit_field): Add op0_mode and value_mode arguments.
11519 Update calls to extract_fixed_bit_field.
11520 (extract_bit_field_using_extv): Add an op0_mode argument.
11521 Use scalar_int_mode internally.
11522 (extract_bit_field_1): Rename the new integer mode from imode to
11523 op0_mode and use it instead of GET_MODE (op0). Update calls to
11524 extract_split_bit_field, extract_bit_field_using_extv and
11525 extract_fixed_bit_field.
11526 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
11527 to extract_split_bit_field and extract_fixed_bit_field_1.
11528 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
11529 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
11530 on the target mode.
11531 (extract_split_bit_field): Add an op0_mode argument. Update call
11532 to extract_fixed_bit_field.
11534 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11535 Alan Hayward <alan.hayward@arm.com>
11536 David Sherwood <david.sherwood@arm.com>
11538 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
11539 * explow.c (hard_function_value): Likewise.
11540 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
11541 convert_to_mode call outside the loop.
11542 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
11543 for the mode iterator. Require the mode specified by max_pieces
11545 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
11547 (copy_blkmode_to_reg): Likewise.
11548 (set_storage_via_setmem): Likewise.
11549 * optabs.c (prepare_cmp_insn): Likewise.
11550 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11551 * stor-layout.c (finish_bitfield_representative): Likewise.
11553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11554 Alan Hayward <alan.hayward@arm.com>
11555 David Sherwood <david.sherwood@arm.com>
11557 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
11558 * expr.c (convert_move): Use them.
11559 (convert_modes): Likewise.
11560 (store_expr_with_bounds): Likewise.
11562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11563 Alan Hayward <alan.hayward@arm.com>
11564 David Sherwood <david.sherwood@arm.com>
11566 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
11567 parameter for the mode of "x". Remove the "known_x", "known_mode"
11568 and "known_ret" arguments. Change the type of the mode argument
11569 to scalar_int_mode.
11570 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
11571 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
11572 (reg_num_sign_bit_copies_for_combine): Likewise.
11573 * rtlanal.c (nonzero_bits1): Likewise.
11574 (num_sign_bit_copies1): Likewise.
11575 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
11576 (reg_num_sign_bit_copies_general): Likewise.
11577 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
11578 (reg_nonzero_bits_general): Likewise.
11580 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11581 Alan Hayward <alan.hayward@arm.com>
11582 David Sherwood <david.sherwood@arm.com>
11584 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
11585 than in subroutines. Return 1 for non-integer modes.
11586 (cached_num_sign_bit_copies): Change the type of the mode parameter
11587 to scalar_int_mode.
11588 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
11589 classes. Handle CONST_INT_P first and then check whether X also
11590 has a scalar integer mode. Check the same thing for inner registers
11591 of a SUBREG and for values that are being extended or truncated.
11593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11594 Alan Hayward <alan.hayward@arm.com>
11595 David Sherwood <david.sherwood@arm.com>
11597 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
11598 in subroutines. Return the mode mask for non-integer modes.
11599 (cached_nonzero_bits): Change the type of the mode parameter
11600 to scalar_int_mode.
11601 (nonzero_bits1): Likewise. Remove early exit for other mode
11602 classes. Handle CONST_INT_P first and then check whether X
11603 also has a scalar integer mode.
11605 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11606 Alan Hayward <alan.hayward@arm.com>
11607 David Sherwood <david.sherwood@arm.com>
11609 * expr.c (widest_int_mode_for_size): Make the comment match the code.
11610 Return a scalar_int_mode and assert that the size is greater than
11612 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
11613 (op_by_pieces_d::op_by_pieces_d): Likewise.
11614 (op_by_pieces_d::run): Likewise.
11615 (can_store_by_pieces): Likewise.
11617 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11618 Alan Hayward <alan.hayward@arm.com>
11619 David Sherwood <david.sherwood@arm.com>
11621 * combine.c (extract_left_shift): Add a mode argument and update
11623 (make_compound_operation_int): Change the type of the mode parameter
11624 to scalar_int_mode and update the call to extract_left_shift.
11626 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11627 Alan Hayward <alan.hayward@arm.com>
11628 David Sherwood <david.sherwood@arm.com>
11630 * combine.c (simplify_and_const_int): Change the type of the mode
11631 parameter to scalar_int_mode.
11632 (simplify_and_const_int_1): Likewise. Update recursive call.
11634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11635 Alan Hayward <alan.hayward@arm.com>
11636 David Sherwood <david.sherwood@arm.com>
11638 * combine.c (simplify_compare_const): Check that the mode is a
11639 scalar_int_mode (rather than VOIDmode) before testing its
11641 (simplify_comparison): Move COMPARISON_P handling out of the
11642 loop and restrict the latter part of the loop to scalar_int_modes.
11643 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
11644 and when considering SUBREG_REGs. Use is_int_mode instead of
11645 checking GET_MODE_CLASS against MODE_INT.
11647 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11648 Alan Hayward <alan.hayward@arm.com>
11649 David Sherwood <david.sherwood@arm.com>
11651 * combine.c (try_widen_shift_mode): Move check for equal modes to...
11652 (simplify_shift_const_1): ...here. Use scalar_int_mode for
11653 shift_unit_mode and for modes involved in scalar shifts.
11655 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11656 Alan Hayward <alan.hayward@arm.com>
11657 David Sherwood <david.sherwood@arm.com>
11659 * combine.c (force_int_to_mode): New function, split out from...
11660 (force_to_mode): ...here. Keep xmode up-to-date and use it
11661 instead of GET_MODE (x).
11663 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11664 Alan Hayward <alan.hayward@arm.com>
11665 David Sherwood <david.sherwood@arm.com>
11667 * optabs-query.h (extraction_insn::struct_mode): Change type to
11668 opt_scalar_int_mode and update comment.
11669 (extraction_insn::field_mode): Change type to scalar_int_mode.
11670 (extraction_insn::pos_mode): Likewise.
11671 * combine.c (make_extraction): Update accordingly.
11672 * optabs-query.c (get_traditional_extraction_insn): Likewise.
11673 (get_optab_extraction_insn): Likewise.
11674 * recog.c (simplify_while_replacing): Likewise.
11675 * expmed.c (narrow_bit_field_mem): Change the type of the mode
11676 parameter to opt_scalar_int_mode.
11678 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11679 Alan Hayward <alan.hayward@arm.com>
11680 David Sherwood <david.sherwood@arm.com>
11682 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
11683 to a scalar_int_mode instead of a machine_mode.
11684 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
11685 (get_best_mode): Return a boolean and use a pointer argument to store
11686 the selected mode. Replace the limit mode parameter with a bit limit.
11687 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
11688 for the values returned by bit_field_mode_iterator::next_mode.
11689 (store_bit_field): Update call to get_best_mode.
11690 (store_fixed_bit_field): Likewise.
11691 (extract_fixed_bit_field): Likewise.
11692 * expr.c (optimize_bitfield_assignment_op): Likewise.
11693 * fold-const.c (optimize_bit_field_compare): Likewise.
11694 (fold_truth_andor_1): Likewise.
11695 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
11696 Update for new type of m_mode.
11697 (get_best_mode): As above.
11699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11700 Alan Hayward <alan.hayward@arm.com>
11701 David Sherwood <david.sherwood@arm.com>
11703 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
11704 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
11705 (store_bit_field): Check is_a <scalar_int_mode> before calling
11706 strict_volatile_bitfield_p.
11707 (extract_bit_field): Likewise.
11709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11710 Alan Hayward <alan.hayward@arm.com>
11711 David Sherwood <david.sherwood@arm.com>
11713 * target.def (cstore_mode): Return a scalar_int_mode.
11714 * doc/tm.texi: Regenerate.
11715 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
11716 * targhooks.h (default_cstore_mode): Likewise.
11717 * targhooks.c (default_cstore_mode): Likewise, using a forced
11719 * expmed.c (emit_cstore): Expect the target of the cstore to be
11722 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11723 Alan Hayward <alan.hayward@arm.com>
11724 David Sherwood <david.sherwood@arm.com>
11726 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
11728 (niter_desc): Likewise mode.
11729 (iv_analyze): Add a mode parameter.
11731 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
11732 and change its type to scalar_int_mode.
11733 * loop-iv.c: Update commentary at head of file.
11734 (iv_constant): Pass the mode paraeter before the rtx it describes
11735 and change its type to scalar_int_mode. Remove VOIDmode handling.
11736 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
11737 (iv_extend): Likewise.
11738 (shorten_into_mode): Likewise.
11739 (iv_add): Use scalar_int_mode.
11740 (iv_mult): Likewise.
11741 (iv_shift): Likewise.
11742 (canonicalize_iv_subregs): Likewise.
11743 (get_biv_step_1): Pass the outer_mode parameter before the rtx
11744 it describes and change its mode to scalar_int_mode. Also change
11745 the type of the returned inner_mode to scalar_int_mode.
11746 (get_biv_step): Likewise, turning outer_mode from a pointer
11747 into a direct parameter. Update call to get_biv_step_1.
11748 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
11749 iv_constant and get_biv_step.
11750 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
11751 and change its type to scalar_int_mode. Don't initialise iv->mode
11752 to VOIDmode and remove later checks for its still being VOIDmode.
11753 Update calls to iv_analyze_op and iv_analyze_expr. Check
11754 is_a <scalar_int_mode> when changing the mode under consideration.
11755 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
11756 Update call to iv_analyze_expr.
11757 (iv_analyze_op): Add a mode parameter. Reject subregs whose
11758 inner register is not also a scalar_int_mode. Update call to
11760 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
11761 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
11762 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
11763 separate mode class checks. Update calls to iv_analyze. Remove
11764 fix-up of VOIDmodes after iv_analyze_biv.
11765 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
11766 don't have a scalar_int_mode. Update call to biv_p.
11768 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11769 Alan Hayward <alan.hayward@arm.com>
11770 David Sherwood <david.sherwood@arm.com>
11772 * cfgexpand.c (convert_debug_memory_address): Use
11773 as_a <scalar_int_mode>.
11774 * combine.c (expand_compound_operation): Likewise.
11775 (make_extraction): Likewise.
11776 (change_zero_ext): Likewise.
11777 (simplify_comparison): Likewise.
11778 * cse.c (cse_insn): Likewise.
11779 * dwarf2out.c (minmax_loc_descriptor): Likewise.
11780 (mem_loc_descriptor): Likewise.
11781 (loc_descriptor): Likewise.
11782 * expmed.c (init_expmed_one_mode): Likewise.
11783 (synth_mult): Likewise.
11784 (emit_store_flag_1): Likewise.
11785 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
11786 of a comparison with size.
11787 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
11788 (reduce_to_bit_field_precision): Likewise.
11789 * function.c (expand_function_end): Likewise.
11790 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
11791 * loop-doloop.c (doloop_modify): Likewise.
11792 * optabs.c (expand_binop): Likewise.
11793 (expand_unop): Likewise.
11794 (expand_copysign_absneg): Likewise.
11795 (prepare_cmp_insn): Likewise.
11796 (maybe_legitimize_operand): Likewise.
11797 * recog.c (const_scalar_int_operand): Likewise.
11798 * rtlanal.c (get_address_mode): Likewise.
11799 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11800 (simplify_cond_clz_ctz): Likewise.
11801 * tree-nested.c (get_nl_goto_field): Likewise.
11802 * tree.c (build_vector_type_for_mode): Likewise.
11803 * var-tracking.c (use_narrower_mode): Likewise.
11805 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11806 Alan Hayward <alan.hayward@arm.com>
11807 David Sherwood <david.sherwood@arm.com>
11809 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
11810 * builtins.c (expand_builtin_signbit): Use it.
11811 * cfgexpand.c (expand_debug_expr): Likewise.
11812 * dojump.c (do_jump): Likewise.
11813 (do_compare_and_jump): Likewise.
11814 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
11815 * expmed.c (make_tree): Likewise.
11816 * expr.c (expand_expr_real_2): Likewise.
11817 (expand_expr_real_1): Likewise.
11818 (try_casesi): Likewise.
11819 * fold-const-call.c (fold_const_call_ss): Likewise.
11820 * fold-const.c (unextend): Likewise.
11821 (extract_muldiv_1): Likewise.
11822 (fold_single_bit_test): Likewise.
11823 (native_encode_int): Likewise.
11824 (native_encode_string): Likewise.
11825 (native_interpret_int): Likewise.
11826 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
11827 * internal-fn.c (expand_addsub_overflow): Likewise.
11828 (expand_neg_overflow): Likewise.
11829 (expand_mul_overflow): Likewise.
11830 (expand_arith_overflow): Likewise.
11831 * match.pd: Likewise.
11832 * stor-layout.c (layout_type): Likewise.
11833 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11834 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11835 * tree-ssanames.c (get_range_info): Likewise.
11836 * tree-switch-conversion.c (array_value_type) Likewise.
11837 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11838 (vect_recog_divmod_pattern): Likewise.
11839 (vect_recog_mixed_size_cond_pattern): Likewise.
11840 * tree-vrp.c (extract_range_basic): Likewise.
11841 (simplify_float_conversion_using_ranges): Likewise.
11842 * tree.c (int_fits_type_p): Likewise.
11843 * ubsan.c (instrument_bool_enum_load): Likewise.
11844 * varasm.c (mergeable_string_section): Likewise.
11845 (narrowing_initializer_constant_valid_p): Likewise.
11846 (output_constant): Likewise.
11848 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11849 Alan Hayward <alan.hayward@arm.com>
11850 David Sherwood <david.sherwood@arm.com>
11852 * machmode.h (NARROWEST_INT_MODE): New macro.
11853 * expr.c (alignment_for_piecewise_move): Use it instead of
11854 GET_CLASS_NARROWEST_MODE (MODE_INT).
11855 (push_block): Likewise.
11856 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11858 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11860 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11861 Alan Hayward <alan.hayward@arm.com>
11862 David Sherwood <david.sherwood@arm.com>
11864 * postreload.c (move2add_valid_value_p): Change the type of the
11865 mode parameter to scalar_int_mode.
11866 (move2add_use_add2_insn): Add a mode parameter and use it instead
11868 (move2add_use_add3_insn): Likewise.
11869 (reload_cse_move2add): Update accordingly.
11871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11872 Alan Hayward <alan.hayward@arm.com>
11873 David Sherwood <david.sherwood@arm.com>
11875 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11877 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11878 * optabs.c (expand_unop): Likewise.
11880 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11881 Alan Hayward <alan.hayward@arm.com>
11882 David Sherwood <david.sherwood@arm.com>
11884 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11885 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11886 (popcount_loc_descriptor): Likewise.
11887 (bswap_loc_descriptor): Likewise.
11888 (rotate_loc_descriptor): Likewise.
11889 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
11890 calling the functions above.
11892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11893 Alan Hayward <alan.hayward@arm.com>
11894 David Sherwood <david.sherwood@arm.com>
11896 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
11898 (try_combine): Likewise.
11899 (simplify_if_then_else): Likewise.
11900 * cse.c (cse_insn): Likewise.
11901 * dwarf2out.c (mem_loc_descriptor): Likewise.
11902 * emit-rtl.c (gen_lowpart_common): Likewise.
11903 * simplify-rtx.c (simplify_truncation): Likewise.
11904 (simplify_binary_operation_1): Likewise.
11905 (simplify_const_relational_operation): Likewise.
11906 (simplify_ternary_operation): Likewise.
11907 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11910 Alan Hayward <alan.hayward@arm.com>
11911 David Sherwood <david.sherwood@arm.com>
11913 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11914 * reload.c (push_reload): Likewise.
11915 (find_reloads): Likewise.
11917 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11918 Alan Hayward <alan.hayward@arm.com>
11919 David Sherwood <david.sherwood@arm.com>
11921 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11922 (make_compound_operation_int): Likewise.
11923 (change_zero_ext): Likewise.
11924 * expr.c (convert_move): Likewise.
11925 (convert_modes): Likewise.
11926 * fwprop.c (forward_propagate_subreg): Likewise.
11927 * loop-iv.c (get_biv_step_1): Likewise.
11928 * optabs.c (widen_operand): Likewise.
11929 * postreload.c (move2add_valid_value_p): Likewise.
11930 * recog.c (simplify_while_replacing): Likewise.
11931 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11932 (simplify_binary_operation_1): Likewise. Remove redundant
11933 mode equality check.
11935 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11936 Alan Hayward <alan.hayward@arm.com>
11937 David Sherwood <david.sherwood@arm.com>
11939 * combine.c (combine_simplify_rtx): Add checks for
11940 is_a <scalar_int_mode>.
11941 (simplify_if_then_else): Likewise.
11942 (make_field_assignment): Likewise.
11943 (simplify_comparison): Likewise.
11944 * ifcvt.c (noce_try_bitop): Likewise.
11945 * loop-invariant.c (canonicalize_address_mult): Likewise.
11946 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11948 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11949 Alan Hayward <alan.hayward@arm.com>
11950 David Sherwood <david.sherwood@arm.com>
11952 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
11953 is_a <scalar_int_mode> instead of != BLKmode.
11955 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11956 Alan Hayward <alan.hayward@arm.com>
11957 David Sherwood <david.sherwood@arm.com>
11959 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
11960 instead of != VOIDmode.
11961 * combine.c (if_then_else_cond): Likewise.
11962 (change_zero_ext): Likewise.
11963 * dwarf2out.c (mem_loc_descriptor): Likewise.
11964 (loc_descriptor): Likewise.
11965 * rtlanal.c (canonicalize_condition): Likewise.
11966 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
11968 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11969 Alan Hayward <alan.hayward@arm.com>
11970 David Sherwood <david.sherwood@arm.com>
11972 * simplify-rtx.c (simplify_binary_operation_1): Use
11973 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
11975 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11976 Alan Hayward <alan.hayward@arm.com>
11977 David Sherwood <david.sherwood@arm.com>
11979 * wide-int.h (int_traits<unsigned char>) New class.
11980 (int_traits<unsigned short>) Likewise.
11981 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11982 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11984 * combine.c (set_nonzero_bits_and_sign_copies): Use
11985 is_a <scalar_int_mode>.
11986 (find_split_point): Likewise.
11987 (combine_simplify_rtx): Likewise.
11988 (simplify_logical): Likewise.
11989 (expand_compound_operation): Likewise.
11990 (expand_field_assignment): Likewise.
11991 (make_compound_operation): Likewise.
11992 (extended_count): Likewise.
11993 (change_zero_ext): Likewise.
11994 (simplify_comparison): Likewise.
11995 * dwarf2out.c (scompare_loc_descriptor): Likewise.
11996 (ucompare_loc_descriptor): Likewise.
11997 (minmax_loc_descriptor): Likewise.
11998 (mem_loc_descriptor): Likewise.
11999 (loc_descriptor): Likewise.
12000 * expmed.c (init_expmed_one_mode): Likewise.
12001 * lra-constraints.c (lra_constraint_offset): Likewise.
12002 * optabs.c (prepare_libcall_arg): Likewise.
12003 * postreload.c (move2add_note_store): Likewise.
12004 * reload.c (operands_match_p): Likewise.
12005 * rtl.h (load_extend_op): Likewise.
12006 * rtlhooks.c (gen_lowpart_general): Likewise.
12007 * simplify-rtx.c (simplify_truncation): Likewise.
12008 (simplify_unary_operation_1): Likewise.
12009 (simplify_binary_operation_1): Likewise.
12010 (simplify_const_binary_operation): Likewise.
12011 (simplify_const_relational_operation): Likewise.
12012 (simplify_subreg): Likewise.
12013 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12014 * var-tracking.c (adjust_mems): Likewise.
12015 (prepare_call_arguments): Likewise.
12017 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12018 Alan Hayward <alan.hayward@arm.com>
12019 David Sherwood <david.sherwood@arm.com>
12021 * machmode.h (is_int_mode): New fuction.
12022 * combine.c (find_split_point): Use it.
12023 (combine_simplify_rtx): Likewise.
12024 (simplify_if_then_else): Likewise.
12025 (simplify_set): Likewise.
12026 (simplify_shift_const_1): Likewise.
12027 (simplify_comparison): Likewise.
12028 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12029 * cse.c (notreg_cost): Likewise.
12030 (cse_insn): Likewise.
12031 * cselib.c (cselib_lookup_1): Likewise.
12032 * dojump.c (do_jump_1): Likewise.
12033 (do_compare_rtx_and_jump): Likewise.
12034 * dse.c (get_call_args): Likewise.
12035 * dwarf2out.c (rtl_for_decl_init): Likewise.
12036 (native_encode_initializer): Likewise.
12037 * expmed.c (emit_store_flag_1): Likewise.
12038 (emit_store_flag): Likewise.
12039 * expr.c (convert_modes): Likewise.
12040 (store_field): Likewise.
12041 (expand_expr_real_1): Likewise.
12042 * fold-const.c (fold_read_from_constant_string): Likewise.
12043 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12044 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12045 * optabs.c (expand_binop): Likewise.
12046 (expand_unop): Likewise.
12047 (expand_abs_nojump): Likewise.
12048 (expand_one_cmpl_abs_nojump): Likewise.
12049 * simplify-rtx.c (mode_signbit_p): Likewise.
12050 (val_signbit_p): Likewise.
12051 (val_signbit_known_set_p): Likewise.
12052 (val_signbit_known_clear_p): Likewise.
12053 (simplify_relational_operation_1): Likewise.
12054 * tree.c (vector_type_mode): Likewise.
12056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12057 Alan Hayward <alan.hayward@arm.com>
12058 David Sherwood <david.sherwood@arm.com>
12060 * machmode.h (smallest_mode_for_size): Fix formatting.
12061 (smallest_int_mode_for_size): New function.
12062 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12063 instead of smallest_mode_for_size.
12064 * combine.c (make_extraction): Likewise.
12065 * config/arc/arc.c (arc_expand_movmem): Likewise.
12066 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12067 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12068 * config/s390/s390.c (s390_expand_insv): Likewise.
12069 * config/sparc/sparc.c (assign_int_registers): Likewise.
12070 * config/spu/spu.c (spu_function_value): Likewise.
12071 (spu_function_arg): Likewise.
12072 * coverage.c (get_gcov_type): Likewise.
12073 (get_gcov_unsigned_t): Likewise.
12074 * dse.c (find_shift_sequence): Likewise.
12075 * expmed.c (store_bit_field_1): Likewise.
12076 * expr.c (convert_move): Likewise.
12077 (store_field): Likewise.
12078 * internal-fn.c (expand_arith_overflow): Likewise.
12079 * optabs-query.c (get_best_extraction_insn): Likewise.
12080 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12081 * stor-layout.c (layout_type): Likewise.
12082 (initialize_sizetypes): Likewise.
12083 * targhooks.c (default_get_mask_mode): Likewise.
12084 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12087 Alan Hayward <alan.hayward@arm.com>
12088 David Sherwood <david.sherwood@arm.com>
12090 * machmode.h (opt_mode::else_blk): New function.
12091 (int_mode_for_mode): Declare.
12092 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12093 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12095 * cfgexpand.c (expand_debug_expr): Likewise.
12096 * combine.c (gen_lowpart_or_truncate): Likewise.
12097 (gen_lowpart_for_combine): Likewise.
12098 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12099 * config/avr/avr.c (avr_to_int_mode): Likewise.
12100 (avr_out_plus_1): Likewise.
12101 (avr_out_plus): Likewise.
12102 (avr_out_round): Likewise.
12103 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12104 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12105 (s390_expand_vcond): Likewise.
12106 * config/spu/spu.c (spu_split_immediate): Likewise.
12107 (spu_expand_mov): Likewise.
12108 * dse.c (get_stored_val): Likewise.
12109 * expmed.c (store_bit_field_1): Likewise.
12110 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12112 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12113 (extract_low_bits): Likewise.
12114 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12115 handling rather than repeating the check.
12116 (emit_group_store): Likewise.
12117 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12118 * optabs.c (expand_absneg_bit): Likewise.
12119 (expand_copysign_absneg): Likewise.
12120 (expand_copysign_bit): Likewise.
12121 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12122 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12123 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12124 * var-tracking.c (prepare_call_arguments): Likewise.
12125 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12126 int_mode_for_mode instead of mode_for_size.
12127 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12129 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12130 Alan Hayward <alan.hayward@arm.com>
12131 David Sherwood <david.sherwood@arm.com>
12133 * machmode.h (int_mode_for_size): New function.
12134 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12135 instead of mode_for_size.
12136 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12138 * combine.c (expand_field_assignment): Use int_mode_for_size
12139 instead of mode_for_size.
12140 (make_extraction): Likewise.
12141 (simplify_shift_const_1): Likewise.
12142 (simplify_comparison): Likewise.
12143 * dojump.c (do_jump): Likewise.
12144 * dwarf2out.c (mem_loc_descriptor): Likewise.
12145 * emit-rtl.c (init_derived_machine_modes): Likewise.
12146 * expmed.c (flip_storage_order): Likewise.
12147 (convert_extracted_bit_field): Likewise.
12148 * expr.c (copy_blkmode_from_reg): Likewise.
12149 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12150 * internal-fn.c (expand_mul_overflow): Likewise.
12151 * lower-subreg.c (simple_move): Likewise.
12152 * optabs-libfuncs.c (init_optabs): Likewise.
12153 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12154 * tree.c (vector_type_mode): Likewise.
12155 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12156 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12157 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12158 * tree-vect-stmts.c (vectorizable_load): Likewise.
12159 (vectorizable_store): Likewise.
12161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12162 Alan Hayward <alan.hayward@arm.com>
12163 David Sherwood <david.sherwood@arm.com>
12165 * coretypes.h (pod_mode): New type.
12166 (scalar_int_mode_pod): New typedef.
12167 * machmode.h (pod_mode): New class.
12168 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12169 * genmodes.c (emit_mode_int_n): Update accordingly.
12170 * lower-subreg.h (target_lower_subreg): Change type to
12171 scalar_int_mode_pod.
12172 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12173 scalar_int_mode_pod.
12175 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12176 Alan Hayward <alan.hayward@arm.com>
12177 David Sherwood <david.sherwood@arm.com>
12179 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12180 machine_mode to scalar_int_mode.
12181 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12182 (rs6000_option_override_internal): Remove cast to int.
12183 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12184 machine_mode to scalar_int_mode.
12185 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12186 (rs6000_option_override_internal): Remove cast to int.
12187 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12188 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12190 * config/s390/s390.c (s390_expand_builtin): Likewise.
12191 * coretypes.h (scalar_int_mode): New type.
12192 (opt_scalar_int_mode): New typedef.
12193 * machmode.h (scalar_int_mode): New class.
12194 (scalar_int_mode::includes_p): New function.
12195 (byte_mode): Change type to scalar_int_mode.
12196 (word_mode): Likewise.
12197 (ptr_mode): Likewise.
12198 * emit-rtl.c (byte_mode): Likewise.
12199 (word_mode): Likewise.
12200 (ptr_mode): Likewise.
12201 (init_derived_machine_modes): Update accordingly.
12202 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12203 and MODE_PARTIAL_INT.
12204 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12205 opt_scalar_int_mode.
12207 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12208 Alan Hayward <alan.hayward@arm.com>
12209 David Sherwood <david.sherwood@arm.com>
12211 * target.def (libgcc_floating_mode_supported_p): Take a
12213 * doc/tm.texi: Regenerate.
12214 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12216 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12217 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12220 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12221 Alan Hayward <alan.hayward@arm.com>
12222 David Sherwood <david.sherwood@arm.com>
12224 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12225 * doc/tm.texi: Regenerate.
12226 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12227 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12228 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12229 * targhooks.h (default_floatn_mode): Likewise.
12230 * targhooks.c (default_floatn_mode): Likewise.
12231 * tree.c (build_common_tree_nodes): Update accordingly.
12233 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12234 Alan Hayward <alan.hayward@arm.com>
12235 David Sherwood <david.sherwood@arm.com>
12237 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12238 (mode_iterator::iterate_p): Likewise.
12239 (mode_iterator::get_wider): Likewise.
12240 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12243 Alan Hayward <alan.hayward@arm.com>
12244 David Sherwood <david.sherwood@arm.com>
12246 * coretypes.h (opt_scalar_float_mode): New typedef.
12247 * machmode.h (float_mode_for_size): New function.
12248 * emit-rtl.c (double_mode): Delete.
12249 (init_emit_once): Use float_mode_for_size.
12250 * stor-layout.c (layout_type): Likewise.
12251 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12254 Alan Hayward <alan.hayward@arm.com>
12255 David Sherwood <david.sherwood@arm.com>
12257 * output.h (assemble_real): Take a scalar_float_mode.
12258 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12259 * config/arm/arm.md (consttable_4): Likewise.
12260 (consttable_8): Likewise.
12261 (consttable_16): Likewise.
12262 * config/mips/mips.md (consttable_float): Likewise.
12263 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12264 * varasm.c (assemble_real): Take a scalar_float_mode.
12265 (output_constant_pool_2): Update accordingly.
12266 (output_constant): Likewise.
12268 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12269 Alan Hayward <alan.hayward@arm.com>
12270 David Sherwood <david.sherwood@arm.com>
12272 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12273 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12274 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12275 (native_encode_real): Likewise.
12276 (native_interpret_real): Likewise.
12277 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12278 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12280 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12281 Alan Hayward <alan.hayward@arm.com>
12282 David Sherwood <david.sherwood@arm.com>
12284 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12285 <scalar_float_mode>. Simplify.
12286 (gen_extend_conv_libfunc): Likewise.
12288 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12289 Alan Hayward <alan.hayward@arm.com>
12290 David Sherwood <david.sherwood@arm.com>
12292 * coretypes.h (scalar_float_mode): New type.
12293 * machmode.h (mode_traits::from_int): Use machine_mode if
12294 USE_ENUM_MODES is defined.
12295 (is_a): New function.
12297 (dyn_cast): Likewise.
12298 (scalar_float_mode): New class.
12299 (scalar_float_mode::includes_p): New function.
12300 (is_float_mode): Likewise.
12301 * gdbhooks.py (MachineModePrinter): New class.
12302 (build_pretty_printer): Use it for scalar_float_mode.
12303 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12304 (format_helper::format_helper): Turn into a template.
12305 * genmodes.c (get_mode_class): New function.
12306 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12307 or machine_mode if none.
12308 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12309 as_a <scalar_float_mode>.
12310 * dwarf2out.c (mem_loc_descriptor): Likewise.
12311 (insert_float): Likewise.
12312 (add_const_value_attribute): Likewise.
12313 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12314 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12315 (expand_unop): Update accordingly.
12316 (expand_abs_nojump): Likewise.
12317 (expand_copysign_absneg): Take a scalar_float_mode.
12318 (expand_copysign_bit): Likewise.
12319 (expand_copysign): Update accordingly.
12321 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12322 Alan Hayward <alan.hayward@arm.com>
12323 David Sherwood <david.sherwood@arm.com>
12325 * coretypes.h (opt_mode): New class.
12326 * machmode.h (opt_mode): Likewise.
12327 (opt_mode::else_void): New function.
12328 (opt_mode::require): Likewise.
12329 (opt_mode::exists): Likewise.
12330 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12331 (GET_MODE_2XWIDER_MODE): Likewise.
12332 (mode_iterator::get_wider): Update accordingly.
12333 (mode_iterator::get_2xwider): Likewise.
12334 (mode_iterator::get_known_wider): Likewise, turning into a template.
12335 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12336 forcing a wider mode to exist.
12337 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12338 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12339 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12340 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12341 * lower-subreg.c (init_lower_subreg): Likewise.
12342 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12343 on the final iteration.
12344 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12345 a wider mode exists before asking for a move pattern.
12346 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12347 forcing a wider mode to exist.
12348 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12349 returning false if no such mode exists.
12350 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12351 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12352 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12353 Avoid checking for a MODE_INT if we already know the mode is not a
12355 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12356 forcing a wider mode to exist.
12357 (expmed_mult_highpart_optab): Likewise.
12358 (expmed_mult_highpart): Likewise.
12359 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12361 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12362 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12363 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12364 * internal-fn.c (expand_mul_overflow): Update use of
12365 GET_MODE_2XWIDER_MODE.
12366 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12367 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12368 GET_MODE_WIDER_MODE.
12369 (convert_plusminus_to_widen): Likewise.
12370 * tree-switch-conversion.c (array_value_type): Likewise.
12371 * var-tracking.c (emit_note_insn_var_location): Likewise.
12372 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12373 Return false inside rather than outside the loop if no wider mode
12375 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12376 and GET_MODE_2XWIDER_MODE
12377 (can_compare_p): Use else_void.
12378 * gdbhooks.py (OptMachineModePrinter): New class.
12379 (build_pretty_printer): Use it for opt_mode.
12381 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12382 Alan Hayward <alan.hayward@arm.com>
12383 David Sherwood <david.sherwood@arm.com>
12385 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
12386 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
12388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12389 Alan Hayward <alan.hayward@arm.com>
12390 David Sherwood <david.sherwood@arm.com>
12392 * machmode.h (mode_traits): New structure.
12393 (get_narrowest_mode): New function.
12394 (mode_iterator::start): Likewise.
12395 (mode_iterator::iterate_p): Likewise.
12396 (mode_iterator::get_wider): Likewise.
12397 (mode_iterator::get_known_wider): Likewise.
12398 (mode_iterator::get_2xwider): Likewise.
12399 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
12400 (FOR_EACH_MODE): Likewise.
12401 (FOR_EACH_MODE_FROM): Likewise.
12402 (FOR_EACH_MODE_UNTIL): Likewise.
12403 (FOR_EACH_WIDER_MODE): Likewise.
12404 (FOR_EACH_2XWIDER_MODE): Likewise.
12405 * builtins.c (expand_builtin_strlen): Use new mode iterators.
12406 * combine.c (simplify_comparison): Likewise
12407 * config/i386/i386.c (type_natural_mode): Likewise.
12408 * cse.c (cse_insn): Likewise.
12409 * dse.c (find_shift_sequence): Likewise.
12410 * emit-rtl.c (init_derived_machine_modes): Likewise.
12411 (init_emit_once): Likewise.
12412 * explow.c (hard_function_value): Likewise.
12413 * expmed.c (extract_fixed_bit_field_1): Likewise.
12414 (extract_bit_field_1): Likewise.
12415 (expand_divmod): Likewise.
12416 (emit_store_flag_1): Likewise.
12417 * expr.c (init_expr_target): Likewise.
12418 (convert_move): Likewise.
12419 (alignment_for_piecewise_move): Likewise.
12420 (widest_int_mode_for_size): Likewise.
12421 (emit_block_move_via_movmem): Likewise.
12422 (copy_blkmode_to_reg): Likewise.
12423 (set_storage_via_setmem): Likewise.
12424 (compress_float_constant): Likewise.
12425 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12426 * optabs-query.c (get_best_extraction_insn): Likewise.
12427 * optabs.c (expand_binop): Likewise.
12428 (expand_twoval_unop): Likewise.
12429 (expand_twoval_binop): Likewise.
12430 (widen_leading): Likewise.
12431 (widen_bswap): Likewise.
12432 (expand_parity): Likewise.
12433 (expand_unop): Likewise.
12434 (prepare_cmp_insn): Likewise.
12435 (prepare_float_lib_cmp): Likewise.
12436 (expand_float): Likewise.
12437 (expand_fix): Likewise.
12438 (expand_sfix_optab): Likewise.
12439 * postreload.c (move2add_use_add2_insn): Likewise.
12440 * reg-stack.c (reg_to_stack): Likewise.
12441 * reginfo.c (choose_hard_reg_mode): Likewise.
12442 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12443 * stor-layout.c (mode_for_size): Likewise.
12444 (smallest_mode_for_size): Likewise.
12445 (mode_for_vector): Likewise.
12446 (finish_bitfield_representative): Likewise.
12447 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
12448 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
12449 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12450 * var-tracking.c (prepare_call_arguments): Likewise.
12452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12453 Alan Hayward <alan.hayward@arm.com>
12454 David Sherwood <david.sherwood@arm.com>
12456 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
12457 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
12458 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
12459 * machmode.h (mode_size): Move earlier in file.
12460 (mode_precision): Likewise.
12461 (mode_inner): Likewise.
12462 (mode_nunits): Likewise.
12463 (mode_unit_size): Likewise.
12464 (unit_unit_precision): Likewise.
12465 (mode_wider): Likewise.
12466 (mode_2xwider): Likewise.
12467 (machine_mode): New class.
12468 (mode_to_bytes): New function.
12469 (mode_to_bits): Likewise.
12470 (mode_to_precision): Likewise.
12471 (mode_to_inner): Likewise.
12472 (mode_to_unit_size): Likewise.
12473 (mode_to_unit_precision): Likewise.
12474 (mode_to_nunits): Likewise.
12475 (GET_MODE_SIZE): Use mode_to_bytes.
12476 (GET_MODE_BITSIZE): Use mode_to_bits.
12477 (GET_MODE_PRECISION): Use mode_to_precision.
12478 (GET_MODE_INNER): Use mode_to_inner.
12479 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
12480 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
12481 (GET_MODE_NUNITS): Use mode_to_nunits.
12482 * system.h (ALWAYS_INLINE): New macro.
12483 * config/powerpcspe/powerpcspe-c.c
12484 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
12485 int for arg1_mode and arg2_mode.
12487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12488 Alan Hayward <alan.hayward@arm.com>
12489 David Sherwood <david.sherwood@arm.com>
12491 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12492 Prefix mode names with E_ in case statements.
12493 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12494 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
12495 (aarch64_split_simd_move): Likewise.
12496 (aarch64_gen_storewb_pair): Likewise.
12497 (aarch64_gen_loadwb_pair): Likewise.
12498 (aarch64_gen_store_pair): Likewise.
12499 (aarch64_gen_load_pair): Likewise.
12500 (aarch64_get_condition_code_1): Likewise.
12501 (aarch64_constant_pool_reload_icode): Likewise.
12502 (get_rsqrte_type): Likewise.
12503 (get_rsqrts_type): Likewise.
12504 (get_recpe_type): Likewise.
12505 (get_recps_type): Likewise.
12506 (aarch64_gimplify_va_arg_expr): Likewise.
12507 (aarch64_simd_container_mode): Likewise.
12508 (aarch64_emit_load_exclusive): Likewise.
12509 (aarch64_emit_store_exclusive): Likewise.
12510 (aarch64_expand_compare_and_swap): Likewise.
12511 (aarch64_gen_atomic_cas): Likewise.
12512 (aarch64_emit_bic): Likewise.
12513 (aarch64_emit_atomic_swap): Likewise.
12514 (aarch64_emit_atomic_load_op): Likewise.
12515 (aarch64_evpc_trn): Likewise.
12516 (aarch64_evpc_uzp): Likewise.
12517 (aarch64_evpc_zip): Likewise.
12518 (aarch64_evpc_ext): Likewise.
12519 (aarch64_evpc_rev): Likewise.
12520 (aarch64_evpc_dup): Likewise.
12521 (aarch64_gen_ccmp_first): Likewise.
12522 (aarch64_gen_ccmp_next): Likewise.
12523 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12524 (alpha_emit_xfloating_libcall): Likewise.
12525 (emit_insxl): Likewise.
12526 (alpha_arg_type): Likewise.
12527 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
12528 (arc_preferred_simd_mode): Likewise.
12529 (arc_secondary_reload): Likewise.
12530 (get_arc_condition_code): Likewise.
12531 (arc_print_operand): Likewise.
12532 (arc_legitimate_constant_p): Likewise.
12533 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12534 * config/arc/arc.md (casesi_load): Likewise.
12535 (casesi_compact_jump): Likewise.
12536 * config/arc/predicates.md (proper_comparison_operator): Likewise.
12537 (cc_use_register): Likewise.
12538 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12539 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12540 (arm_init_iwmmxt_builtins): Likewise.
12541 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
12542 (neon_expand_vector_init): Likewise.
12543 (arm_attr_length_move_neon): Likewise.
12544 (maybe_get_arm_condition_code): Likewise.
12545 (arm_emit_vector_const): Likewise.
12546 (arm_preferred_simd_mode): Likewise.
12547 (arm_output_iwmmxt_tinsr): Likewise.
12548 (thumb1_output_casesi): Likewise.
12549 (thumb2_output_casesi): Likewise.
12550 (arm_emit_load_exclusive): Likewise.
12551 (arm_emit_store_exclusive): Likewise.
12552 (arm_expand_compare_and_swap): Likewise.
12553 (arm_evpc_neon_vuzp): Likewise.
12554 (arm_evpc_neon_vzip): Likewise.
12555 (arm_evpc_neon_vrev): Likewise.
12556 (arm_evpc_neon_vtrn): Likewise.
12557 (arm_evpc_neon_vext): Likewise.
12558 (arm_validize_comparison): Likewise.
12559 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
12560 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
12561 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
12562 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
12563 (c6x_preferred_simd_mode): Likewise.
12564 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
12565 (epiphany_rtx_costs): Likewise.
12566 * config/epiphany/predicates.md (proper_comparison_operator):
12568 * config/frv/frv.c (condexec_memory_operand): Likewise.
12569 (frv_emit_move): Likewise.
12570 (output_move_single): Likewise.
12571 (output_condmove_single): Likewise.
12572 (frv_hard_regno_mode_ok): Likewise.
12573 (frv_matching_accg_mode): Likewise.
12574 * config/h8300/h8300.c (split_adds_subs): Likewise.
12575 (h8300_rtx_costs): Likewise.
12576 (h8300_print_operand): Likewise.
12577 (compute_mov_length): Likewise.
12578 (output_logical_op): Likewise.
12579 (compute_logical_op_length): Likewise.
12580 (compute_logical_op_cc): Likewise.
12581 (h8300_shift_needs_scratch_p): Likewise.
12582 (output_a_shift): Likewise.
12583 (compute_a_shift_length): Likewise.
12584 (compute_a_shift_cc): Likewise.
12585 (expand_a_rotate): Likewise.
12586 (output_a_rotate): Likewise.
12587 * config/i386/i386.c (classify_argument): Likewise.
12588 (function_arg_advance_32): Likewise.
12589 (function_arg_32): Likewise.
12590 (function_arg_64): Likewise.
12591 (function_value_64): Likewise.
12592 (ix86_gimplify_va_arg): Likewise.
12593 (ix86_legitimate_constant_p): Likewise.
12594 (put_condition_code): Likewise.
12595 (split_double_mode): Likewise.
12596 (ix86_avx256_split_vector_move_misalign): Likewise.
12597 (ix86_expand_vector_logical_operator): Likewise.
12598 (ix86_split_idivmod): Likewise.
12599 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
12600 (ix86_build_const_vector): Likewise.
12601 (ix86_build_signbit_mask): Likewise.
12602 (ix86_match_ccmode): Likewise.
12603 (ix86_cc_modes_compatible): Likewise.
12604 (ix86_expand_branch): Likewise.
12605 (ix86_expand_sse_cmp): Likewise.
12606 (ix86_expand_sse_movcc): Likewise.
12607 (ix86_expand_int_sse_cmp): Likewise.
12608 (ix86_expand_vec_perm_vpermi2): Likewise.
12609 (ix86_expand_vec_perm): Likewise.
12610 (ix86_expand_sse_unpack): Likewise.
12611 (ix86_expand_int_addcc): Likewise.
12612 (ix86_split_to_parts): Likewise.
12613 (ix86_vectorize_builtin_gather): Likewise.
12614 (ix86_vectorize_builtin_scatter): Likewise.
12615 (avx_vpermilp_parallel): Likewise.
12616 (inline_memory_move_cost): Likewise.
12617 (ix86_tieable_integer_mode_p): Likewise.
12618 (x86_maybe_negate_const_int): Likewise.
12619 (ix86_expand_vector_init_duplicate): Likewise.
12620 (ix86_expand_vector_init_one_nonzero): Likewise.
12621 (ix86_expand_vector_init_one_var): Likewise.
12622 (ix86_expand_vector_init_concat): Likewise.
12623 (ix86_expand_vector_init_interleave): Likewise.
12624 (ix86_expand_vector_init_general): Likewise.
12625 (ix86_expand_vector_set): Likewise.
12626 (ix86_expand_vector_extract): Likewise.
12627 (emit_reduc_half): Likewise.
12628 (ix86_emit_i387_round): Likewise.
12629 (ix86_mangle_type): Likewise.
12630 (ix86_expand_round_sse4): Likewise.
12631 (expand_vec_perm_blend): Likewise.
12632 (canonicalize_vector_int_perm): Likewise.
12633 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
12634 (expand_vec_perm_1): Likewise.
12635 (expand_vec_perm_interleave3): Likewise.
12636 (expand_vec_perm_even_odd_pack): Likewise.
12637 (expand_vec_perm_even_odd_1): Likewise.
12638 (expand_vec_perm_broadcast_1): Likewise.
12639 (ix86_vectorize_vec_perm_const_ok): Likewise.
12640 (ix86_expand_vecop_qihi): Likewise.
12641 (ix86_expand_mul_widen_hilo): Likewise.
12642 (ix86_expand_sse2_abs): Likewise.
12643 (ix86_expand_pextr): Likewise.
12644 (ix86_expand_pinsr): Likewise.
12645 (ix86_preferred_simd_mode): Likewise.
12646 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
12647 * config/i386/sse.md (*andnot<mode>3): Likewise.
12648 (<mask_codefor><code><mode>3<mask_name>): Likewise.
12649 (*<code><mode>3): Likewise.
12650 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
12651 (ia64_expand_atomic_op): Likewise.
12652 (ia64_arg_type): Likewise.
12653 (ia64_mode_to_int): Likewise.
12654 (ia64_scalar_mode_supported_p): Likewise.
12655 (ia64_vector_mode_supported_p): Likewise.
12656 (expand_vec_perm_broadcast): Likewise.
12657 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
12658 (iq2000_function_arg_advance): Likewise.
12659 (iq2000_function_arg): Likewise.
12660 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
12661 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
12662 (m68k_libcall_value): Likewise.
12663 (m68k_function_value): Likewise.
12664 (sched_attr_op_type): Likewise.
12665 * config/mcore/mcore.c (mcore_output_move): Likewise.
12666 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
12668 (microblaze_function_arg): Likewise.
12669 * config/mips/mips.c (mips16_build_call_stub): Likewise.
12670 (mips_print_operand): Likewise.
12671 (mips_mode_ok_for_mov_fmt_p): Likewise.
12672 (mips_vector_mode_supported_p): Likewise.
12673 (mips_preferred_simd_mode): Likewise.
12674 (mips_expand_vpc_loongson_even_odd): Likewise.
12675 (mips_expand_vec_unpack): Likewise.
12676 (mips_expand_vi_broadcast): Likewise.
12677 (mips_expand_vector_init): Likewise.
12678 (mips_expand_vec_reduc): Likewise.
12679 (mips_expand_msa_cmp): Likewise.
12680 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
12681 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
12682 (cc_flags_for_mode): Likewise.
12683 * config/msp430/msp430.c (msp430_print_operand): Likewise.
12684 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
12685 (nds32_output_casesi_pc_relative): Likewise.
12686 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12687 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
12688 (nvptx_gen_unpack): Likewise.
12689 (nvptx_gen_pack): Likewise.
12690 (nvptx_gen_shuffle): Likewise.
12691 (nvptx_gen_wcast): Likewise.
12692 (nvptx_preferred_simd_mode): Likewise.
12693 * config/pa/pa.c (pa_secondary_reload): Likewise.
12694 * config/pa/predicates.md (base14_operand): Likewise.
12695 * config/powerpcspe/powerpcspe-c.c
12696 (altivec_resolve_overloaded_builtin): Likewise.
12697 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
12699 (rs6000_preferred_simd_mode): Likewise.
12700 (output_vec_const_move): Likewise.
12701 (rs6000_expand_vector_extract): Likewise.
12702 (rs6000_split_vec_extract_var): Likewise.
12703 (reg_offset_addressing_ok_p): Likewise.
12704 (rs6000_legitimate_offset_address_p): Likewise.
12705 (rs6000_legitimize_address): Likewise.
12706 (rs6000_emit_set_const): Likewise.
12707 (rs6000_const_vec): Likewise.
12708 (rs6000_emit_move): Likewise.
12709 (spe_build_register_parallel): Likewise.
12710 (rs6000_darwin64_record_arg_recurse): Likewise.
12711 (swap_selector_for_mode): Likewise.
12712 (spe_init_builtins): Likewise.
12713 (paired_init_builtins): Likewise.
12714 (altivec_init_builtins): Likewise.
12715 (do_load_for_compare): Likewise.
12716 (rs6000_generate_compare): Likewise.
12717 (rs6000_expand_float128_convert): Likewise.
12718 (emit_load_locked): Likewise.
12719 (emit_store_conditional): Likewise.
12720 (rs6000_output_function_epilogue): Likewise.
12721 (rs6000_handle_altivec_attribute): Likewise.
12722 (rs6000_function_value): Likewise.
12723 (emit_fusion_gpr_load): Likewise.
12724 (emit_fusion_p9_load): Likewise.
12725 (emit_fusion_p9_store): Likewise.
12726 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
12727 (fusion_gpr_mem_load): Likewise.
12728 (fusion_addis_mem_combo_load): Likewise.
12729 (fusion_addis_mem_combo_store): Likewise.
12730 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
12731 (fusion_gpr_mem_load): Likewise.
12732 (fusion_addis_mem_combo_load): Likewise.
12733 (fusion_addis_mem_combo_store): Likewise.
12734 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12736 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
12737 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
12738 (rs6000_preferred_simd_mode): Likewise.
12739 (output_vec_const_move): Likewise.
12740 (rs6000_expand_vector_extract): Likewise.
12741 (rs6000_split_vec_extract_var): Likewise.
12742 (reg_offset_addressing_ok_p): Likewise.
12743 (rs6000_legitimate_offset_address_p): Likewise.
12744 (rs6000_legitimize_address): Likewise.
12745 (rs6000_emit_set_const): Likewise.
12746 (rs6000_const_vec): Likewise.
12747 (rs6000_emit_move): Likewise.
12748 (rs6000_darwin64_record_arg_recurse): Likewise.
12749 (swap_selector_for_mode): Likewise.
12750 (paired_init_builtins): Likewise.
12751 (altivec_init_builtins): Likewise.
12752 (rs6000_expand_float128_convert): Likewise.
12753 (emit_load_locked): Likewise.
12754 (emit_store_conditional): Likewise.
12755 (rs6000_output_function_epilogue): Likewise.
12756 (rs6000_handle_altivec_attribute): Likewise.
12757 (rs6000_function_value): Likewise.
12758 (emit_fusion_gpr_load): Likewise.
12759 (emit_fusion_p9_load): Likewise.
12760 (emit_fusion_p9_store): Likewise.
12761 * config/rx/rx.c (rx_gen_move_template): Likewise.
12762 (flags_from_mode): Likewise.
12763 * config/s390/predicates.md (s390_alc_comparison): Likewise.
12764 (s390_slb_comparison): Likewise.
12765 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
12766 (s390_vector_mode_supported_p): Likewise.
12767 (s390_cc_modes_compatible): Likewise.
12768 (s390_match_ccmode_set): Likewise.
12769 (s390_canonicalize_comparison): Likewise.
12770 (s390_emit_compare_and_swap): Likewise.
12771 (s390_branch_condition_mask): Likewise.
12772 (s390_rtx_costs): Likewise.
12773 (s390_secondary_reload): Likewise.
12774 (__SECONDARY_RELOAD_CASE): Likewise.
12775 (s390_expand_cs): Likewise.
12776 (s390_preferred_simd_mode): Likewise.
12777 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
12778 * config/sh/sh.c (sh_print_operand): Likewise.
12779 (dump_table): Likewise.
12780 (sh_secondary_reload): Likewise.
12781 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12782 * config/sh/sh.md (casesi_worker_1): Likewise.
12783 (casesi_worker_2): Likewise.
12784 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
12785 (fcc_comparison_operator): Likewise.
12786 * config/sparc/sparc.c (sparc_expand_move): Likewise.
12787 (emit_soft_tfmode_cvt): Likewise.
12788 (sparc_preferred_simd_mode): Likewise.
12789 (output_cbranch): Likewise.
12790 (sparc_print_operand): Likewise.
12791 (sparc_expand_vec_perm_bmask): Likewise.
12792 (vector_init_bshuffle): Likewise.
12793 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
12794 (spu_vector_mode_supported_p): Likewise.
12795 (spu_expand_insv): Likewise.
12796 (spu_emit_branch_or_set): Likewise.
12797 (spu_handle_vector_attribute): Likewise.
12798 (spu_builtin_splats): Likewise.
12799 (spu_builtin_extract): Likewise.
12800 (spu_builtin_promote): Likewise.
12801 (spu_expand_sign_extend): Likewise.
12802 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
12803 (tilegx_simd_int): Likewise.
12804 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
12805 (tilepro_simd_int): Likewise.
12806 * config/v850/v850.c (const_double_split): Likewise.
12807 (v850_print_operand): Likewise.
12808 (ep_memory_offset): Likewise.
12809 * config/vax/vax.c (vax_rtx_costs): Likewise.
12810 (vax_output_int_move): Likewise.
12811 (vax_output_int_add): Likewise.
12812 (vax_output_int_subtract): Likewise.
12813 * config/visium/predicates.md (visium_branch_operator): Likewise.
12814 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
12815 (visium_print_operand_address): Likewise.
12816 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12817 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
12818 (xtensa_expand_conditional_branch): Likewise.
12819 (xtensa_copy_incoming_a7): Likewise.
12820 (xtensa_output_literal): Likewise.
12821 * dfp.c (decimal_real_maxval): Likewise.
12822 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12824 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12825 Alan Hayward <alan.hayward@arm.com>
12826 David Sherwood <david.sherwood@arm.com>
12828 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
12829 (mode_nunits_inline): Likewise.
12830 (mode_inner_inline): Likewise.
12831 (mode_unit_size_inline): Likewise.
12832 (mode_unit_precision_inline): Likewise.
12833 (emit_insn_modes_h): Likewise. Also emit a #define of the
12835 (emit_mode_wider): Add an E_ prefix to mode names.
12836 (emit_mode_complex): Likewise.
12837 (emit_mode_inner): Likewise.
12838 (emit_mode_adjustments): Likewise.
12839 (emit_mode_int_n): Likewise.
12840 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12841 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12842 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12843 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12844 (CRC32_BUILTIN, ENTRY): Likewise.
12845 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12846 (aarch64_pop_regs): Likewise.
12847 (aarch64_process_components): Likewise.
12848 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12849 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12850 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12851 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12852 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12853 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12855 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12856 (pcmpistr): Likewise.
12857 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12858 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12859 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12861 * config/rl78/rl78.c (mduc_regs): Likewise.
12862 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12863 (htm_expand_builtin): Likewise.
12864 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12865 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12866 * config/xtensa/xtensa.c (print_operand): Likewise.
12867 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12868 (NUM_MODE_VECTOR_INT): Likewise.
12869 * genoutput.c (null_operand): Likewise.
12870 (output_operand_data): Likewise.
12871 * genrecog.c (print_parameter_value): Likewise.
12872 * lra.c (debug_operand_data): Likewise.
12874 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12876 * dwarf2out.c (scompare_loc_descriptor_wide)
12877 (scompare_loc_descriptor_narrow): New functions, split out from...
12878 (scompare_loc_descriptor): ...here.
12879 * expmed.c (emit_store_flag_int): New function, split out from...
12880 (emit_store_flag): ...here.
12882 2017-08-30 Richard Biener <rguenther@suse.de>
12884 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12885 (dwarf2out_early_finish): Move setting of AT_pubnames from
12886 early debug output to early finish.
12888 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12890 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
12891 and -mdata-region to the assembler.
12893 2017-08-30 Richard Biener <rguenther@suse.de>
12895 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
12897 (gen_subprogram_die): Add DW_AT_object_pointer only early.
12898 (dwarf2out_early_global_decl): Only generate a DIE for the
12899 abstract origin if it doesn't already exist or is a declaration DIE.
12900 (resolve_addr): Do not add the linkage name twice when
12901 generating a stub DIE for the DW_TAG_GNU_call_site target.
12903 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12905 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12906 Use machine_mode rather than int for arg1_mode.
12908 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12911 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12912 that the second argument of the built-in functions to unpack
12913 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12914 switch statement instead a lot of if statements.
12915 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12916 Allow 64-bit values to be in Altivec registers as well as
12917 traditional floating point registers.
12918 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12920 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12922 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12923 MAX_REGS_PER_ADDRESS == 1.
12925 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12927 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12928 * config/i386/i386.c (ix86_option_override_internal): Simplify
12929 setting of opts->x_flag_entry.
12931 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12932 Jakub Jelinek <jakub@redhat.com>
12933 Richard Biener <rguenther@suse.de>
12935 PR tree-optimization/81503
12936 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
12937 folded constant fits in the target type; reorder tests for clarity.
12939 2017-08-29 Martin Liska <mliska@suse.cz>
12941 * passes.def: Include pass_lower_switch.
12942 * stmt.c (dump_case_nodes): Remove and move to
12943 tree-switch-conversion.
12944 (case_values_threshold): Likewise.
12945 (expand_switch_as_decision_tree_p): Likewise.
12946 (emit_case_decision_tree): Likewise.
12947 (expand_case): Likewise.
12948 (balance_case_nodes): Likewise.
12949 (node_has_low_bound): Likewise.
12950 (node_has_high_bound): Likewise.
12951 (node_is_bounded): Likewise.
12952 (emit_case_nodes): Likewise.
12953 (struct simple_case_node): New struct.
12954 (add_case_node): Remove.
12955 (emit_case_dispatch_table): Use vector instead of case_list.
12956 (reset_out_edges_aux): Remove.
12957 (compute_cases_per_edge): Likewise.
12958 (expand_case): Build list of simple_case_node.
12959 (expand_sjlj_dispatch_table): Use it.
12960 * tree-switch-conversion.c (struct case_node): Moved from
12961 stmt.c and adjusted.
12962 (emit_case_nodes): Likewise.
12963 (node_has_low_bound): Likewise.
12964 (node_has_high_bound): Likewise.
12965 (node_is_bounded): Likewise.
12966 (case_values_threshold): Likewise.
12967 (reset_out_edges_aux): Likewise.
12968 (compute_cases_per_edge): Likewise.
12969 (add_case_node): Likewise.
12970 (dump_case_nodes): Likewise.
12971 (balance_case_nodes): Likewise.
12972 (expand_switch_as_decision_tree_p): Likewise.
12973 (emit_jump): Likewise.
12974 (emit_case_decision_tree): Likewise.
12975 (try_switch_expansion): Likewise.
12976 (do_jump_if_equal): Likewise.
12977 (emit_cmp_and_jump_insns): Likewise.
12978 (fix_phi_operands_for_edge): New function.
12979 (record_phi_operand_mapping): Likewise.
12980 (class pass_lower_switch): New pass.
12981 (pass_lower_switch::execute): New function.
12982 (make_pass_lower_switch): Likewise.
12983 (conditional_probability):
12984 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12985 * tree-pass.h: Add make_pass_lower_switch.
12987 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12990 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
12993 2017-08-29 Richard Biener <rguenther@suse.de>
12995 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
12996 we do not add a DW_AT_inline attribute twice.
12997 (gen_subprogram_die): Remove code setting DW_AT_inline on
12998 DECL_ABSTRACT_P nodes.
13000 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13002 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13003 calls to internal functions.
13004 (gimplify_modify_expr): Likewise.
13005 * tree-call-cdce.c (use_internal_fn): Likewise.
13006 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13007 (convert_to_divmod): Set the nothrow flag.
13008 * tree-if-conv.c (predicate_mem_writes): Likewise.
13009 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13010 (vectorizable_call): Likewise.
13011 (vectorizable_store): Likewise.
13012 (vectorizable_load): Likewise.
13013 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13014 (vect_recog_mask_conversion_pattern): Likewise.
13016 2017-08-29 Martin Liska <mliska@suse.cz>
13019 * gcc.c (driver_handle_option): Add new argument.
13020 * opts-common.c (handle_option): Pass
13021 target_option_override_hook.
13022 * opts-global.c (lang_handle_option): Add new option.
13023 (set_default_handlers): Add new argument.
13024 (decode_options): Likewise.
13025 * opts.c (target_handle_option): Likewise.
13026 (common_handle_option): Call target_option_override_hook.
13027 * opts.h (struct cl_option_handler_func): Add hook for
13028 target option override.
13029 (struct cl_option_handlers): Likewise.
13030 (set_default_handlers): Add new argument.
13031 (decode_options): Likewise.
13032 (common_handle_option): Likewise.
13033 (target_handle_option): Likewise.
13034 * toplev.c (toplev::main): Pass targetm.target_option.override
13037 2017-08-29 Richard Biener <rguenther@suse.de>
13038 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13040 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13041 life to the active subtree.
13043 2017-08-28 Jeff Law <law@redhat.com>
13045 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13046 derive_equivalences.
13047 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13048 Code moved into....
13049 (edge_info::derive_equivalences): New private member function
13051 * tree-ssa-dom.c (class edge_info): Changed from a struct
13052 to a class. Add ctor/dtor, methods and data members.
13053 (edge_info::edge_info): Renamed from allocate_edge_info.
13054 Initialize additional members.
13055 (edge_info::~edge_info): New.
13056 (free_dom_edge_info): Delete the edge info.
13057 (record_edge_info): Use new class & associated member functions.
13058 Tighten forms for testing for edge equivalences.
13059 (record_temporary_equivalences): Iterate over the simple
13060 equivalences rather than assuming there's only one per edge.
13061 (cprop_into_successor_phis): Iterate over the simple
13062 equivalences rather than assuming there's only one per edge.
13063 (optimize_stmt): Use operand_equal_p rather than pointer
13064 equality for mini-DSE code.
13066 2017-08-28 Nathan Sidwell <nathan@acm.org>
13068 * gcc.c (execute): Fold SIGPIPE handling into switch
13069 statement. Adjust internal error message.
13071 2017-08-28 Richard Biener <rguenther@suse.de>
13074 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13075 Do nothing for removed DIEs.
13077 2017-08-28 Richard Biener <rguenther@suse.de>
13079 PR tree-optimization/81977
13080 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13083 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13086 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13087 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13088 using targetm.gen_mem_thread_fence.
13090 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13093 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13094 predicate to register_operand. Reorder operands.
13095 (*btr<mode>): Ditto.
13096 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13097 (*btr<mode>_mask): Ditto.
13099 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13101 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13102 * config/rs6000/xmmintrin.h: New file.
13103 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13105 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13108 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13109 parameter and_insn and return it.
13110 (recombine_lvx_pattern): Insert a copy to ensure availability of
13111 the base register of the copied masking operation at the point of
13112 the instruction replacement.
13113 (recombine_stvx_pattern): Likewise.
13115 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13117 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13118 undocumented switches.
13119 (-mpower9-dform-vector): Likewise.
13120 (-mpower9-dform): Likewise.
13121 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13122 comments to delete references to -mpower9-dform* switches.
13123 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13124 Delete reference to -mpower9-dform* switches, test for
13125 -mpower9-vector instead.
13126 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13127 (OTHER_P9_VECTOR_MASKS): Likewise.
13128 (POWERPC_MASKS): Likewise.
13129 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13130 tests against -mpower9-dform* to -mpower9-vector. Delete code
13131 that checked for -mpower9-dform* consistancy with other options.
13132 Add test for -mpower9-misc to enable other power9 switches.
13133 (rs6000_init_hard_regno_mode_ok): Likewise.
13134 (rs6000_option_override_internal): Likewise.
13135 (rs6000_emit_prologue): Likewise.
13136 (rs6000_emit_epilogue): Likewise.
13137 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13138 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13139 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13141 (emit_fusion_p9_store): Likewise.
13142 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13143 resetting these macros if the assembler does not support ISA 3.0
13145 (TARGET_P9_DFORM_VECTOR): Likewise.
13146 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13147 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13149 2017-08-25 Alan Modra <amodra@gmail.com>
13151 PR rtl-optimization/81747
13152 * cse.c (cse_extended_basic_block): Don't attempt to record
13153 equivalences for degenerate conditional branches that branch
13154 to their fall-through.
13156 2017-08-24 Martin Sebor <msebor@redhat.com>
13158 PR middle-end/81908
13159 * gimple-fold.c (size_must_be_zero_p): New function.
13160 (gimple_fold_builtin_memory_op): Call it.
13162 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13164 * config/rs6000/mm_malloc.h: New file.
13166 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13168 PR tree-optimization/81913
13169 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13170 analysis when either IVs in condition can wrap.
13172 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13174 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13175 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13177 2017-08-24 Richard Biener <rguenther@suse.de>
13180 * targhooks.c (default_target_can_inline_p): Properly
13181 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13182 is present and always compare.
13183 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13184 infer -mfpmath=sse from TARGET_SSE_P.
13185 (ix86_can_inline_p): Properly use target_option_default_node when
13186 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13188 2017-08-24 Richard Biener <rguenther@suse.de>
13191 * dwarf2out.c (output_die): Handle flag_generate_offload like
13193 (output_comp_unit): Likewise.
13194 (gen_array_type_die): Likewise.
13195 (dwarf2out_early_finish): Likewise.
13196 (note_variable_value_in_expr): Likewise.
13197 (dwarf2out_finish): Likewise. Adjust assert.
13198 * cgraphunit.c (symbol_table::compile): Move setting of
13199 flag_generate_offload earlier ...
13200 (symbol_table::finalize_compilation_unit): ... here, before
13201 early debug finalization.
13203 2017-08-24 Richard Biener <rguenther@suse.de>
13205 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13206 and ipa-fnsummary.h.
13207 (ix86_can_inline_p): When ix86_fpmath flags do not match
13208 check whether the callee uses FP math at all.
13210 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13212 PR middle-end/81931
13213 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13214 instead of TYPE_PRECISION.
13216 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13217 Alan Hayward <alan.hayward@arm.com>
13218 David Sherwood <david.sherwood@arm.com>
13220 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13222 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13224 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13226 (ix86_frame::stack_realign_allocate): New field.
13227 (struct machine_frame_state): Modify comments.
13228 (machine_frame_state::sp_realigned_fp_end): New field.
13229 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13230 layout calculation.
13231 (sp_valid_at): Add assertion to assure no attempt to access invalid
13232 offset of a realigned stack.
13233 (fp_valid_at): Likewise.
13234 (choose_baseaddr): Modify comments.
13235 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13236 ix86_expand_prologue.
13237 (ix86_expand_prologue): Modify stack realignment and allocation.
13238 (ix86_expand_epilogue): Modify comments.
13239 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13240 avx2_runtime, avx512f, and avx512f_runtime.
13242 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13244 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13245 (mstackrealign): Do not init to -1.
13246 * config/i386/i386.c (ix86_option_override_internal):
13247 Check opts_set, not opts when setting default value of
13248 opts->x_ix86_force_align_arg_pointer.
13250 2017-08-23 Richard Biener <rguenther@suse.de>
13252 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13253 lang_hooks.decl_printable_name.
13254 * print-rtl-function.c (print_rtx_function): Likewise.
13255 * tree-pretty-print.c (dump_function_header): Likewise.
13257 2017-08-23 Richard Biener <rguenther@suse.de>
13260 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13261 -g0 at compile-time.
13263 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13265 PR middle-end/19706
13266 * doc/sourcebuild.texi (Other hardware attributes):
13269 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13271 PR middle-end/19706
13272 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13273 Fix single-use check.
13275 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13277 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13279 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13281 * doc/install.texi: Modify to add more details on running selected
13284 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13286 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13287 is combined with -mabi=ms.
13288 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13291 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13293 PR tree-optimization/81488
13294 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13295 and cached_basis fields.
13296 (MAX_SPREAD): New constant.
13297 (alloc_cand_and_find_basis): Initialize new fields.
13298 (clear_visited): New function.
13299 (create_phi_basis_1): Rename from create_phi_basis, set visited
13300 and cached_basis fields.
13301 (create_phi_basis): New wrapper function.
13302 (phi_add_costs_1): Rename from phi_add_costs, add spread
13303 parameter, set visited field, short-circuit when limits reached.
13304 (phi_add_costs): New wrapper function.
13305 (record_phi_increments_1): Rename from record_phi_increments, set
13307 (record_phi_increments): New wrapper function.
13308 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13309 (phi_incr_cost): New wrapper function.
13310 (all_phi_incrs_profitable_1): Rename from
13311 all_phi_incrs_profitable, set visited field.
13312 (all_phi_incrs_profitable): New wrapper function.
13314 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13315 Alan Hayward <alan.hayward@arm.com>
13316 David Sherwood <david.sherwood@arm.com>
13318 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13319 that takes the outer and inner modes.
13320 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13321 comparison as the canonical test for a paradoxical subreg.
13322 * combine.c (simplify_set): Use paradoxical_subreg_p.
13323 (make_extraction): Likewise.
13324 (force_to_mode): Likewise.
13325 (rtx_equal_for_field_assignment_p): Likewise.
13326 (gen_lowpart_for_combine): Likewise.
13327 (simplify_comparison): Likewise.
13328 * cse.c (equiv_constant): Likewise.
13329 * expmed.c (store_bit_field_1): Likewise.
13330 * final.c (alter_subreg): Likewise.
13331 * fwprop.c (propagate_rtx): Likewise.
13332 (forward_propagate_subreg): Likewise.
13333 * ira-conflicts.c (ira_build_conflicts): Likewise.
13334 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13335 * lra-constraints.c (curr_insn_transform): Likewise.
13336 (split_reg): Likewise.
13337 * lra-eliminations.c (move_plus_up): Likewise.
13338 (lra_eliminate_regs_1): Likewise.
13339 * recog.c (general_operand): Likewise.
13340 * ree.c (combine_reaching_defs): Likewise.
13341 * reload.c (push_reload): Likewise.
13342 (find_reloads): Likewise.
13343 * reload1.c (elimination_effects): Likewise.
13344 (compute_reload_subreg_offset): Likewise.
13345 (choose_reload_regs): Likewise.
13346 * rtlanal.c (subreg_lsb_1): Likewise.
13347 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13348 (simplify_subreg): Likewise.
13349 * var-tracking.c (track_loc_p): Likewise.
13350 * emit-rtl.c (byte_lowpart_offset): Likewise.
13351 (paradoxical_subreg_p): Delete out-of-line definition.
13353 2017-08-22 Jeff Law <law@redhat.com>
13355 PR tree-optimization/81741
13356 PR tree-optimization/71947
13357 * tree-ssa-dom.c: Include tree-inline.h.
13358 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13359 equivalences if one is more expensive to compute than the other.
13360 * tree-ssa-scopedtables.h (class const_or_copies): Make
13361 record_const_or_copy_raw method private.
13362 (class avail_exprs_stack): New method simplify_binary_operation.
13363 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13364 avail_exprs_stack::simplify_binary_operation as needed.
13365 (avail_exprs_stack::simplify_binary_operation): New function.
13367 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13369 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13370 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13371 (DOT_SYMBOLS): Likewise.
13372 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13373 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13374 (RS6000_ABI_NAME): Likewise.
13375 (TARGET_CMODEL): Likewise.
13376 (TOC_SECTION_ASM_OP): Likewise.
13377 (SET_CMODEL): New macro.
13378 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13380 2017-08-22 Richard Biener <rguenther@suse.de>
13382 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13383 to allow for free-lang-data replacements similar to verify_type_variant.
13385 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13386 Alan Hayward <alan.hayward@arm.com>
13387 David Sherwood <david.sherwood@arm.com>
13389 * config/aarch64/aarch64.md (casesi): Use DImode rather than
13390 VOIDmode for the LABEL_REF.
13392 2017-08-22 Richard Biener <rguenther@suse.de>
13394 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
13396 2017-08-22 Richard Biener <rguenther@suse.de>
13398 * common.opt (feliminate-dwarf2-dups): Ignore.
13399 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
13400 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
13401 same_die_p_wrap, compute_section_prefix,
13402 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
13403 (comdat_symbol_id, comdat_symbol_number): Likewise.
13404 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
13406 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
13407 (output_die): Mark unreachable path unreachable.
13408 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
13409 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
13410 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
13411 (dwarf2out_early_finish): Likewise.
13413 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
13415 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
13417 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
13420 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
13421 not VAR_P. Filter attribute warnings with OPT_Wattributes.
13422 (avr_attribute_table) <io, io_low, address>: Initialize
13423 .decl_required with true.
13425 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
13427 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
13428 undocumented debugging options.
13429 (-mvsx-scalar-double): Likewise.
13430 (-mallow-df-permute): Likewise.
13431 (-mvectorize-builtins): Likewise.
13432 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13433 (rs6000_builtin_vectorized_function): Likewise.
13434 (rs6000_builtin_md_vectorized_function): Likewise.
13435 (rs6000_opt_vars): Likewise.
13437 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
13440 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
13441 (*btrq_imm): Rename from *btrq.
13442 (*btcq_imm): Rename from *btcq.
13443 (btsc): New code attribute.
13444 (*<btsc><mode>): New insn pattern.
13445 (*btr<mode>): Ditto.
13446 (*<btsc><mode>_mask): New insn_and_split pattern.
13447 (*btr<mode>_mask): Ditto.
13449 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13450 Alan Hayward <alan.hayward@arm.com>
13451 David Sherwood <david.sherwood@arm.com>
13453 * function.c (pad_below): Simplify padding calculation.
13455 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13456 Alan Hayward <alan.hayward@arm.com>
13457 David Sherwood <david.sherwood@arm.com>
13459 * target.def (function_prologue): Remove frame size argument.
13460 (function_epilogue): Likewise.
13461 * doc/tm.texi: Regenerate.
13462 * final.c (final_start_function): Update call to function_prologue.
13463 (final_end_function): Update call to function_epilogue.
13464 (default_function_pro_epilogue): Remove frame size argument.
13465 * output.h (default_function_pro_epilogue): Likewise.
13466 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
13467 (arm_output_function_prologue): Likewise.
13468 * config/frv/frv.c (frv_function_prologue): Likewise.
13469 (frv_function_epilogue): Likewise.
13470 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
13471 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
13472 (ia64_output_function_epilogue): Likewise.
13473 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
13474 (m32r_output_function_epilogue): Likewise.
13475 * config/microblaze/microblaze.c (microblaze_function_prologue)
13476 (microblaze_function_epilogue): Likewise.
13477 * config/mips/mips.c (mips_output_function_prologue): Likewise.
13478 (mips_output_function_epilogue): Likewise.
13479 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
13480 (mmix_target_asm_function_epilogue): Likewise.
13481 * config/msp430/msp430.c (msp430_start_function): Likewise.
13482 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
13483 (nds32_asm_function_epilogue): Likewise.
13484 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
13485 * config/pa/pa.c (pa_output_function_prologue): Likewise.
13486 (pa_output_function_epilogue): Likewise.
13487 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
13488 (rs6000_output_function_epilogue): Likewise.
13489 * config/rl78/rl78.c (rl78_start_function): Likewise.
13490 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
13491 (rs6000_output_function_epilogue): Likewise.
13492 * config/rx/rx.c (rx_output_function_prologue): Likewise.
13493 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
13494 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
13495 (sparc_asm_function_epilogue): Likewise.
13497 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13499 * tree.h (type_has_mode_precision_p): New function.
13500 * convert.c (convert_to_integer_1): Use it.
13501 * expr.c (expand_expr_real_2): Likewise.
13502 (expand_expr_real_1): Likewise.
13503 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
13504 * match.pd: Likewise.
13505 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
13506 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
13507 * tree-tailcall.c (process_assignment): Likewise.
13508 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13509 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
13510 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
13511 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13512 (vectorizable_assignment): Likewise.
13513 (vectorizable_shift): Likewise.
13514 (vectorizable_operation): Likewise.
13515 * tree-vrp.c (register_edge_assert_for_2): Likewise.
13517 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
13519 * match.pd: Add pow (C, x) simplification.
13521 2017-08-21 Richard Biener <rguenther@suse.de>
13523 PR tree-optimization/81900
13524 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
13525 for blocks with abnormal predecessors.
13526 (compute_antic): Do not set visited flag prematurely.
13528 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
13531 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
13533 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13535 * stor-layout.h (vector_type_mode): Move to...
13536 * tree.h (vector_type_mode): ...here.
13537 * stor-layout.c (vector_type_mode): Move to...
13538 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
13540 2017-08-21 Richard Biener <rguenther@suse.de>
13542 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
13543 register_external_die hooks.
13544 (debug_false_tree_charstarstar_uhwistar): Declare.
13545 (debug_nothing_tree_charstar_uhwi): Likewise.
13546 * debug.c (do_nothing_debug_hooks): Adjust.
13547 (debug_false_tree_charstarstar_uhwistar): New do nothing.
13548 (debug_nothing_tree_charstar_uhwi): Likewise.
13549 * dbxout.c (dbx_debug_hooks): Adjust.
13550 (xcoff_debug_hooks): Likewise.
13551 * sdbout.c (sdb_debug_hooks): Likewise.
13552 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13553 * dwarf2out.c (macinfo_label_base): New global.
13554 (dwarf2out_register_external_die): New function for the
13555 register_external_die hook.
13556 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
13557 (dwarf2_debug_hooks): Use them.
13558 (dwarf2_lineno_debug_hooks): Adjust.
13559 (struct die_struct): Add with_offset flag.
13560 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
13561 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
13562 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
13563 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
13564 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
13565 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
13566 defining section names for the early LTO debug variants.
13567 (reset_indirect_string): New helper.
13568 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
13569 (print_dw_val): Add support for offsetted symbol references.
13570 (get_ultimate_context): Split out from is_cxx.
13571 (is_cxx): Use get_ultimate_context.
13572 (is_fortran): Add decl overload.
13573 (compute_comp_unit_symbol): Split out worker from
13574 compute_section_prefix.
13575 (compute_section_prefix): Call compute_comp_unit_symbol and
13576 set comdat_type_p here.
13577 (output_die): Skip DIE symbol output for the LTO added one.
13578 Handle DIE symbol references with offset.
13579 (output_comp_unit): Guard section name mangling properly.
13580 For LTO debug sections emit a symbol at the section beginning
13581 which we use to refer to its DIEs.
13582 (add_abstract_origin_attribute): For DIEs registered via
13583 dwarf2out_register_external_die directly refer to the early
13584 DIE rather than indirectly through the shadow one we created.
13585 Remove obsolete call to dwarf2out_abstract_function for
13586 non-function/block origins.
13587 (gen_array_type_die): When generating early LTO debug do
13588 not emit DW_AT_string_length.
13589 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
13590 late when in LTO. As suggested place a gcc_unreachable for
13591 the DECL_ABSTRACT_P case.
13592 (gen_subprogram_die): Avoid another specification DIE
13593 for early built declarations/definitions for the late LTO case.
13594 (gen_variable_die): Add type references for late duplicated VLA dies
13596 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
13597 we have the abstract instance already.
13598 (process_scope_var): Adjust decl DIE contexts in LTO which
13599 first puts them in limbo.
13600 (gen_decl_die): Do not generate type DIEs late apart from
13601 types for VLAs or for decls we do not yet have a DIE. Do not
13602 call dwarf2out_abstract_function late.
13603 (dwarf2out_early_global_decl): Make sure to create DIEs
13604 for abstract instances of a decl first.
13605 (dwarf2out_late_global_decl): Adjust comment.
13606 (output_macinfo_op): With multiple macro sections use
13607 macinfo_label_base to distinguish labels.
13608 (output_macinfo): Likewise. Update macinfo_label_base.
13609 Pass in the line info label.
13610 (note_variable_value_in_expr): When generating LTO resolve
13611 all variable values here by generating DIEs as needed.
13612 (init_sections_and_labels): Add early LTO debug flag parameter
13613 and generate different sections and names if set. Add generation
13614 counter for the labels so we can have multiple of them.
13615 (reset_dies): Helper to allow DIEs to be output multiple times.
13616 (dwarf2out_finish): When outputting DIEs to the fat part of an
13617 LTO object first reset DIEs.
13618 (dwarf2out_early_finish): Output early DIEs when generating LTO.
13619 (modified_type_die): Check for decl_ultimate_origin being self
13621 (gen_type_die_with_usage): Likewise.
13622 (gen_typedef_die): Allow decl_ultimate_origin being self.
13623 (set_decl_abstract_flags): Remove.
13624 (set_block_abstract_flags): Likewise.
13625 (dwarf2out_abstract_function): Treat the early generated DIEs
13626 as the abstract copy and only add DW_AT_inline and
13627 DW_AT_artificial here and call set_decl_origin_self.
13628 If the DIE has an abstract origin don't do anything.
13629 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
13630 if we have none yet (Go fails to build one, PR78628).
13631 (variably_modified_type_p): Prevent endless recursion for Ada
13632 cyclic pointer types.
13633 * lto-streamer-in.c: Include debug.h.
13634 (dref_queue): New global.
13635 (lto_read_tree_1): Stream in DIE references.
13636 (lto_input_tree): Register DIE references.
13637 (input_function): Stream DECL_DEBUG_ARGS.
13638 * lto-streamer-out.c: Include debug.h.
13639 (lto_write_tree_1): Output DIE references.
13640 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
13641 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
13642 (output_function): Stream DECL_DEBUG_ARGS.
13643 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
13644 Stream DECL_ABSTRACT_ORIGIN.
13645 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
13646 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
13647 DECL_CONTEXT for file-scope decls.
13648 * lto-streamer.h (struct dref_entry): Declare.
13649 (dref_queue): Likewise.
13650 * cfgexpand.c (pass_expand::execute): Do not call the
13651 outlining_inline_function hook here.
13652 * lto-wrapper.c (debug_obj): New global.
13653 (tool_cleanup): Unlink it if required.
13654 (debug_objcopy): New function.
13655 (run_gcc): Handle early debug sections in the IL files by
13656 extracting them to separate files, partially linkin them and
13657 feeding the result back as result to the linker.
13658 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
13659 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
13660 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
13661 sections into a separate segment.
13662 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
13664 (darwin_asm_dwarf_section): Likewise.
13665 (darwin_asm_output_dwarf_offset): Likewise.
13666 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
13668 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13669 Alan Hayward <alan.hayward@arm.com>
13670 David Sherwood <david.sherwood@arm.com>
13672 * read-md.h (md_reader::record_potential_iterator_use): Replace
13673 pointer argument with an rtx and an index.
13674 * read-rtl.c (iterator_group::apply_iterator): Likewise.
13675 (apply_mode_iterator): Likewise.
13676 (apply_code_iterator): Likewise.
13677 (apply_int_iterator): Likewise.
13678 (apply_subst_iterator): Likewise.
13679 (record_iterator_use): Likewise.
13680 (record_attribute_use): Likewise.
13681 (md_reader::record_potential_iterator_use): Likewise. Update calls
13682 to record_iterator_use and apply_iterator.
13683 (iterator_use): Replace ptr with x and index.
13684 (attribute_use): Likewise.
13685 (apply_attribute_uses): Update calls to apply_iterator.
13686 (apply_iterators): Likewise. Update initialization of iterator_use.
13687 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
13688 and record_potential_iterator_use.
13689 (rtx_reader::read_rtx_operand): Likewise.
13691 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13692 Alan Hayward <alan.hayward@arm.com>
13693 David Sherwood <david.sherwood@arm.com>
13695 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
13698 2017-08-21 Richard Biener <rguenther@suse.de>
13700 PR middle-end/81884
13701 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
13702 at struct end conservatively when comparing common bases.
13704 2017-08-21 Richard Biener <rguenther@suse.de>
13706 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
13707 (mem_ref_in_stmt): Remove.
13708 (determine_max_movement): Use ref index to get at the reference.
13709 (invariantness_dom_walker::before_dom_children): Deal with
13710 lim data already initialized.
13711 (gather_mem_refs_stmt): Initialize lim data and record ref index.
13713 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13715 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
13716 (TARGET_ISA_ROUND): Ditto.
13717 (TARGET_ROUND): Ditto.
13718 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
13719 * config/i386/i386.md: Ditto.
13720 * config/i386/sse.md: Ditto.
13721 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
13722 with OPTION_MASK_ISA_SSE4_1.
13724 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
13727 * doc/extend.texi (x86 Built-in Functions): Correct the name of
13728 __builtin_ia32_lzcnt_u16.
13730 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
13733 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
13734 (rs6000_set_current_function): Rewrite function to use it.
13736 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
13739 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
13740 and TYPE_WARN_IF_NOT_ALIGN.
13741 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
13742 (handle_warn_if_not_align): New.
13743 (place_union_field): Call handle_warn_if_not_align.
13744 (place_field): Call handle_warn_if_not_align.
13745 Copy TYPE_WARN_IF_NOT_ALIGN.
13746 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
13747 (layout_type): Likewise.
13748 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
13750 (tree_decl_common): Add warn_if_not_align.
13751 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
13752 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
13753 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
13754 (DECL_WARN_IF_NOT_ALIGN): Likewise.
13755 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
13756 * doc/extend.texi: Document warn_if_not_aligned attribute.
13757 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
13759 2017-08-17 Martin Liska <mliska@suse.cz>
13762 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
13763 (get_data_dependence): Use it as pointer type.
13764 (distribute_loop): Likewise.
13766 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13768 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
13769 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
13770 (p8_vmrgow_<mode>_direct): New define_insn.
13771 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
13772 handle endianness for vmrgew and vmrgow permute patterns.
13774 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13776 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
13777 * config/rs6000/rs6000-cpus.def: Remove comment.
13778 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
13779 (POWERPC_MASKS): Likewise.
13780 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
13781 use of TARGET_VSX_TIMODE.
13782 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
13783 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
13784 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
13785 (rs6000_option_override_internal): Remove dead code.
13786 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
13787 (rs6000_legitimize_reload_address): Likewise.
13788 (rs6000_legitimate_address_p): Likewise.
13789 (rs6000_opt_masks): Delete "vsx-timode".
13790 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
13791 from function comment.
13792 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
13793 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
13794 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
13796 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
13797 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
13798 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
13799 (VSX_TI): Likewise.
13801 (define_peephole2): Likewise.
13803 2017-08-17 Martin Sebor <msebor@redhat.com>
13806 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
13807 past the end of an array.
13808 (test_pp_format): Add test cases.
13810 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
13812 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
13813 missing ECF_NOTHROW flags.
13815 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13818 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
13819 operands residing in integer registers.
13820 (*vsx_le_perm_load_<mode>): Likewise.
13821 (*vsx_le_perm_store_<mode>): Likewise.
13822 (define_peephole2): Add peepholes to optimize the above.
13824 2017-08-17 Marek Polacek <polacek@redhat.com>
13826 PR middle-end/81814
13827 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
13828 to mimic what shorten_compare did. Change the return type to bool.
13829 (fold_cond_expr_with_comparison): Update call to
13830 operand_equal_for_comparison_p.
13831 (fold_ternary_loc): Likewise.
13833 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13835 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13837 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13839 2017-08-17 Richard Biener <rguenther@suse.de>
13841 * tree-ssa-structalias.c (solve_graph): When propagating
13842 to successors update the graphs succ edges and avoid duplicate work.
13844 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13847 * config/i386/i386.c (ix86_option_override_internal): Save target
13848 specific options after ix86_stack_protector_guard_reg was changed.
13850 2017-08-17 Richard Biener <rguenther@suse.de>
13852 PR tree-optimization/81827
13853 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13854 (new_var_info): Initialize it conservatively.
13855 (get_call_vi): Mark register vars.
13856 (new_scalar_tmp_constraint_exp): Likewise.
13857 (handle_rhs_call): Likewise.
13858 (handle_const_call): Likewise.
13859 (create_function_info_for): Likewise.
13860 (solve_constraints): Sort varinfos to separate register from
13861 non-register vars to pack points-to solution bitmaps during
13864 2017-08-17 Marek Polacek <polacek@redhat.com>
13866 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13868 2017-08-17 Richard Biener <rguenther@suse.de>
13870 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13871 to true when overflow is undefined and we saturated the result.
13873 2017-08-17 Alan Modra <amodra@gmail.com>
13876 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13877 Don't use store multiple if only one reg needs saving.
13878 (interesting_frame_related_regno): New function.
13879 (rs6000_frame_related): Don't emit frame info for regs that
13881 (rs6000_emit_epilogue): Likewise.
13883 2017-08-16 Nathan Sidwell <nathan@acm.org>
13885 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13886 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13887 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
13888 * tree.c (free_lang_data_in_type): Use else-if chain. Always
13889 clear TYPE_LANG_1. Remove obsolete member-function stripping.
13890 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
13891 (verify_type): Adjust for TYPE_BINFO move.
13892 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
13893 process TYPE_BINFO directly.
13894 (hash_tree): Likewise.
13895 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13897 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13900 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13902 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13904 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13907 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13908 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13909 Add ix86_binary_operator_ok to insn constraint.
13910 (*iordi_1_bts): Ditto.
13911 (*xordi_1_btc): Ditto.
13912 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13913 Update corresponding peephole2 pattern.
13917 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13919 PR tree-optimization/81832
13920 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13921 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13923 2017-08-16 Marek Polacek <polacek@redhat.com>
13926 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13927 regarding size_zero_node.
13929 2017-08-16 Martin Liska <mliska@suse.cz>
13932 * config.gcc: Respect previously set extra_objs in case
13935 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
13937 PR tree-optimization/81835
13938 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
13939 the phi SSA_NAME. Check that the condition in a COND_EXPR does
13940 not depend on the phi.
13942 2017-08-16 Alan Modra <amodra@gmail.com>
13944 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
13947 2017-08-16 Alan Modra <amodra@gmail.com>
13949 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
13950 (save_reg_p): ..into this. Update all callers.
13951 (first_reg_to_save): Simplify.
13953 2017-08-16 Alan Modra <amodra@gmail.com>
13955 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
13958 2017-08-15 Joseph Myers <joseph@codesourcery.com>
13962 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
13963 constant count if that count is less than 32.
13965 2017-08-15 Nathan Sidwell <nathan@acm.org>
13967 * gcc.c (execute): Emit friendlier message if inferior is killed
13968 by an external cause.
13970 2017-08-15 Richard Biener <rguenther@suse.de>
13972 PR tree-optimization/81790
13973 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
13974 CONSTRUCTORs from simplifying and VN.
13976 2017-08-14 Martin Sebor <msebor@redhat.com>
13978 * builtin-attrs.def: Add comments.
13980 2017-08-14 Martin Sebor <msebor@redhat.com>
13983 * doc/extend.texi (attribute nonstring): Document new attribute.
13985 2017-08-14 Martin Sebor <msebor@redhat.com>
13988 * tree-diagnostic.c (default_tree_printer): Handle %G.
13989 * gimple-pretty-print.h (percent_G_format): Declare new function.
13990 * gimple-pretty-print.c (percent_G_format): Define.
13991 * tree-pretty-print.c (percent_K_format): Add argument.
13993 2017-08-14 Martin Sebor <msebor@redhat.com>
13995 PR translation/79998
13996 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
13997 Remove a stray space.
13999 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14002 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14003 (*iordi_1_bts): Ditto.
14004 (*xordi_1_btc): Ditto.
14006 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14009 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14010 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14012 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14013 quoted strings, and make more translator-friendly.
14014 (darwin_rs6000_override_options): Likewise.
14015 (rs6000_option_override_internal): Likewise.
14016 (rs6000_return_in_memory): Fix overlong line.
14017 (init_cmulative_args): Use quoted strings, and make more
14018 translator-friendly.
14019 (rs6000_pass_by_reference): Fix overlong line.
14020 (def_builtin): Use quoted strings.
14021 (altivec_expand_predicate_builtin): Use quoted strings, and make
14022 more translator-friendly.
14023 (htm_expand_builtin): Use quoted strings.
14024 (cpu_expand_builtin): Use quoted strings, and make more
14025 translator-friendly.
14026 (altivec_expand_builtin): Likewise.
14027 (paired_expand_predicate_builtin): Likewise.
14028 (rs6000_invalid_builtin): Likewise.
14029 (builtin_function_type): Use quoted strings.
14030 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14031 more translator-friendly.
14032 (rs6000_trampoline_init): Likewise.
14033 (rs6000_handle_altivec_attribute): Likewise.
14034 (rs6000_inner_target_options): Use quoted strings.
14035 (rs6000_disable_incompatible_switches): Likewise.
14036 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14037 strings, and make more translator-friendly.
14038 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14040 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14042 PR tree-optimization/81799
14043 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14044 cond_expr to simple gimple operand.
14046 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14048 PR middle-end/46932
14049 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14051 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14055 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14056 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14057 TARGET_GASISR_PROLOGUES.
14058 * config/avr/avr.c (avr_option_override): Same.
14059 (avr_pass_pre_proep::execute): Same.
14061 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14064 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14065 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14067 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14069 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14070 define_insn_and_split. Split to a memory load from 0 in
14071 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14072 using PTR mode iterator.
14073 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14074 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14075 (*add_tp_<mode>): Redefine as define_insn_and_split.
14076 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14077 address space. Merge with *add_tp_x32 using PTR mode iterator.
14078 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14079 Split to an add with a memory load from 0 in
14080 DEFAULT_TLS_SEG_REG address space.
14082 2017-08-12 Andrew Pinski <apinski@cavium.com>
14084 * config/aarch64/aarch64-option-extensions.def (rdma):
14085 Fix feature string to what Linux prints out in /proc/cpuinfo.
14087 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14090 * dwarf2out.c (modified_type_die): For C typedef types that have
14091 an ultimate origin, process the ultimate origin instead of the
14093 (gen_typedef_die): Assert that input DECLs have no ultimate
14095 (gen_type_die_with_usage): For typedef variants that have an
14096 ultimate origin, just call gen_decl_die on the original DECL.
14097 (process_scope_var): Avoid creating DIEs for local typedefs and
14098 concrete static variables.
14100 2017-08-12 Alan Modra <amodra@gmail.com>
14104 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14105 match gnu-user.h startfile.
14106 (ENDFILE_LINUX_SPEC): Similarly.
14108 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14111 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14113 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14115 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14116 * config/aarch64/aarch64.md (mov<mode>): Change.
14117 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14118 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14119 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14121 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14123 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14124 for storage order barriers.
14126 2017-08-11 Martin Liska <mliska@suse.cz>
14129 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14130 variables of void type.
14132 2017-08-11 Martin Liska <mliska@suse.cz>
14134 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14135 TARGET_SUPPORTS_ALIASES.
14136 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14137 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14138 (optimize_weakref): Likewise.
14139 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14140 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14141 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14143 2017-08-11 Martin Liska <mliska@suse.cz>
14146 * config/i386/i386.c (make_resolver_func): Do complete
14147 refactoring of the function.
14149 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14152 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14153 * config/i386/i386.c (ix86_stack_protect_guard): Use
14154 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14155 * doc/invoke.texi (x86 Options): Document
14156 -mstack-protector-guard-symbol= option.
14158 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14160 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14161 * config/i386/i386.c (ix86_split_stack_guard): New function.
14162 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14163 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14164 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14165 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14166 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14167 (split_stack_space_check): Call ix86_split_stack_guard.
14169 2017-08-10 Martin Sebor <msebor@redhat.com>
14171 * print-tree.c (print_node): Print location using the established
14173 Replace spaces with colons.
14174 (debug_raw, debug): Ditto.
14176 2017-08-10 Martin Sebor <msebor@redhat.com>
14179 * pretty-print.c (pp_format): Correct the handling of %s precision.
14181 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14184 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14186 (ix86_finalize_stack_frame_flags): This. Also clear
14187 frame_pointer_needed if -fno-omit-frame-pointer is used without
14189 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14190 with ix86_finalize_stack_frame_flags.
14191 (ix86_expand_epilogue): Likewise.
14192 (ix86_expand_split_stack_prologue): Likewise.
14193 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14195 2017-08-10 Martin Liska <mliska@suse.cz>
14198 * c-attribs.c (handle_target_attribute):
14199 Report warning for an empty string argument of target attribute.
14201 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14204 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14206 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14207 or DECL_NONLOCAL labels.
14208 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14209 or DECL_NONLOCAL labels here.
14211 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14213 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14214 to indicate when early gimple folding has been disabled.
14215 (rs6000_gimple_fold_builtin): Add debug content.
14216 (rs6000_invalid_builtin): Fix whitespace.
14217 (rs6000_expand_builtin): Fix whitespace.
14218 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14220 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14223 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14224 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14226 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14228 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14229 (qdf24xx): Likewise.
14230 * config/aarch64/aarch64.md: Include falkor.md.
14231 * config/aarch64/falkor.md: New.
14233 2017-08-09 Marek Polacek <polacek@redhat.com>
14236 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14237 * diagnostic.c (emit_diagnostic): Add a comment.
14238 (emit_diagnostic_valist): New function.
14240 2017-08-09 Marek Polacek <polacek@redhat.com>
14243 * input.c (make_location): New overload.
14244 * input.h (make_location): Declare.
14246 2017-08-08 Alan Modra <amodra@gmail.com>
14247 H.J. Lu <hongjiu.lu@intel.com>
14250 * gcc.c (NO_PIE_SPEC): Delete.
14251 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14253 (LINK_PIE_SPEC): ..to here.
14254 (LINK_COMMAND_SPEC): Support -no-pie.
14255 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14256 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14258 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14259 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14260 (ENDFILE_CRTEND_SPEC): Similarly.
14262 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14265 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14266 (mstack-protector-guard-offset=): Ditto.
14267 * config/i386/i386.c (ix86_option_override): Handle
14268 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14270 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14271 ix86_stack_protect_guard_offset variables.
14272 (TARGET_STACK_PROTECT_GUARD): Always define.
14273 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14274 and -mstack-protector-guard-offset= options.
14276 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14278 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14279 boundary case for the last candidate.
14281 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14283 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14284 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14286 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14288 PR middle-end/19706
14289 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14290 * config/aarch64/aarch64-builtins.c
14291 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14292 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14293 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14295 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14296 Andrew Pinski <pinskia@gmail.com>
14298 PR middle-end/19706
14299 * internal-fn.def (XORSIGN): New.
14300 * optabs.def (xorsign_optab): New.
14301 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14302 (convert_expand_mult_copysign): New.
14303 (pass_optimize_widening_mul::execute): Call
14304 convert_expand_mult_copysign.
14306 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14308 PR tree-optimization/81354
14309 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14310 Insert on edges rather than explicitly creating landing pads.
14311 (analyze_candidates_and_replace): Commit edge inserts.
14313 2017-08-08 Richard Biener <rguenther@suse.de>
14315 PR middle-end/81719
14316 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14317 (expand_simple_operations): Also look through ADDR_EXPRs with
14318 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14320 2017-08-08 Richard Biener <rguenther@suse.de>
14322 PR tree-optimization/81723
14323 * tree-vect-slp.c (struct bst_traits): New hash traits.
14324 (bst_fail): New global.
14325 (vect_build_slp_tree_2): New worker, split out from ...
14326 (vect_build_slp_tree): ... this now wrapping it with using
14327 bst_fail set to cache SLP tree build fails. Properly handle
14329 (vect_analyze_slp_instance): Allocate and free bst_fail.
14331 2017-08-08 Martin Liska <mliska@suse.cz>
14334 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14335 LABEL_DECLs that can be from a different function.
14337 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14339 PR tree-optimization/81744
14340 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14341 loop's number of iterations.
14343 2017-08-08 Martin Liska <mliska@suse.cz>
14345 * asan.c: Include header files.
14346 * attribs.c (build_decl_attribute_variant): New function moved
14348 (build_type_attribute_qual_variant): Likewise.
14349 (cmp_attrib_identifiers): Likewise.
14350 (simple_cst_list_equal): Likewise.
14351 (omp_declare_simd_clauses_equal): Likewise.
14352 (attribute_value_equal): Likewise.
14353 (comp_type_attributes): Likewise.
14354 (build_type_attribute_variant): Likewise.
14355 (lookup_ident_attribute): Likewise.
14356 (remove_attribute): Likewise.
14357 (merge_attributes): Likewise.
14358 (merge_type_attributes): Likewise.
14359 (merge_decl_attributes): Likewise.
14360 (merge_dllimport_decl_attributes): Likewise.
14361 (handle_dll_attribute): Likewise.
14362 (attribute_list_equal): Likewise.
14363 (attribute_list_contained): Likewise.
14364 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14365 (lookup_attribute_by_prefix): Likewise.
14366 * bb-reorder.c: Include header files.
14367 * builtins.c: Likewise.
14368 * calls.c: Likewise.
14369 * cfgexpand.c: Likewise.
14370 * cgraph.c: Likewise.
14371 * cgraphunit.c: Likewise.
14372 * convert.c: Likewise.
14373 * dwarf2out.c: Likewise.
14374 * final.c: Likewise.
14375 * fold-const.c: Likewise.
14376 * function.c: Likewise.
14377 * gimple-expr.c: Likewise.
14378 * gimple-fold.c: Likewise.
14379 * gimple-pretty-print.c: Likewise.
14380 * gimple.c: Likewise.
14381 * gimplify.c: Likewise.
14382 * hsa-common.c: Likewise.
14383 * hsa-gen.c: Likewise.
14384 * internal-fn.c: Likewise.
14385 * ipa-chkp.c: Likewise.
14386 * ipa-cp.c: Likewise.
14387 * ipa-devirt.c: Likewise.
14388 * ipa-fnsummary.c: Likewise.
14389 * ipa-inline.c: Likewise.
14390 * ipa-visibility.c: Likewise.
14392 * lto-cgraph.c: Likewise.
14393 * omp-expand.c: Likewise.
14394 * omp-general.c: Likewise.
14395 * omp-low.c: Likewise.
14396 * omp-offload.c: Likewise.
14397 * omp-simd-clone.c: Likewise.
14398 * opts-global.c: Likewise.
14399 * passes.c: Likewise.
14400 * predict.c: Likewise.
14401 * sancov.c: Likewise.
14402 * sanopt.c: Likewise.
14403 * symtab.c: Likewise.
14404 * toplev.c: Likewise.
14405 * trans-mem.c: Likewise.
14406 * tree-chkp.c: Likewise.
14407 * tree-eh.c: Likewise.
14408 * tree-into-ssa.c: Likewise.
14409 * tree-object-size.c: Likewise.
14410 * tree-parloops.c: Likewise.
14411 * tree-profile.c: Likewise.
14412 * tree-ssa-ccp.c: Likewise.
14413 * tree-ssa-live.c: Likewise.
14414 * tree-ssa-loop.c: Likewise.
14415 * tree-ssa-sccvn.c: Likewise.
14416 * tree-ssa-structalias.c: Likewise.
14417 * tree-ssa.c: Likewise.
14418 * tree-streamer-in.c: Likewise.
14419 * tree-vectorizer.c: Likewise.
14420 * tree-vrp.c: Likewise.
14421 * tsan.c: Likewise.
14422 * ubsan.c: Likewise.
14423 * varasm.c: Likewise.
14424 * varpool.c: Likewise.
14425 * tree.c: Remove functions moved to attribs.[ch].
14426 * tree.h: Likewise.
14427 * config/aarch64/aarch64.c: Add attrs.h header file.
14428 * config/alpha/alpha.c: Likewise.
14429 * config/arc/arc.c: Likewise.
14430 * config/arm/arm.c: Likewise.
14431 * config/avr/avr.c: Likewise.
14432 * config/bfin/bfin.c: Likewise.
14433 * config/c6x/c6x.c: Likewise.
14434 * config/cr16/cr16.c: Likewise.
14435 * config/cris/cris.c: Likewise.
14436 * config/darwin.c: Likewise.
14437 * config/epiphany/epiphany.c: Likewise.
14438 * config/fr30/fr30.c: Likewise.
14439 * config/frv/frv.c: Likewise.
14440 * config/ft32/ft32.c: Likewise.
14441 * config/h8300/h8300.c: Likewise.
14442 * config/i386/winnt.c: Likewise.
14443 * config/ia64/ia64.c: Likewise.
14444 * config/iq2000/iq2000.c: Likewise.
14445 * config/lm32/lm32.c: Likewise.
14446 * config/m32c/m32c.c: Likewise.
14447 * config/m32r/m32r.c: Likewise.
14448 * config/m68k/m68k.c: Likewise.
14449 * config/mcore/mcore.c: Likewise.
14450 * config/microblaze/microblaze.c: Likewise.
14451 * config/mips/mips.c: Likewise.
14452 * config/mmix/mmix.c: Likewise.
14453 * config/mn10300/mn10300.c: Likewise.
14454 * config/moxie/moxie.c: Likewise.
14455 * config/msp430/msp430.c: Likewise.
14456 * config/nds32/nds32-isr.c: Likewise.
14457 * config/nds32/nds32.c: Likewise.
14458 * config/nios2/nios2.c: Likewise.
14459 * config/nvptx/nvptx.c: Likewise.
14460 * config/pa/pa.c: Likewise.
14461 * config/pdp11/pdp11.c: Likewise.
14462 * config/powerpcspe/powerpcspe.c: Likewise.
14463 * config/riscv/riscv.c: Likewise.
14464 * config/rl78/rl78.c: Likewise.
14465 * config/rx/rx.c: Likewise.
14466 * config/s390/s390.c: Likewise.
14467 * config/sh/sh.c: Likewise.
14468 * config/sol2.c: Likewise.
14469 * config/sparc/sparc.c: Likewise.
14470 * config/spu/spu.c: Likewise.
14471 * config/stormy16/stormy16.c: Likewise.
14472 * config/tilegx/tilegx.c: Likewise.
14473 * config/tilepro/tilepro.c: Likewise.
14474 * config/v850/v850.c: Likewise.
14475 * config/vax/vax.c: Likewise.
14476 * config/visium/visium.c: Likewise.
14477 * config/xtensa/xtensa.c: Likewise.
14479 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14482 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
14483 constraints since the -mupper-regs-* switches have been
14485 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
14486 into a vector from a double word element that was extracted from
14487 another vector, and eliminate extra XXPERMDI instructions.
14488 (vsx_concat_<mode>_2): Likewise.
14489 (vsx_concat_<mode>_3): Likewise.
14490 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
14491 concat to allow optimizing inserts from previous extracts.
14493 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
14495 * config/i386/i386.c (ix86_stack_protect_guard): Generate
14496 memory reference to a SSP offset in TLS address space.
14497 (ix86_print_operand) <case '@'>: Remove.
14498 (ix86_print_operand_punct_valid_p): Remove '@' code.
14499 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
14500 UNSPEC_SP_TLS_TEST.
14501 (stack_tls_protect_set_<mode>): Remove.
14502 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
14503 (stack_tls_protect_test_<mode>): Remove.
14504 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
14506 2017-08-07 Olivier Hainque <hainque@adacore.com>
14509 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
14511 2017-08-07 Douglas Rupp <rupp@adacore.com>
14513 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
14514 variable was referenced as multidir in command.
14516 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14519 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
14522 2017-08-07 Martin Liska <mliska@suse.cz>
14524 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
14525 * config/rl78/rl78.c: Add include of attribs.h.
14526 * config/sh/sh.c: Likewise.
14527 * config/v850/v850.c: Likewise.
14529 2017-08-07 Tom de Vries <tom@codesourcery.com>
14531 PR middle-end/78266
14532 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
14534 2017-08-07 Martin Liska <mliska@suse.cz>
14536 * config/mips/mips.c: Include attribs.h.
14538 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
14541 * doc/invoke.texi: Document change in behvaior for -Ofast for
14544 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
14546 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
14548 (aarch64_pop_regs): Likewise.
14549 (aarch64_gen_load_pair): Likewise.
14550 (aarch64_save_callee_saves): Likewise.
14551 (aarch64_restore_callee_saves): Likewise.
14553 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14555 * config/i386/i386.c: Revert the last change.
14557 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14560 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14562 (ix86_finalize_stack_frame_flags): This. Also clear
14563 frame_pointer_needed if -fno-omit-frame-pointer is used without
14565 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14566 with ix86_finalize_stack_frame_flags.
14567 (ix86_expand_epilogue): Likewise.
14568 (ix86_expand_split_stack_prologue): Likewise.
14570 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14573 * config/i386/i386.c (get_builtin_code_for_version): Set priority
14574 to P_AES for Westmere.
14576 2017-08-07 Jonathan Yong <10walls@gmail.com>
14578 * config/i386/mingw.opt (fset-stack-executable): Removed.
14579 * config/i386/cygming.opt (fset-stack-executable): Moved
14581 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
14583 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
14585 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
14587 2017-08-07 Marek Polacek <polacek@redhat.com>
14589 PR middle-end/81737
14590 * fold-const.c (fold_indirect_ref_1): Check type_domain.
14592 2017-08-07 Martin Liska <mliska@suse.cz>
14594 * attribs.h (canonicalize_attr_name): New function.
14595 (cmp_attribs): Move from c-format.c and adjusted.
14596 (is_attribute_p): Moved from tree.h.
14597 * tree-inline.c: Add new includes.
14598 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
14599 (private_is_attribute_p): Remove.
14600 (private_lookup_attribute): Likewise.
14601 (private_lookup_attribute_by_prefix): Simplify.
14602 (remove_attribute): Use is_attribute_p.
14603 * tree.h: Remove removed declarations.
14605 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14607 PR middle-end/81698
14608 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
14609 instead of computing it in the function. Formatting fix.
14610 (expand_case): Don't rely on default_edge being the first edge,
14611 clear it if removing it, pass default_edge to
14612 emit_case_dispatch_table.
14613 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
14616 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
14618 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
14619 insn in the function, emit NOP after the insn.
14621 2017-08-06 Tom de Vries <tom@codesourcery.com>
14623 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
14626 2017-08-06 Tom de Vries <tom@codesourcery.com>
14628 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
14631 2017-08-04 Yury Gribov <tetra2005@gmail.com>
14633 PR tree-optimization/57371
14634 * match.pd: New pattern.
14636 2017-08-04 Marek Polacek <polacek@redhat.com>
14638 PR middle-end/81695
14639 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
14640 perform the computation in offset_int.
14642 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14644 PR tree-optimization/81136
14645 * tree-vectorizer.h: Include tree-hash-traits.h.
14646 (vec_base_alignments): New typedef.
14647 (vec_info): Add a base_alignments field.
14648 (vect_record_base_alignments): Declare.
14649 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
14651 (DR_IS_CONDITIONAL_IN_STMT): New macro.
14652 (create_data_ref): Add an is_conditional_in_stmt argument.
14653 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
14654 the is_conditional_in_stmt field.
14655 (data_ref_loc): Add an is_conditional_in_stmt field.
14656 (get_references_in_stmt): Set the is_conditional_in_stmt field.
14657 (find_data_references_in_stmt): Update call to create_data_ref.
14658 (graphite_find_data_references_in_stmt): Likewise.
14659 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
14660 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
14661 (vect_record_base_alignment): New function.
14662 (vect_record_base_alignments): Likewise.
14663 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
14664 for nested statements even if we fail to compute a misalignment.
14665 Use pooled base alignments for unconditional references.
14666 (vect_find_same_alignment_drs): Compare base addresses instead
14668 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
14669 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
14671 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14673 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
14674 Add an explicit name for the enum. Use auto_vec for slp_instances
14675 and grouped_stores.
14676 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
14678 (_bb_vec_info): Add a constructor and destructor.
14679 (vinfo_for_stmt): Return NULL for uids of -1 as well.
14680 (destroy_loop_vec_info): Delete.
14681 (vect_destroy_datarefs): Likewise.
14682 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
14683 (vec_info::vec_info): New function.
14684 (vec_info::~vec_info): Likewise.
14685 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
14686 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
14687 destroy_loop_vec_info.
14688 * tree-vect-loop.c (new_loop_vec_info): Replace with...
14689 (_loop_vec_info::_loop_vec_info): ...this.
14690 (destroy_loop_vec_info): Replace with...
14691 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
14692 the stmt_vec_infos. Leave handling of vec_info information to its
14693 destructor. Remove explicit vector releases.
14694 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
14695 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
14696 * tree-vect-slp.c (new_bb_vec_info): Replace with...
14697 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
14698 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
14699 (destroy_bb_vec_info): Replace with...
14700 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
14701 information to its destructor.
14702 (vect_slp_analyze_bb_1): Use new and delete instead of
14703 new_bb_vec_info and destroy_bb_vec_info.
14704 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
14707 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14709 * tree-data-ref.h (subscript): Add access_fn field.
14710 (data_dependence_relation): Add could_be_independent_p.
14711 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
14712 (same_access_functions): Move to tree-data-ref.c.
14713 * tree-data-ref.c (ref_contains_union_access_p): New function.
14714 (access_fn_component_p): Likewise.
14715 (access_fn_components_comparable_p): Likewise.
14716 (dr_analyze_indices): Add a reference to access_fn_component_p.
14717 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
14719 (constant_access_functions): Likewise.
14720 (add_other_self_distances): Likewise.
14721 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
14722 (initialize_data_dependence_relation): Use XCNEW and remove
14723 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
14724 of access functions that have the same type. Allow the
14725 subsequence to end with different bases in some circumstances.
14726 Record the chosen access functions in SUB_ACCESS_FN.
14727 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
14728 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
14729 (subscript_dependence_tester_1): Likewise dra and drb.
14730 (build_classic_dist_vector): Update calls accordingly.
14731 (subscript_dependence_tester): Likewise.
14732 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
14733 DDR_COULD_BE_INDEPENDENT_P.
14734 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
14735 comp_alias_ddrs instead of may_alias_ddrs.
14736 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14738 (vect_analyze_data_ref_dependence): Use it if
14739 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
14740 distance vectors if that fails.
14741 (dependence_distance_ge_vf): New function.
14742 (vect_prune_runtime_alias_test_list): Use it. Don't clear
14743 LOOP_VINFO_MAY_ALIAS_DDRS.
14745 2017-08-04 Richard Biener <rguenther@suse.de>
14747 PR middle-end/81705
14748 * fold-const.c (fold_binary_loc): Properly restrict
14749 minus_var0 && minus_var1 case when associating undefined overflow
14752 2017-08-04 Tom de Vries <tom@codesourcery.com>
14754 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
14756 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14758 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14759 Don't start diagnostic messages with a capital letter.
14760 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14762 (rs6000_invalid_builtin): Likewise.
14763 (rs6000_trampoline_init): Likewise.
14765 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14768 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
14769 after setting changeable df flags.
14771 2017-08-03 Richard Biener <rguenther@suse.de>
14773 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
14774 up if the use is in USE - X.
14776 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
14778 * toplev.c (dumpfile.h): New include.
14779 (internal_error_reentered): New static function. Use it...
14780 (internal_error_function): ...here to handle reentered internal_error.
14782 2017-08-03 Richard Biener <rguenther@suse.de>
14784 PR middle-end/81148
14785 * fold-const.c (split_tree): Add minus_var and minus_con
14786 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
14787 here but always use minus_*.
14788 (associate_trees): Assert we never associate with MINUS_EXPR
14789 and NULL first operand. Do not recurse for PLUS_EXPR operands
14790 when associating as MINUS_EXPR either.
14791 (fold_binary_loc): Track minus_var and minus_con.
14793 2017-08-03 Tom de Vries <tom@codesourcery.com>
14796 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
14797 ACCEL_COMPILER, apply finish_options on
14798 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
14800 2017-08-03 Tom de Vries <tom@codesourcery.com>
14803 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
14804 function_entry_patch_area_size > 0.
14806 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14809 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
14810 instead of 10??LU, perform unit multiplication in wide_int,
14811 don't change alloc_object_size_limit if the limit is larger
14814 PR tree-optimization/81655
14815 PR tree-optimization/81588
14816 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
14817 the case when ranges[i].low and high are 1 for unsigned type with
14820 PR middle-end/81052
14821 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
14822 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
14824 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14826 * tree-vrp.h: Add include guard.
14828 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
14831 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14832 (ud2): New insn pattern.
14833 * config/i386/i386.c (ix86_expand_epilogue):
14834 For naked functions, generate ud2 instead of trap insn.
14836 2017-08-02 Marek Polacek <polacek@redhat.com>
14839 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14841 2017-08-02 Tom de Vries <tom@codesourcery.com>
14842 Cesar Philippidis <cesar@codesourcery.com>
14844 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14845 Add missing edge probabilities.
14847 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14849 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14850 Correct endianness.
14852 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14854 PR middle-end/79499
14855 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14856 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14857 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14858 prologue_seq sequences - if any.
14860 2017-08-02 Richard Biener <rguenther@suse.de>
14862 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14863 via vectors if supported, integer extracts via punning if supported
14864 or otherwise vector extracts.
14866 2017-08-02 Richard Biener <rguenther@suse.de>
14868 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14870 (bitmap_insert_into_set): ... this.
14872 2017-08-02 Richard Biener <rguenther@suse.de>
14874 PR tree-optimization/81633
14876 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14878 PR tree-optimization/71752
14879 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14881 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14883 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14884 (machine_function::call_ms2sysv_pad_out): Remove field.
14885 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14886 (ix86_compute_frame_layout): Likewise.
14888 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
14891 * config/i386/i386.c (ix86_set_func_type): Disallow naked
14892 attribute with interrupt attribute.
14894 2017-08-01 Andrew Pinski <apinski@cavium.com>
14896 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
14897 BIT_INSERT_EXPR's operand 1
14898 to see if the types precision matches.
14900 2017-08-01 Martin Liska <mliska@suse.cz>
14902 PR middle-end/70140
14903 * builtins.c (expand_builtin_memcpy_args): Remove.
14904 (expand_builtin_memcpy): Call newly added function
14905 expand_builtin_memory_copy_args.
14906 (expand_builtin_memcpy_with_bounds): Likewise.
14907 (expand_builtin_mempcpy): Remove last argument.
14908 (expand_builtin_mempcpy_with_bounds): Likewise.
14909 (expand_builtin_memory_copy_args): New function created from
14910 expand_builtin_mempcpy_args with small modifications.
14911 (expand_builtin_mempcpy_args): Remove.
14912 (expand_builtin_stpcpy): Remove unused argument.
14913 (expand_builtin): Likewise.
14914 (expand_builtin_with_bounds): Likewise.
14916 2017-08-01 Martin Liska <mliska@suse.cz>
14919 Make mempcpy more optimal (PR middle-end/70140).
14921 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14924 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14925 __builtin_vec_cmpne verify both arguments are compatible vectors
14926 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14927 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14928 move computation of aligned to after checking the argument types.
14932 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14933 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
14936 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
14937 Doug Rupp <rupp@adacore.com>
14938 Olivier Hainque <hainque@adacore.com>
14940 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
14941 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
14943 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
14944 for TARGET_OS_CPP_BUILTIN.
14945 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
14946 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
14948 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
14949 passing required abi options to the assembler for EABI configurations.
14950 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
14951 of .text.hot and .text.unlikely sections for kernel modules when
14952 using ARM style exceptions.
14953 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
14954 options. Add EXTRA_CC1_SPEC.
14955 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
14957 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
14959 (ARM_TARGET2_DWARF_FORMAT): Define.
14960 * config/arm/t-vxworks: Adjust multilib control to removal of the
14961 Diab command line options.
14963 2017-08-01 Martin Liska <mliska@suse.cz>
14965 PR gcov-profile/81561
14966 * gcov.c (unblock): Make unblocking safe as we need to preserve
14967 index correspondence of blocks and block_lists.
14969 2017-08-01 Richard Biener <rguenther@suse.de>
14971 PR tree-optimization/81181
14972 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
14973 (compute_antic): ... end of iteration here.
14975 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
14977 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
14978 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
14979 (ftree-slp-vectorize): Likewise.
14980 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
14981 can no longer be set independent of flag_tree_loop_vectorize.
14982 * omp-general.c (emp_max_vf): Likewise.
14983 * opts.c (enable_fdo_optimizations): Remove references to
14984 flag_tree_vectorize, these are now implicit.
14985 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14986 and leave it for the options machinery.
14988 2017-08-01 Martin Liska <mliska@suse.cz>
14990 PR middle-end/70140
14991 * builtins.c (expand_builtin_memcpy_args): Remove.
14992 (expand_builtin_memcpy): Call newly added function
14993 expand_builtin_memory_copy_args.
14994 (expand_builtin_memcpy_with_bounds): Likewise.
14995 (expand_builtin_mempcpy): Remove last argument.
14996 (expand_builtin_mempcpy_with_bounds): Likewise.
14997 (expand_builtin_memory_copy_args): New function created from
14998 expand_builtin_mempcpy_args with small modifications.
14999 (expand_builtin_mempcpy_args): Remove.
15000 (expand_builtin_stpcpy): Remove unused argument.
15001 (expand_builtin): Likewise.
15002 (expand_builtin_with_bounds): Likewise.
15004 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15007 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15008 print "ds:" only for immediates in generic address space.
15010 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15013 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15014 (ix86_function_ok_for_sibcall): Return false for naked functions.
15016 2017-08-01 Richard Biener <rguenther@suse.de>
15018 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15019 (compute_antic): Seed worklist with exit block predecessors.
15020 * cfganal.c (dfs_find_deadend): For a cycle return the source
15021 of the edge closing it.
15023 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15025 * config/aarch64/aarch64.c
15026 (aarch64_can_const_movi_rtx_p): Move 0 check.
15028 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15030 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15031 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15033 * match.pd: Ditto in address comparison pattern.
15035 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15037 PR tree-optimization/81627
15038 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15039 closed ssa form for store-store chain.
15041 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15043 PR tree-optimization/81620
15044 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15045 for store-store chain.
15047 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15049 PR tree-optimization/81588
15050 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15051 ranges[i].in_p, invert comparison code ccode. For >/>=,
15052 swap rhs1 and rhs2 and comparison code unconditionally,
15053 for </<= don't do that. Don't swap rhs1/rhs2 again if
15054 ranges[i].in_p, instead invert comparison code ccode if
15055 opcode or oe->rank is BIT_IOR_EXPR.
15058 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15059 a direct optab to conversion optab.
15060 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15061 with GET_MODE_INNER as last argument instead of optab_handler.
15062 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15063 vector extraction if possible and optab is available.
15064 * expr.c (store_constructor): Use convert_optab_handler instead
15065 of optab_handler. Use vector initialization from smaller
15066 vectors if possible and optab is available.
15067 * tree-vect-stmts.c (vectorizable_load): Likewise.
15068 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15069 now have two modes.
15070 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15071 of vec_init from half-sized vectors with the same element mode.
15072 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15073 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15074 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15075 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15076 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15077 after mode in gen_vec_extract* calls.
15078 (vec_extract<mode>): Renamed to ...
15079 (vec_extract<mode><ssescalarmodelower>): ... this.
15080 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15081 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15082 element mode after mode in gen_vec_init* calls.
15083 (VEC_INIT_HALF_MODE): New mode iterator.
15084 (vec_init<mode>): Renamed to ...
15085 (vec_init<mode><ssescalarmodelower>): ... this.
15086 (vec_init<mode><ssehalfvecmodelower>): New expander.
15087 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15088 (vec_extractv2sfsf): ... this.
15089 (vec_initv2sf): Renamed to ...
15090 (vec_initv2sfsf): ... this.
15091 (vec_extractv2si): Renamed to ...
15092 (vec_extractv2sisi): ... this.
15093 (vec_initv2si): Renamed to ...
15094 (vec_initv2sisi): ... this.
15095 (vec_extractv4hi): Renamed to ...
15096 (vec_extractv4hihi): ... this.
15097 (vec_initv4hi): Renamed to ...
15098 (vec_initv4hihi): ... this.
15099 (vec_extractv8qi): Renamed to ...
15100 (vec_extractv8qiqi): ... this.
15101 (vec_initv8qi): Renamed to ...
15102 (vec_initv8qiqi): ... this.
15103 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15104 (vec_init<mode>): Renamed to ...
15105 (vec_init<mode><VEC_base_l>): ... this.
15106 (vec_extract<mode>): Renamed to ...
15107 (vec_extract<mode><VEC_base_l>): ... this.
15108 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15109 (vec_initv2sfsf): ... this.
15110 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15111 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15112 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15113 element mode after mode in gen_vec_init* calls.
15114 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15115 (vec_init<mode><Vel>): ... this.
15116 (vec_extract<mode>): Renamed to ...
15117 (vec_extract<mode><Vel>): ... this.
15118 * config/aarch64/iterators.md (Vel): New mode attribute.
15119 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15120 Add element mode after mode in gen_vec_extract* calls.
15121 * config/s390/vector.md (non_vec_l): New mode attribute.
15122 (vec_extract<mode>): Renamed to ...
15123 (vec_extract<mode><non_vec_l>): ... this.
15124 (vec_init<mode>): Renamed to ...
15125 (vec_init<mode><non_vec_l>): ... this.
15126 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15127 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15129 * config/arm/iterators.md (V_elem_l): New mode attribute.
15130 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15131 (vec_extract<mode><V_elem_l>): ... this.
15132 (vec_extractv2di): Renamed to ...
15133 (vec_extractv2didi): ... this.
15134 (vec_init<mode>): Renamed to ...
15135 (vec_init<mode><V_elem_l>): ... this.
15136 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15137 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15138 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15139 Add element mode after gen_vec_extract* calls.
15140 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15141 (vec_init<mode><unitmode>): ... this.
15142 (vec_extract<mode>): Renamed to ...
15143 (vec_extract<mode><unitmode>): ... this.
15144 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15145 (vec_init<mode><unitmode>): ... this.
15146 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15147 (vec_initv2sfsf): ... this.
15148 (vec_extractv2sf): Renamed to ...
15149 (vec_extractv2sfsf): ... this.
15150 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15151 Add element mode after gen_vec_extract* calls.
15152 * config/mips/mips.md (unitmode): New mode iterator.
15153 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15154 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15155 * config/spu/spu.md (inner_l): New mode attribute.
15156 (vec_init<mode>): Renamed to ...
15157 (vec_init<mode><inner_l>): ... this.
15158 (vec_extract<mode>): Renamed to ...
15159 (vec_extract<mode><inner_l>): ... this.
15160 * config/sparc/sparc.md (veltmode): New mode iterator.
15161 (vec_init<VMALL:mode>): Renamed to ...
15162 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15163 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15164 (vec_initv2sisi): ... this.
15165 (vec_initv2sf): Renamed to ...
15166 (vec_initv2sfsf): ... this.
15167 (vec_extractv2sf): Renamed to ...
15168 (vec_extractv2sfsf): ... this.
15169 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15170 (vec_init<mode>): Renamed to ...
15171 (vec_init<mode><VEC_base_l>): ... this.
15172 (vec_extract<mode>): Renamed to ...
15173 (vec_extract<mode><VEC_base_l>): ... this.
15174 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15175 (vec_initv2sfsf): ... this.
15176 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15177 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15178 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15179 gen_vec_init* calls.
15181 2017-08-01 Richard Biener <rguenther@suse.de>
15183 PR tree-optimization/81297
15184 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15185 TREE_OVERFLOW from INTEGER_CSTs.
15187 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15189 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15191 2017-07-31 Carl Love <cel@us.ibm.com>
15193 * config/rs6000/rs6000-c: Add support for built-in functions
15194 vector signed char vec_xl_be (signed long long, signed char *);
15195 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15196 vector signed int vec_xl_be (signed long long, signed int *);
15197 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15198 vector signed long long vec_xl_be (signed long long, signed long long *);
15199 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15200 vector signed short vec_xl_be (signed long long, signed short *);
15201 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15202 vector double vec_xl_be (signed long long, double *);
15203 vector float vec_xl_be (signed long long, float *);
15204 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15205 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15206 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15208 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15209 (altivec_expand_builtin): Add switch statement to call
15210 altivec_expand_xl_be for each builtin.
15211 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15212 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15213 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15214 __builtin_vsx_le_be_v16qi.
15215 * doc/extend.texi: Update the built-in documentation file for the
15216 new built-in functions.
15218 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15221 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15223 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15225 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15227 * config.gcc: Add z14.
15228 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15229 CPU model numbers for z13s and z14.
15230 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15232 * config/s390/s390-opts.h (enum processor_type): Rename
15233 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15234 * config/s390/s390.c (processor_table): Add field for CPU name to
15235 be passed to Binutils.
15236 (s390_asm_output_machine_for_arch): Use the new field in
15237 processor_table for Binutils.
15238 (s390_expand_builtin): Replace arch12 with z14.
15239 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15240 (s390_get_sched_attrmask): Likewise.
15241 (s390_get_unit_mask): Likewise.
15242 * config/s390/s390.opt: Add z14 to processor_type enum.
15244 2017-07-31 Martin Jambor <mjambor@suse.cz>
15247 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15248 regardless of optimization level.
15250 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15251 Martin Liska <mliska@suse.cz>
15253 * predict.def: Remove old comment and adjust probability.
15254 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15255 PREDICT statements.
15257 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15260 * config/i386/i386.c (ix86_function_naked): New function.
15261 (ix86_can_use_return_insn_p): Return false for naked functions.
15262 (ix86_expand_prologue): Skip prologue for naked functions.
15263 (ix86_expand_epilogue): Skip epilogue for naked functions
15264 and emit trap instruction.
15265 (ix86_warn_func_return): New function.
15266 (ix86_attribute_table): Add "naked" attribute specification.
15267 (TARGET_WARN_FUNC_RETURN): Define.
15268 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15270 2017-07-31 Martin Liska <mliska@suse.cz>
15272 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15273 (dump_gimple_bb_header): Always dump BB info.
15274 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15276 2017-07-31 Martin Liska <mliska@suse.cz>
15279 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15280 also with current_function_decl non-null equality.
15282 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15285 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15286 change type to the element type, instead add eltype variable and
15287 use it where we are interested in the element type.
15289 PR tree-optimization/81603
15290 * ipa-polymorphic-call.c
15291 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15292 offset arithmetic in offset_int, bail out if the resulting bit offset
15293 doesn't fit into shwi.
15295 2017-07-31 Martin Liska <mliska@suse.cz>
15297 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15298 (gimplify_save_expr): Fix comment.
15300 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15303 * config/i386/i386.c (ix86_function_arg): Update arguments for
15305 (ix86_compute_frame_layout): Set the initial stack offset to
15306 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15307 INCOMING_FRAME_SP_OFFSET.
15308 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15309 stack before exception handler returns.
15310 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15311 the 'ERROR_CODE' for exception handler.
15313 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15315 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15316 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15317 (ASM_OUTPUT_REG_POP): Ditto.
15318 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15319 instead of asm_fprintf to output pure string.
15321 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15323 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15324 to imported_module_or_decl hook.
15325 (debug_nothing_tree_tree_tree_bool): Remove.
15326 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15327 * debug.c (do_nothing_debug_hooks): Use
15328 debug_nothing_tree_tree_tree_bool_bool instead of
15329 debug_nothing_tree_tree_tree_bool.
15330 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15331 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15332 * sdbout.c (sdb_debug_hooks): Likewise.
15333 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15334 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15336 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15337 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15338 attribute, don't add anything.
15340 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15342 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15343 (fold_build2_stat_loc): Likewise.
15344 (fold_build3_stat_loc): Likewise.
15345 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15346 (fold_build1_loc): Remove macro.
15347 (fold_build2_loc): Likewise.
15348 (fold_build3_loc): Likewise.
15350 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15352 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15353 (gimple_build_debug_bind_source_stat): Likewise.
15354 * gimple.h (gimple_build_debug_bind): Remove macro.
15355 (gimple_build_debug_bind_source): Likewise.
15357 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15359 * bitmap.c (bitmap_alloc): Adjust.
15360 (bitmap_gc_alloc): Likewise.
15361 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15363 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15365 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15366 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15367 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15368 (bitmap_gc_alloc_stat): Likewise.
15369 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15371 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15373 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15374 * rtl.h (shallow_copy_rtx): Remove macro.
15376 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15378 * emit-rtl.c (gen_raw_REG): Adjust.
15379 * gengenrtl.c (gendef): Likewise.
15380 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15381 * rtl.h (rtx_alloc): Remove macro.
15383 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15385 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
15386 (build_tree_list_stat): Likewise.
15387 * tree.h (build_tree_list): Remove macro.
15388 (build_tree_list_vec): Likewise.
15390 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15392 * tree.c (make_vector_stat): Remove _stat from name.
15393 (build_vector_stat): Likewise.
15394 * tree.h (make_vector_stat): Remove macro.
15395 (build_vector_stat): Likewise.
15397 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15399 * tree.h (build_var_debug_value): Remove prototype.
15401 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15403 * tree.c (tree_cons_stat): Remove _stat from name.
15404 * tree.h (tree_cons): Remove macro.
15406 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15408 * tree.c (build_vl_exp_stat): Remove _stat from name.
15409 * tree.h (build_vl_exp): Remove macro.
15411 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15413 * tree.c (build_decl_stat): Remove _stat from name.
15414 * tree.h (build_decl): Remove macro.
15416 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15418 * gimple.c (gimple_build_with_ops_stat): Adjust.
15419 (gimple_alloc_stat): Remove _stat from name.
15420 * gimple.h (gimple_alloc): Remove macro.
15422 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15424 * tree.c (make_tree_vec_stat): Remove _stat from name.
15425 (grow_tree_vec_stat): Likewise.
15426 * tree.h (make_tree_vec_stat): Adjust prototype.
15427 (grow_tree_vec_stat): Likewise.
15428 (make_tree_vec): Remove macro.
15429 (grow_tree_vec): Likewise.
15431 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15433 * fold-const.c (fold_build1_stat_loc): Adjust.
15434 (fold_build2_stat_loc): Likewise.
15435 (fold_build3_stat_loc): Likewise.
15436 * tree.c (build0_stat): Remove _stat from name.
15437 (build1_stat): Likewise.
15438 (build2_stat): Likewise.
15439 (build3_stat): Likewise.
15440 (build4_stat): Likewise.
15441 (build5_stat): Likewise.
15442 * tree.h (build1_loc): Remove macro, and rename _stat function
15444 (build2_loc): Likewise.
15445 (build3_loc): Likewise.
15446 (build4_loc): Likewise.
15447 (build5_loc): Likewise.
15449 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15451 * tree.c (make_int_cst_stat): Remove _stat from name.
15452 * tree.h (make_int_cst_stat): Adjust prototype.
15453 (make_int_cst): Remove macro.
15455 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15457 * tree.c (make_tre_binfo_stat): Remove _stat from name.
15458 * tree.h (make_tree_binfo_stat): Adjust prototype.
15459 (make_tree_binfo): Remove.
15461 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15463 * tree.c (copy_node_stat): Rename to copy_node.
15464 (build_distinct_type_copy): Adjust.
15465 * tree.h (copy_node_stat): Adjust prototype.
15466 (copy_node): Remove macro.
15468 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15470 * tree.c (make_node_stat): rename to make_node.
15471 (build_tree_list_stat): Adjust.
15472 (build0_stat): Likewise.
15473 (build2_stat): Likewise.
15474 (build3_stat): Likewise.
15475 (build4_stat): Likewise.
15476 (build5_stat): Likewise.
15477 (build_decl_stat): Likewise.
15478 * tree.h (make_node_stat): Adjust prototype.
15479 (make_node): remove macro.
15481 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
15483 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
15484 (PPC_FEATURE2_SCV): Likewise.
15485 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
15487 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15489 * config/aarch64/aarch64.c
15490 (aarch64_internal_mov_immediate): Add new special pattern.
15491 * config/aarch64/aarch64.md (*movdi_aarch64):
15492 Add reg/32bit const mov case.
15494 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15495 Richard Sandiford <richard.sandiford@linaro.org>
15497 * config/aarch64/aarch64.md (mov<mode>): Generalize.
15498 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15499 Add integer and movi cases.
15500 (movi-split-hf-df-sf split, fp16): New.
15501 (enabled): Added TARGET_FP_F16INST.
15502 * config/aarch64/iterators.md (GPF_HF): New.
15503 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
15505 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15507 * config/aarch64/aarch64.c
15508 (aarch64_simd_container_mode): Add prototype.
15509 (aarch64_expand_mov_immediate): Add HI support.
15510 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
15511 (aarch64_can_const_movi_rtx_p): New.
15512 (aarch64_preferred_reload_class):
15513 Remove restrictions of using FP registers for certain SIMD operations.
15514 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
15515 (aarch64_valid_floating_const): Add integer move validation.
15516 (aarch64_simd_imm_scalar_p): Remove.
15517 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
15518 (aarch64_legitimate_constant_p): Expand list of supported cases.
15519 * config/aarch64/aarch64-protos.h
15520 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
15521 (aarch64_reinterpret_float_as_int): New.
15522 (aarch64_simd_imm_scalar_p): Remove.
15523 * config/aarch64/constraints.md (Uvi): New.
15524 (Dd): Split into Ds and new Dd.
15525 * config/aarch64/aarch64.md (*movsi_aarch64):
15527 (*movdi_aarch64): Add SIMD mov case.
15529 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15531 * tree-predcom.c: (struct chain): Handle store-store chain in which
15532 stores for elimination only store loop invariant values.
15533 (execute_pred_commoning_chain): Ditto.
15534 (prepare_initializers_chain_store_elim): Ditto.
15535 (prepare_finalizers): Ditto.
15536 (is_inv_store_elimination_chain): New function.
15537 (initialize_root_vars_store_elim_1): New function.
15539 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15541 * tree-predcom.c: Revise general description of the pass.
15542 (enum chain_type): New enum type for store elimination.
15543 (struct chain): New field supporting store elimination.
15544 (struct component): Ditto.
15545 (dump_chain): Dump store-stores chain.
15546 (release_chain): Release resources.
15547 (split_data_refs_to_components): Compute and create component
15548 contains only stores for elimination.
15549 (get_chain_last_ref_at): New function.
15550 (make_invariant_chain): Initialization.
15551 (make_rooted_chain): Specify chain type in parameter and record it.
15552 (add_looparound_copies): Skip for store-stores chain.
15553 (determine_roots_comp): Compute type of chain and pass it to
15555 (initialize_root_vars_store_elim_2): New function.
15556 (finalize_eliminated_stores): New function.
15557 (remove_stmt): Handle store for elimination.
15558 (execute_pred_commoning_chain): Execute predictive commoning on
15559 store-store chains.
15560 (determine_unroll_factor): Skip unroll for store-stores chain.
15561 (prepare_initializers_chain_store_elim): New function.
15562 (prepare_initializers_chain): Hanlde store-store chain.
15563 (prepare_finalizers_chain, prepare_finalizers): New function.
15564 (tree_predictive_commoning_loop): Return integer value indicating
15565 if loop is unrolled or lcssa form is corrupted.
15566 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
15568 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15570 * tree-predcom.c (initialize_root): Delete.
15571 (execute_pred_commoning_chain): Initialize root vars and replace
15572 reference of non-combined chain directly, rather than call above
15575 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15577 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
15578 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
15580 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15582 * tree-predcom.c (struct chain): New field init_seq.
15583 (release_chain): Release init_seq.
15584 (prepare_initializers_chain): Record intialization stmts in above
15586 (insert_init_seqs): New function.
15587 (tree_predictive_commoning_loop): Call insert_init_seqs.
15589 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15591 * tree-predcom.c (determine_roots_comp): Skip trivial components.
15593 2017-07-28 Richard Biener <rguenther@suse.de>
15595 * match.pd: Remove superfluous :c.
15596 * genmatch.c (simplify::id): Add member.
15597 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
15599 (current_id): New global.
15600 (dt_node::parent): Move from ...
15601 (dt_operand::parent): ... here. Add for_id member.
15602 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
15603 (decision_tree::find_node): Relax order requirement when
15604 merging DT_TRUE nodes to ones inbetween the current simplify
15605 and the one we try to merge with. Add diagnostic whenever
15606 we need to enforce pattern order by not merging.
15607 (decision_tree::insert): Set current_id.
15608 (decision_tree::print_node): Dump parent node and for_id.
15609 (parser::last_id): Add member.
15610 (parser::push_simplify): Assign unique id.
15611 (parser::parser): Initialize last_id.
15613 2017-07-28 Martin Liska <mliska@suse.cz>
15616 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
15617 gimple_build_debug_bind.
15619 2017-07-28 Richard Biener <rguenther@suse.de>
15621 PR tree-optimization/81502
15622 * match.pd: Add pattern combining BIT_INSERT_EXPR with
15624 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
15626 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
15627 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
15628 for BIT_FIELD_REF args.
15629 * fold-const.c (make_bit_field_ref): Likewise.
15630 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
15632 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15635 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
15636 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
15637 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
15638 Or it into SANITIZER_UNDEFINED.
15639 * ubsan.c: Include gimple-fold.h and varasm.h.
15640 (ubsan_expand_ptr_ifn): New function.
15641 (instrument_pointer_overflow): New function.
15642 (maybe_instrument_pointer_overflow): New function.
15643 (instrument_object_size): Formatting fix.
15644 (pass_ubsan::execute): Call instrument_pointer_overflow
15645 and maybe_instrument_pointer_overflow.
15646 * internal-fn.c (expand_UBSAN_PTR): New function.
15647 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
15648 * sanitizer.def (__ubsan_handle_pointer_overflow,
15649 __ubsan_handle_pointer_overflow_abort): New builtins.
15650 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
15651 * internal-fn.def (UBSAN_PTR): New internal function.
15652 * opts.c (sanitizer_opts): Add pointer-overflow.
15653 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
15654 * fold-const.c (build_range_check): Compute pointer range check in
15655 integral type if pointer arithmetics would be needed. Formatting
15658 2017-07-28 Martin Liska <mliska@suse.cz>
15661 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
15662 parameters that are of a variable-length.
15664 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
15666 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
15668 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
15669 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
15670 (CRT_CALL_STATIC_FUNCTION): Likewise.
15671 (ASM_DEFAULT_SPEC): New define.
15672 (ASM_SPEC32): Likewise.
15673 (ASM_SPEC64): Likewise.
15674 (ASM_SPEC_COMMON): Likewise.
15675 (ASM_SPEC): Likewise.
15676 (INVALID_64BIT): Likewise.
15677 (LINK_OS_DEFAULT_SPEC): Likewise.
15678 (LINK_OS_SPEC32): Likewise.
15679 (LINK_OS_SPEC64): Likewise.
15680 (POWERPC_LINUX): Likewise.
15681 (PTRDIFF_TYPE): Likewise.
15682 (RESTORE_FP_PREFIX): Likewise.
15683 (RESTORE_FP_SUFFIX): Likewise.
15684 (SAVE_FP_PREFIX): Likewise.
15685 (SAVE_FP_SUFFIX): Likewise.
15686 (SIZE_TYPE): Likewise.
15687 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
15688 (TARGET_64BIT): Likewise.
15689 (TARGET_64BIT): Likewise.
15690 (TARGET_AIX): Likewise.
15691 (WCHAR_TYPE_SIZE): Likewise.
15692 (WCHAR_TYPE): Undefine.
15693 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
15694 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
15695 (CPP_OS_RTEMS_SPEC): Delete.
15696 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
15697 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
15699 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
15701 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15703 PR tree-optimization/81578
15704 * tree-parloops.c (build_new_reduction): Bail out if
15705 reduction_code isn't one of the standard OpenMP reductions.
15706 Move the details printing after that decision.
15708 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15710 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
15711 related to reload_in_progress.
15712 (splat_input_operand): Likewise.
15713 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
15715 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
15717 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
15718 (TARGET_INSTANTIATE_DECLS): Likewise.
15719 (legitimate_indexed_address_p): Delete reload_in_progress code.
15720 (rs6000_debug_legitimate_address_p): Likewise.
15721 (rs6000_eliminate_indexed_memrefs): Likewise.
15722 (rs6000_emit_le_vsx_store): Likewise.
15723 (rs6000_emit_move_si_sf_subreg): Likewise.
15724 (rs6000_emit_move): Likewise.
15725 (register_to_reg_type): Likewise.
15726 (rs6000_pre_atomic_barrier): Likewise.
15727 (rs6000_machopic_legitimize_pic_address): Likewise.
15728 (rs6000_allocate_stack_temp): Likewise.
15729 (rs6000_address_for_fpconvert): Likewise.
15730 (rs6000_address_for_altivec): Likewise.
15731 (rs6000_secondary_memory_needed_rtx): Delete function.
15732 (rs6000_check_sdmode): Likewise.
15733 (rs6000_alloc_sdmode_stack_slot): Likewise.
15734 (rs6000_instantiate_decls): Likewise.
15735 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
15736 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
15737 Delete reload_in_progress.
15738 (*vec_reload_and_plus_<mptrsize>): Likewise.
15739 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
15740 (vsx_div_v2di): Likewise.
15741 (vsx_udiv_v2di): Likewise.
15743 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15745 * config/rs6000/rs6000.opt (mlra): Replace with stub.
15746 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
15747 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
15748 (rs6000_debug_reg_global): Delete print of LRA status.
15749 (rs6000_option_override_internal): Delete dead LRA related code.
15750 (rs6000_lra_p): Delete function.
15751 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
15753 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
15755 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
15756 * config/riscv/rtems.h: New file.
15758 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15759 Sudakshina Das <sudi.das@arm.com>
15761 * config/aarch64/aarch64.md
15762 (define_split for and<mode>3nr_compare): Move
15763 non aarch64_logical_operand to a register.
15764 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
15765 register immediate operand to a register.
15766 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
15768 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
15770 PR middle-end/81564
15771 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
15773 2017-07-27 Richard Biener <rguenther@suse.de>
15775 PR tree-optimization/81573
15776 PR tree-optimization/81494
15777 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
15778 multi defuse cycle case.
15780 2017-07-27 Richard Biener <rguenther@suse.de>
15782 PR tree-optimization/81571
15783 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
15786 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
15788 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
15789 earlier and only if MASK_FPU is set. Adjust formatting.
15791 2017-07-27 Martin Liska <mliska@suse.cz>
15793 * opt-functions.awk: Add validation of value of Init.
15794 * optc-gen.awk: Pass new argument.
15796 2017-07-27 Martin Liska <mliska@suse.cz>
15798 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
15799 Fix wrong condition.
15801 2017-07-27 Martin Liska <mliska@suse.cz>
15803 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
15804 BBs and edges seen by autoFDO.
15806 2017-07-27 Richard Biener <rguenther@suse.de>
15808 PR tree-optimization/81502
15809 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
15810 with incompatible but same sized type.
15811 (execute_update_addresses_taken): Likewise.
15813 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
15815 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
15816 flag_tree_loop_vectorize rather than flag_tree_vectorize.
15818 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15821 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
15822 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
15823 Change s_operand to memory_operand.
15825 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
15827 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
15828 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
15829 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
15830 Emit instructions rather than returning an expression. Handle TFmode
15831 and KFmode by casting to TImode.
15832 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15833 (rs6000_emit_le_vsx_store): Likewise.
15834 * config/rs6000/vsx.md (VSX_TI): New iterator.
15835 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15836 (*vsx_le_undo_permute_<mode>): Likewise.
15837 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15838 emit the split sequence.
15839 (*vsx_le_perm_store_<mode>): Likewise.
15841 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15843 PR tree-optimization/81555
15844 PR tree-optimization/81556
15845 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15846 if true, force CHANGED for the recursive invocation.
15847 (reassociate_bb): Remember original length of ops array, pass
15848 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15850 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15851 attributes for noipa attribute. For naked attribute use
15852 lookup_attribute first before lookup_attribute_spec.
15853 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15855 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15856 for functions with noipa attribute.
15857 (cgraph_externally_visible_p): Return true for functions with noipa
15859 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15860 for functions with noipa attribute.
15861 * doc/extend.texi: Document noipa function attribute.
15862 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15863 also for functions with noipa attribute.
15864 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15866 2017-07-26 Andrew Pinski <apinski@cavium.com>
15868 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15869 vec_unalign_load_cost and vec_unalign_store_cost.
15871 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15873 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15874 -mvsx-small-integer option.
15875 (ISA_3_0_MASKS_IEEE): Likewise.
15876 (OTHER_VSX_VECTOR_MASKS): Likewise.
15877 (POWERPC_MASKS): Likewise.
15878 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15879 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15880 code, only testing for DImode being allowed in non-VSX floating
15882 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15883 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15885 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15886 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15887 TARGET_P8_VECTOR test.
15888 (rs6000_secondary_reload_simple_move): Likewise.
15889 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
15890 since TARGET_P9_VECTOR was already tested.
15891 (rs6000_opt_masks): Remove -mvsx-small-integer.
15892 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
15893 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15895 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
15896 test for TARGET_VEXTRACTUB was used, and that uses
15898 (p9 extract splitter): Likewise.
15899 (vsx_extract_<mode>_di_p9): Likewise.
15900 (vsx_extract_<mode>_store_p9): Likewise.
15901 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15902 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15903 the elimination of TARGET_VSX_SMALL_INTEGER.
15904 (vsx_extract_<mode>_p8): Likewise.
15905 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15906 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15907 (vsx_set_<mode>_p9): Likewise.
15908 (vsx_set_v4sf_p9): Likewise.
15909 (vsx_set_v4sf_p9_zero): Likewise.
15910 (vsx_insert_extract_v4sf_p9): Likewise.
15911 (vsx_insert_extract_v4sf_p9_2): Likewise.
15912 * config/rs6000/rs6000.md (sign extend splitter): Change
15913 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15914 (floatsi<mode>2_lfiwax_mem): Likewise.
15915 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15916 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15917 since a test for TARGET_P9_VECTOR was used.
15918 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15919 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15920 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15921 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15922 TARGET_P8_VECTOR test.
15923 (fix_trunc<mode>si2_stfiwx): Likewise.
15924 (fix_trunc<mode>si2_internal): Likewise.
15925 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15926 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15928 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15929 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15930 TARGET_P8_VECTOR test.
15931 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15932 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15933 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15935 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15936 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
15937 since a test for TARGET_P9_VECTOR was used.
15938 (splitter for loading small constants): Likewise.
15940 2017-07-26 Andrew Pinski <apinski@cavium.com>
15942 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
15945 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
15948 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
15949 (fp_valid_at): Likewise.
15951 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15953 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
15954 (qdf24xx_addrcost_table): Likewise.
15955 (cortexa57_tunings): Update to use generic_branch_cost.
15956 (cortexa72_tunings): Likewise.
15957 (cortexa73_tunings): Likewise.
15958 (qdf24xx_tunings): Likewise.
15960 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15962 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
15963 (thunderx2t99_branch_cost): Likewise.
15964 (cortexa35_tunings): Update to use generic_branch_cost.
15965 (cortexa53_tunings): Likewise.
15966 (cortexa57_tunings): Likewise.
15967 (cortexa72_tunings): Likewise.
15968 (cortexa73_tunings): Likewise.
15969 (thunderx2t99_tunings): Likewise.
15971 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15973 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
15974 (sparc_option_override): Honour MASK_FSMULD.
15975 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
15976 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
15977 * config/sparc/sparc.opt (mfsmuld): New option.
15978 * doc/invoke.texi (mfsmuld): Document option.
15980 2017-07-26 Marek Polacek <polacek@redhat.com>
15982 PR middle-end/70992
15983 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15985 2017-07-26 Richard Biener <rguenther@suse.de>
15987 * gimple-match-head.c (do_valueize): Return OP if valueize
15989 (get_def): New helper to get at the def stmt of a SSA name
15990 if valueize allows.
15991 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
15992 do_valueize to get at the def stmt.
15993 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
15995 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
15997 PR middle-end/46932
15998 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16000 2017-07-26 Martin Liska <mliska@suse.cz>
16003 * function.c (expand_function_start): Make expansion of
16004 nonlocal_goto_save_area after parm_birth_insn.
16006 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16008 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16009 from all CPU target flags enable members.
16011 2017-07-26 Richard Biener <rguenther@suse.de>
16013 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16014 (decision_tree::gen): Make 'type' const.
16015 (write_predicate): Likewise.
16017 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16019 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16020 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16021 (rs6000_option_override_internal): Likewise.
16022 (rs6000_expand_vector_set): Likewise.
16023 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16024 (TARGET_UPPER_REGS_SF): Likewise.
16025 (TARGET_UPPER_REGS_DI): Likewise.
16026 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16027 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16028 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16029 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16030 (Splitters for DI constants in Altivec registers): Likewise.
16031 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16032 (vsx_set_v4sf_p9): Likewise.
16033 (vsx_set_v4sf_p9_zero): Likewise.
16034 (vsx_insert_extract_v4sf_p9): Likewise.
16035 (vsx_insert_extract_v4sf_p9_2): Likewise.
16037 2017-07-25 Carl Love <cel@us.ibm.com>
16039 * doc/extend.texi: Update the built-in documentation file for the
16040 existing built-in functions
16041 vector signed char vec_cnttz (vector signed char);
16042 vector unsigned char vec_cnttz (vector unsigned char);
16043 vector signed short vec_cnttz (vector signed short);
16044 vector unsigned short vec_cnttz (vector unsigned short);
16045 vector signed int vec_cnttz (vector signed int);
16046 vector unsigned int vec_cnttz (vector unsigned int);
16047 vector signed long long vec_cnttz (vector signed long long);
16048 vector unsigned long long vec_cnttz (vector unsigned long long);
16050 2017-07-25 Andrew Pinski <apinski@cavium.com>
16052 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16053 accesses where the use is for the first operand of a BIT_INSERT.
16055 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16058 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16059 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16061 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16063 * config/i386/gstabs.h: Delete.
16064 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16066 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16068 * config/i386/i386.c (ix86_decompose_address): Do not check for
16069 register RTX when looking at index_reg or base_reg.
16070 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16072 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16074 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16075 to update EH info here.
16077 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16079 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16081 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16083 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16085 2017-07-25 Torsten Duwe <duwe@suse.de>
16087 * common.opt: Introduce -fpatchable-function-entry
16088 command line option, and its variables function_entry_patch_area_size
16089 and function_entry_patch_area_start.
16090 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16091 including a two-value parser.
16092 * target.def (print_patchable_function_entry): New target hook.
16093 * targhooks.h (default_print_patchable_function_entry): New function.
16094 * targhooks.c (default_print_patchable_function_entry): Likewise.
16095 * toplev.c (process_options): Switch off IPA-RA if
16096 patchable function entries are being generated.
16097 * varasm.c (assemble_start_function): Look at the
16098 patchable-function-entry command line switch and current
16099 function attributes and maybe generate NOP instructions by
16100 calling the print_patchable_function_entry hook.
16101 * doc/extend.texi: Document patchable_function_entry attribute.
16102 * doc/invoke.texi: Document -fpatchable_function_entry
16103 command line option.
16104 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16106 * doc/tm.texi: Re-generate.
16108 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16111 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16112 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16114 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16116 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16118 2017-07-25 Richard Biener <rguenther@suse.de>
16120 PR tree-optimization/81455
16121 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16122 not walk in cycles when looking for guards.
16124 2017-07-25 Richard Biener <rguenther@suse.de>
16126 PR tree-optimization/81529
16127 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16128 when optimizing backedge uses.
16130 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16132 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16134 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16135 to dl_section_ref. On AIX, append an expression to subtract
16136 the size of the section length to dl_section_ref.
16138 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16140 * configure.ac: If any of the config.* scripts fail, exit 1.
16141 * configure: Regenerate.
16143 2017-07-25 Richard Biener <rguenther@suse.de>
16145 PR middle-end/81546
16146 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16147 of immediate uses, be more verbose on errors.
16149 2017-07-25 Richard Biener <rguenther@suse.de>
16151 PR tree-optimization/81510
16152 * tree-vect-loop.c (vect_is_simple_reduction): When the
16153 reduction stmt is not inside the loop bail out.
16155 2017-07-25 Richard Biener <rguenther@suse.de>
16157 PR tree-optimization/81303
16158 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16159 profitability check against LOOP_VINFO_NITERSM1.
16161 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16163 * domwalk.c (cmp_bb_postorder): Simplify.
16164 (sort_bbs_postorder): New function. Use it...
16165 (dom_walker::walk): ...here to optimize common cases.
16167 2017-07-25 Martin Liska <mliska@suse.cz>
16170 * ipa-visibility.c (function_and_variable_visibility): Make the
16171 redirection just on target that supports aliasing.
16172 Fix GNU coding style.
16174 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16177 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16179 * config/arm/rtems.h: Likewise.
16180 * config/bfin/rtems.h: Likewise.
16181 * config/i386/rtemself.h: Likewise.
16182 * config/lm32/rtems.h: Likewise.
16183 * config/m32c/rtems.h: Likewise.
16184 * config/m68k/rtemself.h: Likewise.
16185 * config/microblaze/rtems.h: Likewise.
16186 * config/mips/rtems.h: Likewise.
16187 * config/moxie/rtems.h: Likewise.
16188 * config/nios2/rtems.h: Likewise.
16189 * config/powerpcspe/rtems.h: Likewise.
16190 * config/rs6000/rtems.h: Likewise.
16191 * config/rtems.h: Likewise.
16192 * config/sh/rtems.h: Likewise.
16193 * config/sh/rtemself.h: Likewise.
16194 * config/sparc/rtemself.h: Likewise.
16196 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16199 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16200 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16201 * tree-ssa-structalias.c (alias_get_name): Same.
16203 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16206 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16207 instructions if no du chain is found.
16209 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16211 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16213 2017-07-25 Richard Biener <rguenther@suse.de>
16215 PR middle-end/81505
16216 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16219 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16221 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16222 upper-regs options.
16223 (ISA_2_7_MASKS_SERVER): Likewise.
16224 (ISA_3_0_MASKS_IEEE): Likewise.
16225 (OTHER_P8_VECTOR_MASKS): Likewise.
16226 (OTHER_VSX_VECTOR_MASKS): Likewise.
16227 (POWERPC_MASKS): Likewise.
16228 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16229 duplicate list of options.
16230 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16231 explicit -mupper-regs options.
16232 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16233 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16234 alias for -mupper-regs-df.
16235 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16236 (rs6000_init_hard_regno_mode_ok): Likewise.
16237 (rs6000_option_override_internal): Likewise.
16238 (rs6000_opt_masks): Likewise.
16239 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16240 options in terms of whether -mvsx or -mpower8-vector was used.
16241 (TARGET_UPPER_REGS_DI): Likewise.
16242 (TARGET_UPPER_REGS_SF): Likewise.
16243 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16244 -mupper-regs-* options.
16246 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16248 * passes.c (emergency_dump_function): Print some empty lines and a
16249 header before the RTL dump.
16251 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16253 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16255 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16258 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16259 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16261 2017-07-24 Carl Love <cel@us.ibm.com>
16263 * config/rs6000/rs6000-c.c: Add support for built-in functions
16264 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16265 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16266 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16267 vec_extract_fp_from_shortl): Add defines for the two builtins.
16268 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16269 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16271 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16272 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16273 * doc/extend.texi: Update the built-in documentation file for the
16274 new built-in function.
16276 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16279 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16281 * doc/generic.texi: Likewise.
16282 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16283 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16285 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16287 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16288 (aarch64_mls_elt_merge<mode>): Likewise.
16290 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16292 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16293 having __cxa_atexit.
16295 2017-07-23 Michael Collison <michael.collison@arm.com>
16297 * config/arm/arm.c (arm_option_override): Deprecate
16298 use of -mstructure-size-boundary.
16299 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16300 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16302 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16305 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16306 Reduce cost estimate for direct moves.
16308 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16311 * config/i386/i386.c (ix86_option_override_internal): Disable
16312 BMI, BMI2 and TBM instructions for -m16.
16314 2017-07-21 Carl Love <cel@us.ibm.com>
16316 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16317 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16318 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16319 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16320 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16321 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16322 VMULOSW): New enum "unspec" values.
16323 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16324 altivec_vmulosw): New patterns.
16325 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16326 VMULOSW): Add definitions.
16328 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16330 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16331 (qdf24xx): Likewise.
16332 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16333 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16334 (AARCH64_FL_V8_1): Renumber.
16335 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16336 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16337 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16338 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16339 rdma to feature modifiers list.
16341 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16343 PR middle-end/56727
16344 * ipa-visibility (function_and_variable_visibility): Convert
16345 recursive PLT call to direct call if appropriate.
16347 2017-07-21 Andrew Pinski <apinski@cavium.com>
16349 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16350 operand 1 to see if the types precision matches.
16351 * fold-const.c (operand_equal_p): Likewise.
16353 2017-07-21 Richard Biener <rguenther@suse.de>
16355 PR tree-optimization/81303
16356 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16357 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16358 (vect_peeling_hash_get_lowest_cost): Adjust.
16359 (vect_enhance_data_refs_alignment): Likewise. Use
16360 vect_get_peeling_costs_all_drs to compute the penalty for no
16361 peeling to match up costs.
16363 2017-07-21 Richard Biener <rguenther@suse.de>
16365 PR tree-optimization/81500
16366 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16367 we didn't identify a reduction path.
16369 2017-07-21 Tom de Vries <tom@codesourcery.com>
16370 Cesar Philippidis <cesar@codesourcery.com>
16372 PR gcov-profile/81442
16373 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16376 2017-07-21 Tom de Vries <tom@codesourcery.com>
16379 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16381 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16382 nvptx_override_options_after_change.
16384 2017-07-21 Ulrich Drepper <drepper@redhat.com>
16386 * dwarf2out.c (output_file_names): Avoid double testing for
16387 dwarf_version >= 5.
16389 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
16391 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
16393 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16395 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
16397 (try_crossjump_to_edge): Do not punt on partitioned functions.
16399 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16401 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
16402 Put all BBs reachable only via paths crossing cold region to cold
16404 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
16406 2016-07-21 Richard Biener <rguenther@suse.de>
16408 PR tree-optimization/81303
16409 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
16410 into account prologue and epilogue iterations when raising
16411 min_profitable_iters to sth at least covering one vector iteration.
16413 2017-07-21 Tamar Christina <tamar.christina@arm.com>
16415 * config/arm/arm.c (arm_test_cpu_arch_dat):
16418 2017-07-20 Nathan Sidwell <nathan@acm.org>
16420 Remove TYPE_METHODS.
16421 * tree.h (TYPE_METHODS): Delete.
16422 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
16423 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
16424 (dbxout_type_methods): Scan TYPE_FIELDS.
16425 (dbxout_type): Don't check TYPE_METHODS here.
16426 * function.c (use_register_for_decl): Always ignore register for
16427 class types when not optimizing.
16428 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
16429 * tree.c (free_lang_data_in_type): Stitch out member functions and
16430 templates from TYPE_FIELDS.
16431 (build_distinct_type_copy, verify_type_variant,
16432 verify_type): Member fns are on TYPE_FIELDS.
16433 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
16434 * tree-pretty-print.c (dump_generic_node): Likewise.
16436 2017-07-20 Jakub Jelinek <jakub@redhat.com>
16439 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
16440 V2TImode and V4TImode.
16441 (ix86_expand_vector_extract): Likewise.
16442 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
16443 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
16444 (ssescalarmode): Handle V4TImode and V2TImode.
16445 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
16446 (*vec_extractv2ti, *vec_extractv4ti): New insns.
16447 (VEXTRACTI128_MODE): New mode iterator.
16448 (splitter for *vec_extractv?ti first element): New.
16449 (VEC_INIT_MODE): New mode iterator.
16450 (vec_init<mode>): Consolidate 3 expanders into one using
16451 VEC_INIT_MODE mode iterator.
16453 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16455 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
16456 non_spilled_static_chain_regno_p.
16458 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16460 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
16462 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
16464 * bb-reorder.c (connect_traces): Allow copying of blocks within
16467 2017-07-20 Richard Biener <rguenther@suse.de>
16469 * gimple.h (gimple_phi_result): Add gphi * overload.
16470 (gimple_phi_result_ptr): Likewise.
16471 (gimple_phi_arg): Likewise. Adjust index assert to only
16472 allow actual argument accesses rather than all slots available
16474 (gimple_phi_arg_def): Add gphi * overload.
16475 * tree-phinodes.c (make_phi_node): Initialize only actual
16477 (resize_phi_node): Clear memory not covered by old node,
16478 do not initialize excess argument slots.
16479 (reserve_phi_args_for_new_edge): Initialize new argument slot
16482 2017-07-20 Bin Cheng <bin.cheng@arm.com>
16484 PR tree-optimization/81388
16486 2016-07-21 Bin Cheng <bin.cheng@arm.com>
16488 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
16489 by removing computation of may_be_zero.
16491 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16492 Tom de Vries <tom@codesourcery.com>
16494 PR middle-end/81030
16495 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
16496 when gimple level profile disagrees with what RTL expander did.
16498 2017-07-20 Richard Biener <rguenther@suse.de>
16500 PR tree-optimization/61171
16501 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
16502 (vect_analyze_stmt): Add slp instance parameter.
16503 (vectorizable_reduction): Likewise.
16504 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
16505 (vect_is_simple_reduction): Deal with chains not detected
16506 as SLP reduction chain, specifically not properly associated
16507 chains containing a mix of plus/minus.
16508 (get_reduction_op): Remove.
16509 (get_initial_defs_for_reduction): Simplify, pass in whether
16510 this is a reduction chain, pass in the SLP node for the PHIs.
16511 (vect_create_epilog_for_reduction): Get the SLP instance as
16513 (vectorizable_reduction): Get the SLP instance as arg.
16514 During analysis remember the SLP node with the PHIs in the
16515 instance. Simplify getting at the vectorized reduction PHIs.
16516 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
16517 through SLP instance.
16518 (vect_slp_analyze_operations): Likewise.
16519 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
16520 (vect_transform_stmt): Likewise.
16522 2017-07-20 Tom de Vries <tom@codesourcery.com>
16524 PR tree-optimization/81489
16525 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
16526 read of phi arg location to before loop that modifies phi.
16528 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16530 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
16533 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16535 PR middle-end/81331
16536 * except.c (execute): Fix ordering issue.
16538 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16540 PR rtl-optimization/81423
16541 * combine.c (make_compound_operation_int): Don't try to optimize
16542 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
16544 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16546 PR rtl-optimization/81423
16547 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
16548 with a constant that is -1 in the truncated to mode.
16550 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16552 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
16553 (determine_unlikely_bbs): ... here.
16554 * predict.h (propagate_unlikely_bbs_forward): Declare.
16555 * cfgexpand.c (pass_expand::execute): Use it.
16556 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
16558 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
16559 propagate_unlikely_bbs_forward.
16561 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16563 PR middle-end/81331
16564 * except.c (maybe_add_nop_after_section_switch): New function.
16567 2017-07-19 Tom de Vries <tom@codesourcery.com>
16569 * gimple.h (gimple_phi_set_arg): Make assert more strict.
16571 2017-07-19 Tom de Vries <tom@codesourcery.com>
16573 * gimple.h (gimple_phi_arg): Make assert more strict.
16575 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
16577 * config.gcc (powerpc*-*-*): Add mmintrin.h.
16578 * config/rs6000/mmintrin.h: New file.
16579 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
16581 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16583 PR tree-optimization/81346
16584 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
16586 2017-07-19 Tom de Vries <tom@codesourcery.com>
16588 * config/nvptx/nvptx.md (VECIM): Add V2DI.
16590 2017-07-19 Tom de Vries <tom@codesourcery.com>
16592 * config/nvptx/nvptx-modes.def: Add V2DImode.
16593 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
16594 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
16595 (nvptx_output_mov_insn): Handle lack of mov.b128.
16596 (nvptx_print_operand): Handle 'H' and 'L' codes.
16597 (nvptx_vector_mode_supported): Allow V2DImode.
16598 (nvptx_preferred_simd_mode): New function.
16599 (nvptx_data_alignment): New function.
16600 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
16601 nvptx_preferred_simd_mode.
16602 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
16604 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
16606 2017-07-19 Tom de Vries <tom@codesourcery.com>
16608 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
16609 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
16610 (nvptx_vector_mode_supported): New function. Allow V2SImode.
16611 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
16612 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
16613 (mov<VECIM>_insn): New define_insn.
16614 (define_expand "mov<VECIM>): New define_expand.
16616 2017-07-19 Tom de Vries <tom@codesourcery.com>
16618 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
16620 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16622 PR tree-optimization/81346
16623 * fold-const.h (fold_div_compare, range_check_type): Declare.
16624 * fold-const.c (range_check_type): New function.
16625 (build_range_check): Use range_check_type.
16626 (fold_div_compare): No longer static, rewritten into
16627 a match.pd helper function.
16628 (fold_comparison): Don't call fold_div_compare here.
16629 * match.pd (X / C1 op C2): New optimization using fold_div_compare
16630 as helper function.
16632 2017-07-19 Nathan Sidwell <nathan@acm.org>
16634 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
16635 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
16636 * tree.c (find_decls_types_r, verify_type): Use
16637 TYPE_{MIN,MAX}_VALUE_RAW.
16638 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
16639 (hash_tree): Likewise.
16640 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16642 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16645 2017-07-18 Tom de Vries <tom@codesourcery.com>
16647 PR middle-end/81464
16648 * omp-expand.c (expand_omp_for_static_chunk): Handle
16649 equal-argument loop exit phi.
16651 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
16654 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
16655 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
16656 operand 2 predicate.
16657 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
16658 operand 2 predicate.
16659 (ror,rol -> rorx splitters): Use const_int_operand as
16660 operand 2 predicate.
16662 2017-06-18 Richard Biener <rguenther@suse.de>
16664 PR tree-optimization/81410
16665 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
16666 the gap in the ! slp_perm SLP case after each group.
16668 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16670 PR middle-end/81463
16671 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
16674 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16676 PR middle-end/81462
16677 * predict.c (set_even_probabilities): Cleanup; do not affect
16678 probabilities that are already known.
16679 (combine_predictions_for_bb): Call even when count is set.
16681 2017-07-18 Nathan Sidwell <nathan@acm.org>
16683 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
16686 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16689 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
16690 optimization for loop niter analysis.
16692 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
16695 * config/avr/avr.c (avr_optimize_casesi): Don't use
16696 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
16698 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
16700 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
16701 body_cost_vec from _vect_peel_extended_info.
16702 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
16703 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
16706 2017-07-18 Bin Cheng <bin.cheng@arm.com>
16708 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
16710 2017-07-18 Richard Biener <rguenther@suse.de>
16712 PR tree-optimization/80620
16713 PR tree-optimization/81403
16714 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
16715 info when re-using a VN table entry.
16717 2017-07-18 Richard Biener <rguenther@suse.de>
16719 PR tree-optimization/81418
16720 * tree-vect-loop.c (vectorizable_reduction): Properly compute
16721 vectype_in. Verify that with lane-reducing reduction operations
16722 we have a single def-use cycle.
16724 2017-07-17 Carl Love <cel@us.ibm.com>
16726 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
16728 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16729 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16730 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16731 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16732 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16733 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16734 VMULOSW): New enum "unspec" values.
16735 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
16736 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
16737 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16738 altivec_vmulosw): New patterns.
16739 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16740 VMULOSW): Add definitions.
16741 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
16743 * config/alpha/alpha.c: Include predict.h.
16745 2017-07-17 Yury Gribov <tetra2005@gmail.com>
16747 * tree-vrp.c (compare_assert_loc): Fix comparison function
16748 to return predictable results.
16750 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16752 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
16754 (subdi3): Likewise.
16755 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
16756 * doc/invoke.texi (mexpand-adddi): Update text.
16758 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16760 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
16761 that also clobbers the CC register. The old expand code is moved
16763 (*arc_clzsi2): ... here.
16764 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
16765 the CC register. The old expand code is moved to ...
16766 (arc_ctzsi2): ... here.
16768 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16770 * config/arc/arc.opt (mindexed-loads): Use initial value
16771 TARGET_INDEXED_LOADS_DEFAULT.
16772 (mauto-modify-reg): Use initial value
16773 TARGET_AUTO_MODIFY_REG_DEFAULT.
16774 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16775 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16776 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
16777 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
16779 2017-07-17 Martin Liska <mliska@suse.cz>
16782 * opts.c (finish_options): Do not allow -fgnu-tm
16783 w/ -fsanitize={kernel-,}address. Say sorry.
16785 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16788 * tree-loop-distribution.c (classify_partition): Only assert on
16789 numer of iterations.
16790 (merge_dep_scc_partitions): Delete prameter. Update function call.
16791 (distribute_loop): Remove code handling loop with unknown niters.
16792 (pass_loop_distribution::execute): Skip loop with unknown niters.
16794 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16797 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
16798 function sort_partitions_by_post_order.
16800 2017-07-17 Bin Cheng <bin.cheng@arm.com>
16802 PR tree-optimization/81374
16803 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
16804 the max index of basic blocks, rather than number of basic blocks.
16806 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16808 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
16810 (arc_legitimate_pic_operand_p): Likewise.
16811 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
16813 (arc_needs_pcl_p): Likewise.
16814 (arc_legitimate_pc_offset_p): Likewise.
16815 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
16816 function is also used in constrains.md.
16817 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
16818 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
16819 PLUS. Only return true/false in known cases, otherwise assert.
16820 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
16821 is already called in arc_legitimate_constant_p.
16822 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
16824 (LEGITIMATE_PIC_OPERAND_P): Use
16825 arc_raw_symbolic_reference_mentioned_p function.
16826 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
16831 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16832 Andrew Burgess <andrew.burgess@embecosm.com>
16834 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16835 (arc_return_address_register): New function.
16836 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16837 (arc_handle_fndecl_attribute): Add naked attribute.
16838 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16839 (TARGET_WARN_FUNC_RETURN): Likewise.
16840 (arc_allocate_stack_slots_for_args): New function.
16841 (arc_warn_func_return): Likewise.
16842 (machine_function): Change type fn_type.
16843 (arc_compute_function_type): Consider new naked function type,
16844 change function return type.
16845 (arc_must_save_register): Adapt to handle new
16846 arc_compute_function_type's return type.
16847 (arc_expand_prologue): Likewise.
16848 (arc_expand_epilogue): Likewise.
16849 (arc_return_address_regs): Delete.
16850 (arc_return_address_register): New function.
16851 (arc_epilogue_uses): Use above function.
16852 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16853 (arc_function_type): Change encoding, add naked type.
16854 (ARC_INTERRUPT_P): Change to handle the new encoding.
16855 (ARC_FAST_INTERRUPT_P): Likewise.
16856 (ARC_NORMAL_P): Define.
16857 (ARC_NAKED_P): Likewise.
16858 (arc_compute_function_type): Delete prototype.
16859 * config/arc/arc.md (in_ret_delay_slot): Use
16860 arc_return_address_register function.
16861 (simple_return): Likewise.
16862 (p_return_i): Likewise.
16864 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16866 PR tree-optimization/81428
16867 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16868 can't be built for those types.
16870 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16872 Remove stuff dead since r239246.
16874 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16875 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16876 (avr_inform_devices): Remove dead stuff.
16878 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16880 * config/arm/arm_neon.h: Fix softp typo.
16882 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16884 PR tree-optimization/81365
16885 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16886 aggregate moves onto bb predecessor edges, make sure there are no
16887 loads that could alias the lhs in between the start of bb and the
16890 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16893 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
16894 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
16895 [LSHIFTRT, outer_code = TRUNCATE]: Same.
16897 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16899 PR tree-optimization/81396
16900 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16901 (init_symbolic_number): Initialize it to 1.
16902 (perform_symbolic_merge): Add n_ops from both operands into the new
16904 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16905 without base_addr as useless if they need more than one operation.
16906 (bswap_replace): Handle !bswap case for NULL base_addr.
16908 2017-07-17 Tom de Vries <tom@codesourcery.com>
16911 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16914 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16916 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16917 conditional builtin define __FIX_LEON3FT_B2BST.
16919 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16921 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16922 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16925 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16927 PR rtl-optimization/81424
16928 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16929 to remove potential trapping from operands if -fnon-call-exceptions.
16931 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16933 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16934 profile_proability for scalling.
16935 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
16937 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16939 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
16941 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16943 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
16944 fixpoint arithmetics.
16946 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16948 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
16949 fixpoint arithmetics.
16951 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16953 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
16954 fixpoint arithmetics.
16956 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16958 * profile-count.h (profile_probability::from_reg_br_prob_note,
16959 profile_probability::to_reg_br_prob_note): New functions.
16960 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
16961 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
16962 * predict.c (probability_reliable_p): Update.
16963 (edge_probability_reliable_p): Update.
16964 (br_prob_note_reliable_p): Update.
16965 (invert_br_probabilities): Update.
16966 (add_reg_br_prob_note): New function.
16967 (combine_predictions_for_insn): Update.
16968 * asan.c (asan_clear_shadow): Update.
16969 * cfgbuild.c (compute_outgoing_frequencies): Update.
16970 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16971 (update_br_prob_note): Update.
16972 (rtl_verify_edges): Update.
16973 (purge_dead_edges): Update.
16974 (fixup_reorder_chain): Update.
16975 * emit-rtl.c (try_split): Update.
16976 * ifcvt.c (cond_exec_process_insns): Update.
16977 (cond_exec_process_if_block): Update.
16978 (dead_or_predicable): Update.
16979 * internal-fn.c (expand_addsub_overflow): Update.
16980 (expand_neg_overflow): Update.
16981 (expand_mul_overflow): Update.
16982 * loop-doloop.c (doloop_modify): Update.
16983 * loop-unroll.c (compare_and_jump_seq): Update.
16984 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16985 * predict.h: Update.
16986 * reorg.c (mostly_true_jump): Update.
16988 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
16989 * config/alpha/alpha.c (emit_unlikely_jump): Update.
16990 * config/arc/arc.c: (emit_unlikely_jump): Update.
16991 * config/arm/arm.c: (emit_unlikely_jump): Update.
16992 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
16993 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
16994 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
16995 (ix86_print_operand): Update.
16996 (ix86_split_fp_branch): Update.
16997 (predict_jump): Update.
16998 * config/ia64/ia64.c (ia64_print_operand): Update.
16999 * config/mmix/mmix.c (mmix_print_operand): Update.
17000 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17001 (rs6000_expand_split_stack_prologue): Update.
17002 * config/rs6000/rs6000.c: Update.
17003 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17004 (s390_expand_vec_movstr): Update.
17005 (s390_expand_cs_tdsi): Update.
17006 (s390_expand_split_stack_prologue): Update.
17007 * config/sh/sh.c (sh_print_operand): Update.
17008 (expand_cbranchsi4): Update.
17009 (expand_cbranchdi4): Update.
17010 * config/sparc/sparc.c (output_v9branch): Update.
17011 * config/spu/spu.c (get_branch_target): Update.
17012 (ea_load_store_inline): Update.
17013 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17014 * config/tilepro/tilepro.c: Update.
17016 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17018 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17019 (gimplify_save_expr): Likewise.
17021 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17023 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17025 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17027 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17029 * timevar.def (TV_IPA_FNSUMMARY): Define.
17031 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17033 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17034 to back store errata sensitive sequence from being generated.
17035 (sqrtdf2_fix): Likewise.
17037 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17039 * tree-ssa-threadupdate.c (compute_path_counts,
17040 update_joiner_offpath_counts): Use profile_probability.
17042 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17045 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17047 * config/arm/arm-c.c (arm_cpu_builtins): Define
17048 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17050 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17052 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17053 array entries to represent __ieee128 versions of the
17054 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17055 scalar_extract_sig, and scalar_insert_exp built-in functions.
17056 (altivec_resolve_overloaded_builtin): Add special case handling
17057 for the __builtin_scalar_insert_exp function, as represented by
17058 the P9V_BUILTIN_VEC_VSIEDP constant.
17059 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17060 exponent support for __ieee128 argument.
17061 (VSESQP): Add scalar extract signature support for __ieee128
17063 (VSTDCNQP): Add scalar test negative support for __ieee128
17065 (VSIEQP): Add scalar insert exponent support for __int128 argument
17066 with __ieee128 result.
17067 (VSIEQPF): Add scalar insert exponent support for __ieee128
17068 argument with __ieee128 result.
17069 (VSTDCQP): Add scalar test data class support for __ieee128
17071 (VSTDCNQP): Add overload support for scalar test negative with
17072 __ieee128 argument.
17073 (VSTDCQP): Add overload support for scalar test data class
17074 __ieee128 argument.
17075 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17076 UNSPEC_VSX_SXSIGDP.
17077 (UNSPEC_VSX_SIEXPQP): New constant.
17078 (xsxexpqp): New insn for VSX scalar extract exponent quad
17080 (xsxsigqp): New insn for VSX scalar extract significand quad
17082 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17083 precision with floating point argument.
17084 (xststdcqp): New expand for VSX scalar test data class quad
17086 (xststdcnegqp): New expand for VSX scalar test negative quad
17088 (xststdcqp): New insn to match expansions for VSX scalar test data
17089 class quad precision and VSX scalar test negative quad precision.
17090 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17091 special case operand checking to enforce that second operand of
17092 VSX scalar test data class with quad precision argument is a 7-bit
17094 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17095 prototypes and descriptions of __ieee128 versions of
17096 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17097 scalar_test_data_class, and scalar_test_neg built-in functions.
17099 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17101 PR tree-optimization/81162
17102 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17103 replace a negate with an add.
17105 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17107 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17109 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17111 * config/arm/arm-c.c (arm_cpu_builtins): Define
17112 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17114 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17116 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17117 (armv8-r): Set ARM Cortex-R52 as default CPU.
17118 * config/arm/arm-tables.opt: Regenerate.
17119 * config/arm/arm-tune.md: Regenerate.
17120 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17122 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17123 extension for -mcpu=cortex-r52.
17125 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17127 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17128 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17129 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17130 (fp-armv8): Define it as FP_ARMv8 only.
17131 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17132 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17134 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17135 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17136 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17137 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17138 than TARGET_FPU_ARMV8.
17139 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17140 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17141 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17143 * config/arm/neon.md (neon_vrint): Likewise.
17144 (neon_vcvt): Likewise.
17145 (neon_<fmaxmin_op><mode>): Likewise.
17146 (<fmaxmin><mode>3): Likewise.
17147 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17148 * config/arm/predicates.md (arm_cond_move_operator): Check against
17149 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17151 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17153 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17154 to top of function.
17156 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17158 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17159 loop in comment with memset.
17161 2017-07-14 Martin Liska <mliska@suse.cz>
17163 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17164 * dwarf2out.c (is_java): Remove the function.
17165 (output_pubname): Remove usage of the function.
17166 (lower_bound_default): Remove usage of DW_LANG_Java.
17167 (gen_compile_unit_die): Likewise.
17168 * gcc.c: Remove compiler defaults for .java and .zip files.
17169 * gimple-expr.c (remove_suffix): Change as there's no longer
17170 extension than 4-letter one.
17171 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17172 (gimplify_save_expr): Likewise.
17173 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17174 as it's possible even for other languages than Java.
17175 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17176 * lto-opts.c (lto_write_options): Remove reference to Java.
17177 * opts.c (strip_off_ending): Update file extension handling.
17178 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17179 * tree-eh.c (lower_resx): Likewise.
17180 * tree.c (free_lang_data_in_type): Remove dead code.
17181 (find_decls_types_r): Likewise.
17182 (build_common_builtin_nodes): Remove Java from a comment.
17183 (verify_type): Remove dead code.
17184 * varasm.c (assemble_external): Remove Java from a comment.
17186 2017-07-14 Martin Liska <mliska@suse.cz>
17188 * opts.c (finish_options): Add quotes.
17189 (common_handle_option): Likewise.
17191 2017-07-14 Martin Liska <mliska@suse.cz>
17193 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17194 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17195 Remove N_SO_PASCAL.
17196 * dwarf2out.c (lower_bound_default): Do not handle
17198 (gen_compile_unit_die): Likewise.
17199 * gcc.c: Remove default extension binding for GNU Pascal.
17200 * stmt.c: Remove Pascal language from a comment.
17201 * xcoffout.c: Likewise.
17203 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17206 * diagnostic-show-locus.c (fixit_cmp): New function.
17207 (layout::layout): Sort m_fixit_hints.
17208 (column_range::column_range): Assert that the values are valid.
17209 (struct char_span): New struct.
17210 (correction::overwrite): New method.
17211 (struct source_line): New struct.
17212 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17213 calls in terms of classes source_line and char_span, and
17214 correction::overwrite.
17215 (selftest::test_overlapped_fixit_printing_2): New function.
17216 (selftest::diagnostic_show_locus_c_tests): Call it.
17218 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17220 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17221 early if there is no lhs.
17223 2017-07-13 Martin Liska <mliska@suse.cz>
17225 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17226 (gen_reference_type_die): Likewise.
17227 * stor-layout.c: Remove Pascal-related comment.
17229 2017-07-13 Martin Liska <mliska@suse.cz>
17231 * opts.c (finish_options): Add quotes to error messages.
17232 (parse_sanitizer_options): Likewise.
17234 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17236 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17238 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17240 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17242 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17244 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17246 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17248 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17251 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17252 provides the hardware capability bits, define the macro
17253 __BUILTIN_CPU_SUPPORTS__.
17254 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17255 if GLIBC does not provide the hardware capability bits. Add a
17256 gcc_unreachable call if the built-in cpu function is neither
17257 __builtin_cpu_is nor __builtin_cpu_supports.
17258 (rs6000_get_function_versions_dispatcher): Change the warning
17259 that an old GLIBC is used which does not export the capability
17260 bits to be an error.
17261 * doc/extend.texi (target_clones attribute): Document the
17262 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17263 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17264 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17265 the macros defined by GCC if the newer GLIBC is available.
17267 2017-07-12 Jeff Law <law@redhat.com>
17269 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17270 remaining includes slightly.
17271 * config/riscv/riscv-builtins.c: Include profile-count.h.
17273 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17276 * config/avr/avr.c (avr_set_current_function): In diagnostic
17277 messages: Quote keywords and (parts of) identifiers.
17278 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17281 2017-07-12 Carl Love <cel@us.ibm.com>
17283 * config/rs6000/rs6000-c.c: Add support for built-in functions
17284 vector bool char vec_revb (vector bool char);
17285 vector bool short vec_revb (vector short char);
17286 vector bool int vec_revb (vector bool int);
17287 vector bool long long vec_revb (vector bool long long);
17288 * doc/extend.texi: Update the built-in documentation file for the
17289 new built-in functions.
17291 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17293 * config/s390/s390.md: Remove movcc splitter.
17295 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17297 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17298 load/store on condition.
17300 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17303 * config/avr/avr.c (avr_encode_section_info)
17304 [progmem && !TREE_READONLY]: Error if progmem object needs
17307 2017-07-11 Michael Collison <michael.collison@arm.com>
17309 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17312 2017-07-11 Carl Love <cel@us.ibm.com>
17314 * config/rs6000/rs6000-c.c: Add support for builtins
17315 vector unsigned int vec_parity_lsbb (vector signed int);
17316 vector unsigned int vec_parity_lsbb (vector unsigned int);
17317 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17318 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17319 vector unsigned long long vec_parity_lsbb (vector signed long long);
17320 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17321 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17322 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17323 * doc/extend.texi: Update the built-in documentation file for the
17324 new built-in functions.
17326 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17328 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17329 (layout::m_primary_loc): New field.
17330 (layout::layout): Initialize new field. Move location filtering
17331 logic from here to...
17332 (layout::maybe_add_location_range): ...this new method. Add
17333 support for filtering to just the lines already specified by other
17335 (layout::will_show_line_p): New method.
17336 (gcc_rich_location::add_location_if_nearby): New method.
17337 (selftest::test_add_location_if_nearby): New test function.
17338 (selftest::diagnostic_show_locus_c_tests): Call it.
17339 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17342 2017-07-11 Tom de Vries <tom@codesourcery.com>
17344 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17345 (bb_first_real_insn): New function.
17346 (nvptx_single): Add extra initialization of broadcasted condition
17349 2017-07-11 Nathan Sidwell <nathan@acm.org>
17351 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17353 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17355 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17356 Binutils doc as TEXI will mess them up.
17357 * doc/invoke.texi (AVR Options): Same here.
17359 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17361 * config/sparc/sparc.opt (mfix-ut700): New option.
17362 (mfix-gr712rc): Likewise.
17363 (sparc_fix_b2bst): New variable.
17364 * doc/invoke.texi (SPARC options): Document them.
17365 (ARM options): Fix warnings.
17366 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17367 instructions to prevent sequences that can trigger the store-store
17368 errata for certain LEON3FT processors.
17369 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17370 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17371 * config/sparc/sparc.md (fix_b2bst): New attribute.
17372 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17374 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17377 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17379 (*rsqrtsf2_sse): Ditto.
17381 (div<mode>3): Macroize insn from divdf3 and divsf3
17382 using MODEF mode iterator.
17384 2017-07-10 Martin Sebor <msebor@redhat.com>
17386 PR tree-optimization/80397
17387 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
17388 instead of testing for equality to INTEGER_TYPE.
17390 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
17392 * config.gcc: Remove uclibc from arc target spec.
17394 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
17396 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
17398 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17401 * lto-wrapper.c (remove_option): New function.
17402 (merge_and_complain): Merge PIC/PIE options more realistically.
17404 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17406 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
17410 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
17411 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
17412 * config.in: Regenerate.
17413 * configure: Regenerate.
17414 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
17415 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
17416 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
17417 (TARGET_GASISR_PROLOGUES): ...target mask.
17418 * common/config/avr/avr-common.c
17419 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
17420 Set -mgas-isr-prologues.
17421 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
17422 INSERT_PASS_BEFORE for it.
17423 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
17424 * config/avr/avr.c (avr_option_override)
17425 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
17426 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
17427 (avr_attribute_table) <no_gccisr>: Add new function attribute.
17428 (avr_set_current_function) <is_no_gccisr>: Init machine field.
17429 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
17431 (make_avr_pass_pre_proep): New function.
17432 (emit_push_sfr) <treg>: Add argument to function and use it
17433 instead of TMP_REG.
17434 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
17435 and set machine->gasisr.yes.
17436 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
17437 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
17438 __gcc_isr.n_pushed to .L__stack_usage.
17439 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
17440 (avr_asm_final_postscan_insn): ...this new static function.
17441 * config/avr/avr.h (machine_function)
17442 <is_no_gccisr, use_L__stack_usage>: New fields.
17443 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
17444 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
17445 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
17446 (gasisr, *gasisr): New expander and insn.
17447 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17448 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
17449 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
17451 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
17453 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
17456 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17458 Move jump-tables out of .text again.
17461 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
17462 (ASM_OUTPUT_ADDR_VEC): New function.
17463 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
17464 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
17465 INSN_ADDRESSes as asm comment.
17466 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
17467 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
17468 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
17469 * config/avr/avr.md (*tablejump): Adjust comment.
17470 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
17471 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
17473 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
17474 (avr_output_addr_vec): New proto.
17475 (avr_log_t) <insn_addresses>: New field.
17477 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
17480 * config/i386/i386.c (ix86_function_arg_advance): Set
17481 outgoing_args_on_stack to true if there are outgoing arguments
17483 (ix86_function_arg): Likewise.
17484 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
17485 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
17486 * config/i386/i386.h (machine_function): Add
17487 outgoing_args_on_stack.
17489 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
17491 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17492 supporting pthreds.
17493 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
17495 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
17497 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
17498 (REAL_H): Remove $(MACHMODE_H).
17499 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
17501 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
17502 $(MACHMODE_H) and double-int.h.
17503 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
17505 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
17508 2017-07-07 Andrew Pinski <apinski@cavium.com>
17510 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
17511 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
17513 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17515 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
17516 Add warning if GCC was not configured to link against a GLIBC that
17517 exports the hardware capability bits.
17518 (make_resolver_func): Make resolver function private and not a
17519 COMDAT function. Create the name with clone_function_name instead
17520 of make_unique_name.
17523 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
17524 correct operand in doing the split.
17526 2017-07-07 Carl Love <cel@us.ibm.com>
17528 * config/rs6000/rs6000-c: Add support for built-in function
17529 vector unsigned short vec_pack_to_short_fp32 (vector float,
17531 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
17532 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
17533 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
17534 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
17535 (convert_4f32_8i16): Add define_expand.
17536 * doc/extend.texi: Update the built-in documentation file for the
17537 new built-in function.
17539 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17541 * config/sparc/m8.md: New file.
17542 * config/sparc/sparc.md: Include m8.md.
17544 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17546 * config/sparc/sparc.opt: New option -mvis4b.
17547 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
17548 (sparc_option_override): Handle VIS4B.
17549 (enum sparc_builtins): Define
17550 SPARC_BUILTIN_DICTUNPACK{8,16,32},
17551 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
17552 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
17553 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
17554 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
17555 (check_constant_argument): New function.
17556 (sparc_vis_init_builtins): Define builtins
17557 __builtin_vis_dictunpack{8,16,32},
17558 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
17559 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
17560 __builtin_vis_fpcmpde{8,16,32}shl and
17561 __builtin_vis_fpcmpur{8,16,32}shl.
17562 (sparc_expand_builtin): Check that the constant operands to
17563 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
17564 constant and in range.
17565 * config/sparc/sparc-c.c (sparc_target_macros): Handle
17567 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
17568 (SPARC_IMM5_P): Likewise.
17569 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
17570 (enabled): Handle vis4b.
17571 (UNSPEC_DICTUNPACK): New unspec.
17572 (UNSPEC_FPCMPSHL): Likewise.
17573 (UNSPEC_FPUCMPSHL): Likewise.
17574 (UNSPEC_FPCMPDESHL): Likewise.
17575 (UNSPEC_FPCMPURSHL): Likewise.
17576 (cpu_feature): New CPU feature `vis4b'.
17577 (dictunpack{8,16,32}): New insns.
17578 (FPCSMODE): New mode iterator.
17579 (fpcscond): New code iterator.
17580 (fpcsucond): Likewise.
17581 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
17582 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
17583 (fpcmpde{8,16,32}{si,di}shl): Likewise.
17584 (fpcmpur{8,16,32}{si,di}shl): Likewise.
17585 * config/sparc/constraints.md: Define constraints `q' for unsigned
17586 2-bit integer constants and `t' for unsigned 5-bit integer
17588 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
17590 (imm5_operand_dictunpack16): Likewise.
17591 (imm5_operand_dictunpack32): Likewise.
17592 (imm2_operand): Likewise.
17593 * doc/invoke.texi (SPARC Options): Document -mvis4b.
17594 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
17595 ditunpack* and fpcmp*shl builtins.
17597 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17599 * config.gcc: Handle m8 in --with-{cpu,tune} options.
17600 * config.in: Add HAVE_AS_SPARC6 define.
17601 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
17603 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
17605 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
17606 (CPP_CPU_SPEC): Handle m8.
17607 (ASM_CPU_SPEC): Likewise.
17608 * config/sparc/sparc-opts.h (enum processor_type): Add
17610 * config/sparc/sparc.c (m8_costs): New struct.
17611 (sparc_option_override): Handle TARGET_CPU_m8.
17612 (sparc32_initialize_trampoline): Likewise.
17613 (sparc64_initialize_trampoline): Likewise.
17614 (sparc_issue_rate): Likewise.
17615 (sparc_register_move_cost): Likewise.
17616 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
17617 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
17618 (ASM_CPU64_DEFAULT_SPEC): Likewise.
17619 (CPP_CPU_SPEC): Handle M8.
17620 (ASM_CPU_SPEC): Likewise.
17621 (AS_M8_FLAG): Define.
17622 * config/sparc/sparc.md: Add m8 to the cpu attribute.
17623 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
17624 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
17626 * configure: Regenerate.
17627 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
17630 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17632 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
17634 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
17635 ("*movdi_insn_sp32"): Do not set v3pipe.
17636 ("*movsi_insn"): Likewise.
17637 ("*movdi_insn_sp64"): Likewise.
17638 ("*movsf_insn"): Likewise.
17639 ("*movdf_insn_sp32"): Likewise.
17640 ("*movdf_insn_sp64"): Likewise.
17641 ("*zero_extendsidi2_insn_sp64"): Likewise.
17642 ("*sign_extendsidi2_insn"): Likewise.
17643 ("*mov<VM32:mode>_insn"): Likewise.
17644 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17645 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17646 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17647 ("<vlop:code><VL:mode>3"): Likewise.
17648 ("*not_<vlop:code><VL:mode>3"): Likewise.
17649 ("*nand<VL:mode>_vis"): Likewise.
17650 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
17651 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
17652 ("one_cmpl<VL:mode>2"): Likewise.
17653 ("faligndata<VM64:mode>_vis"): Likewise.
17654 ("alignaddrsi_vis"): Likewise.
17655 ("alignaddrdi_vis"): Likweise.
17656 ("alignaddrlsi_vis"): Likewise.
17657 ("alignaddrldi_vis"): Likewise.
17658 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17659 ("bmaskdi_vis"): Likewise.
17660 ("bmasksi_vis"): Likewise.
17661 ("bshuffle<VM64:mode>_vis"): Likewise.
17662 ("cmask8<P:mode>_vis"): Likewise.
17663 ("cmask16<P:mode>_vis"): Likewise.
17664 ("cmask32<P:mode>_vis"): Likewise.
17665 ("pdistn<P:mode>_vis"): Likewise.
17666 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17668 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17670 * config/sparc/sparc.md ("subtype"): New insn attribute.
17671 ("*wrgsr_sp64"): Set insn subtype.
17672 ("*rdgsr_sp64"): Likewise.
17673 ("alignaddrsi_vis"): Likewise.
17674 ("alignaddrdi_vis"): Likewise.
17675 ("alignaddrlsi_vis"): Likewise.
17676 ("alignaddrldi_vis"): Likewise.
17677 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17678 ("fexpand_vis"): Likewise.
17679 ("fpmerge_vis"): Likewise.
17680 ("faligndata<VM64:mode>_vis"): Likewise.
17681 ("bshuffle<VM64:mode>_vis"): Likewise.
17682 ("cmask8<P:mode>_vis"): Likewise.
17683 ("cmask16<P:mode>_vis"): Likewise.
17684 ("cmask32<P:mode>_vis"): Likewise.
17685 ("fchksm16_vis"): Likewise.
17686 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
17687 ("fmean16_vis"): Likewise.
17688 ("fp<plusminus_insn>64_vis"): Likewise.
17689 ("<plusminus_insn>v8qi3"): Likewise.
17690 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
17691 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
17692 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
17693 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
17694 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
17695 ("*movqi_insn"): Likewise.
17696 ("*movhi_insn"): Likewise.
17697 ("*movsi_insn"): Likewise.
17698 ("movsi_pic_gotdata_op"): Likewise.
17699 ("*movdi_insn_sp32"): Likewise.
17700 ("*movdi_insn_sp64"): Likewise.
17701 ("movdi_pic_gotdata_op"): Likewise.
17702 ("*movsf_insn"): Likewise.
17703 ("*movdf_insn_sp32"): Likewise.
17704 ("*movdf_insn_sp64"): Likewise.
17705 ("*zero_extendhisi2_insn"): Likewise.
17706 ("*zero_extendqihi2_insn"): Likewise.
17707 ("*zero_extendqisi2_insn"): Likewise.
17708 ("*zero_extendqidi2_insn"): Likewise.
17709 ("*zero_extendhidi2_insn"): Likewise.
17710 ("*zero_extendsidi2_insn_sp64"): Likewise.
17711 ("ldfsr"): Likewise.
17712 ("prefetch_64"): Likewise.
17713 ("prefetch_32"): Likewise.
17714 ("tie_ld32"): Likewise.
17715 ("tie_ld64"): Likewise.
17716 ("*tldo_ldub_sp32"): Likewise.
17717 ("*tldo_ldub1_sp32"): Likewise.
17718 ("*tldo_ldub2_sp32"): Likewise.
17719 ("*tldo_ldub_sp64"): Likewise.
17720 ("*tldo_ldub1_sp64"): Likewise.
17721 ("*tldo_ldub2_sp64"): Likewise.
17722 ("*tldo_ldub3_sp64"): Likewise.
17723 ("*tldo_lduh_sp32"): Likewise.
17724 ("*tldo_lduh1_sp32"): Likewise.
17725 ("*tldo_lduh_sp64"): Likewise.
17726 ("*tldo_lduh1_sp64"): Likewise.
17727 ("*tldo_lduh2_sp64"): Likewise.
17728 ("*tldo_lduw_sp32"): Likewise.
17729 ("*tldo_lduw_sp64"): Likewise.
17730 ("*tldo_lduw1_sp64"): Likewise.
17731 ("*tldo_ldx_sp64"): Likewise.
17732 ("*mov<VM32:mode>_insn"): Likewise.
17733 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17734 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17736 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17738 * config/sparc/sparc.md ("type"): New insn type viscmp.
17739 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
17741 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
17742 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
17743 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
17744 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
17746 ("n7_vis_logical_11cycle"): Likewise.
17747 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
17748 * config/sparc/niagara2.md ("niag3_vis": Likewise.
17749 * config/sparc/niagara.md ("niag_vis"): Likewise.
17750 * config/sparc/ultra3.md ("us3_fga"): Likewise.
17751 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
17753 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17755 * config/sparc/sparc.md: New instruction type `bmask'.
17756 (bmaskdi_vis): Use the `bmask' type.
17757 (bmasksi_vis): Likewise.
17758 * config/sparc/ultra3.md (us3_array): Likewise.
17759 * config/sparc/niagara7.md (n7_array): Likewise.
17760 * config/sparc/niagara4.md (n4_array): Likewise.
17761 * config/sparc/niagara2.md (niag2_vis): Likewise.
17762 (niag3_vis): Likewise.
17763 * config/sparc/niagara.md (niag_vis): Likewise.
17765 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17767 * ipa-comdats.c: Remove optimize check from gate.
17768 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
17769 for functions not optimized.
17770 (ipa_fn_summary_read): Skip optimize check.
17771 (ipa_fn_summary_write): Likewise.
17772 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
17774 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
17776 (can_inline_edge_p): Check flag_pcc_struct_return for match.
17777 (check_callers): Give up on caller which is not optimized.
17778 (inline_small_functions): Likewise.
17779 (ipa_inline): Do not give up when not optimizing.
17780 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
17781 away unoptimizes cdtors.
17782 (whole_program_function_and_variable_visibility): Do
17783 ipa_discover_readonly_nonaddressable_vars in LTO mode.
17784 * ipa.c (process_references): Do not check optimize.
17785 (symbol_table::remove_unreachable_nodes): Update optimize check.
17786 (set_writeonly_bit): Update optimize check.
17787 (pass_ipa_cdtor_merge::gate): Do not check optimize.
17788 (pass_ipa_single_use::gate): Remove.
17790 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17792 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
17793 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
17794 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
17795 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
17796 permute_load, permute_store, adjust_extract, adjust_splat,
17797 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
17798 replace_swap_with_copy, dump_swap_insn_table,
17799 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
17800 recombine_lvx_pattern, recombine_stvx_pattern,
17801 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
17802 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
17803 to file rs6000-p8swap.c.
17804 * config/rs6000/rs6000-p8swap.c: New file.
17805 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
17806 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
17807 and rs6000*-*-* targets.
17809 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17811 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
17813 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17815 * lto-wrapper.c (merge_and_complain): Do not merge
17816 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
17817 fsigned_zeros, ftrapping_math, fwrapv.
17818 (append_compiler_options): Do not track these options.
17819 (append_linker_options): Likewie
17821 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17823 * cgraphunit.c (cgraph_node::finalize_function): When
17824 !flag_toplevel_reorde set no_reorder flag.
17825 (varpool_node::finalize_decl): Likewise.
17826 (symbol_table::compile): Drop no toplevel reorder path.
17828 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17830 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
17831 edges; zero probability is not better than uninitialized.
17833 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17835 * asan.h (asan_sanitize_allocas_p): Declare.
17836 * asan.c (asan_sanitize_allocas_p): New function.
17837 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17838 (handle_builtin_alloca): Likewise.
17839 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17840 if !asan_sanitize_allocas_p.
17841 * params.def (asan-instrument-allocas): Add new option.
17842 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17843 * opts.c (common_handle_option): Disable allocas sanitization for
17846 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17848 * asan.c: Include gimple-fold.h.
17849 (get_last_alloca_addr): New function.
17850 (handle_builtin_stackrestore): Likewise.
17851 (handle_builtin_alloca): Likewise.
17852 (asan_emit_allocas_unpoison): Likewise.
17853 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17854 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17855 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17856 (instrument_builtin_call): Pass gimple iterator to
17857 get_mem_refs_of_builtin_call.
17858 (last_alloca_addr): New global.
17859 * asan.h (asan_emit_allocas_unpoison): Declare.
17860 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17861 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17862 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17863 if function calls alloca.
17864 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17865 * gimple-fold.h (replace_call_with_value): Declare.
17866 * internal-fn.c: Include asan.h.
17867 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17868 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17870 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17872 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17873 (C_SELFTEST_FLAGS): New.
17874 (CPP_SELFTEST_FLAGS): New.
17875 (SELFTEST_DEPS): New, from deps of s-selftest.
17876 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17877 (CPP_SELFTEST_DEPS): New.
17878 (selftest): Add dependency on s-selftest-c++.
17879 (s-selftest): Rename to...
17880 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17881 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17882 than SELFTEST_FLAGS.
17883 (selftest-gdb): Rename to...
17884 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17886 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17887 (selftest-valgrind): Rename to...
17888 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
17890 (selftest-valgrind): Reintroduce as an alias for
17891 selftest-c-valgrind.
17892 (s-selftest-c++): New.
17893 (selftest-c++-gdb): New.
17894 (selftest-c++-valgrind): New.
17896 2017-07-06 Olivier Hainque <hainque@adacore.com>
17898 * gcc.c (process_command): When deciding if undefined variables
17899 should be ignored when processing specs, accept "gcc -v" as well.
17901 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17903 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17904 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17906 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17908 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17909 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17910 * config/arm/arm-tables.opt: Regenerate.
17911 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17913 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17915 2017-07-06 Carl Love <cel@us.ibm.com>
17917 * ChangeLog: Clean up from mid air collision
17919 2017-07-06 Carl Love <cel@us.ibm.com>
17921 * config/rs6000/rs6000-c.c: Add support for built-in functions
17922 vector signed int vec_subc (vector signed int, vector signed int);
17923 vector signed __int128 vec_subc (vector signed __int128,
17924 vector signed __int128);
17925 vector unsigned __int128 vec_subc (vector unsigned __int128,
17926 vector unsigned __int128);
17927 vector signed int vec_sube (vector signed int, vector signed int,
17928 vector signed int);
17929 vector unsigned int vec_sube (vector unsigned int,
17930 vector unsigned int,
17931 vector unsigned int);
17932 vector signed __int128 vec_sube (vector signed __int128,
17933 vector signed __int128,
17934 vector signed__int128);
17935 vector unsigned __int128 vec_sube (vector unsigned __int128,
17936 vector unsigned __int128,
17937 vector unsigned __int128);
17938 vector signed int vec_subec (vector signed int, vector signed int,
17939 vector signed int);
17940 vector unsigned int vec_subec (vector unsigned int,
17941 vector unsigned int,
17942 vector unsigned int);
17943 vector signed __int128 vec_subec (vector signed __int128,
17944 vector signed __int128,
17945 vector signed__int128);
17946 vector unsigned __int128 vec_subec (vector unsigned __int128,
17947 vector unsigned __int128,
17948 vector unsigned __int128);
17949 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
17950 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
17951 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
17952 BU_ALTIVEC_OVERLOAD_X definitions.
17953 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
17954 * doc/extend.texi: Update the built-in documentation file for the new
17955 built-in functions.
17957 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17960 * diagnostic-show-locus.c (layout::layout): Use start and finish
17961 spelling location for the start and finish of each range.
17962 * genmatch.c (linemap_client_expand_location_to_spelling_point):
17963 Add unused aspect param.
17964 * input.c (expand_location_1): Add "aspect" param, and use it
17965 to access the correct part of the location.
17966 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
17968 (expand_location_to_spelling_point): Likewise.
17969 (linemap_client_expand_location_to_spelling_point): Add "aspect"
17970 param, and pass it to expand_location_1.
17972 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
17974 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
17975 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
17976 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
17977 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
17978 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
17979 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
17980 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
17981 _mm_maskz_getmant_ss): New intrinsics.
17982 (__builtin_ia32_getexpss128_mask): Changed to ...
17983 __builtin_ia32_getexpss128_round ... this.
17984 (__builtin_ia32_getexpsd128_mask): Changed to ...
17985 __builtin_ia32_getexpsd128_round ... this.
17986 * config/i386/i386-builtin-types.def
17987 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
17988 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
17989 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
17990 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
17991 __builtin_ia32_getmantss_mask_round): New builtins.
17992 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
17993 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
17994 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
17995 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
17996 * config/i386/sse.md
17997 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
17998 avx512f_sgetexp<mode><mask_scalar_name>
17999 <round_saeonly_scalar_name> ... this.
18000 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18001 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18002 vgetexp<ssescalarmodesuffix>
18003 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18004 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18005 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18006 avx512f_vgetmant<mode><mask_scalar_name>
18007 <round_saeonly_scalar_name> ... this.
18008 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18009 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18010 vgetmant<ssescalarmodesuffix>
18011 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18012 %0<mask_scalar_operand4>, %1, %2
18013 <round_saeonly_scalar_mask_op4>, %3} ... this.
18014 * config/i386/subst.md (mask_scalar_operand4,
18015 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18016 round_saeonly_scalar_nimm_predicate): New subst attributes.
18018 2017-07-06 Julia Koval <julia.koval@intel.com>
18020 * config/i386/i386.c (ix86_erase_embedded_rounding):
18021 Remove code for old rounding pattern.
18023 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18025 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18027 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18029 * doc/sourcebuild.texi (Test Directives, Variants of
18030 dg-require-support): Add documentation for dg-require-stack-check.
18032 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18034 * config/i386/subst.md (mask_scalar, round_scalar,
18035 round_saeonly_scalar): New meta-templates.
18036 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18037 round_scalar_mask_operand3, round_scalar_mask_op3,
18038 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18039 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18040 round_saeonly_scalar_constraint,
18041 round_saeonly_scalar_prefix): New subst attribute.
18042 * config/i386/sse.md
18043 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18044 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18045 <round_scalar_name> ... this.
18046 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18047 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18048 <round_scalar_name> ... this.
18049 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18050 <sse>_vm<code><mode>3<mask_scalar_name>
18051 <round_saeonly_scalar_name> ... this.
18052 (v<plusminus_mnemonic><ssescalarmodesuffix>
18053 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18054 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18055 v<plusminus_mnemonic><ssescalarmodesuffix>
18056 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18057 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18058 (v<multdiv_mnemonic><ssescalarmodesuffix>
18059 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18060 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18061 v<multdiv_mnemonic><ssescalarmodesuffix>
18062 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18063 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18064 (v<maxmin_float><ssescalarmodesuffix>
18065 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18066 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18067 v<maxmin_float><ssescalarmodesuffix>
18068 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18069 %0<mask_scalar_operand3>, %1, %<iptr>2
18070 <round_saeonly_scalar_mask_op3>} ... this.
18072 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18074 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18075 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18077 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18078 Alan Hayward <alan.hayward@arm.com>
18079 David Sherwood <david.sherwood@arm.com>
18081 * combine.c (simplify_if_then_else): Remove "enum" before
18083 * compare-elim.c (can_eliminate_compare): Likewise.
18084 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18086 (aarch64_lookup_simd_builtin_type): Likewise.
18087 (aarch64_simd_builtin_type): Likewise.
18088 (aarch64_init_simd_builtin_types): Likewise.
18089 (aarch64_simd_expand_args): Likewise.
18090 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18092 (aarch64_reverse_mask): Likewise.
18093 (aarch64_simd_emit_reg_reg_move): Likewise.
18094 (aarch64_gen_adjusted_ldpstp): Likewise.
18095 (aarch64_ccmp_mode_to_code): Likewise.
18096 (aarch64_operands_ok_for_ldpstp): Likewise.
18097 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18098 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18100 (aarch64_min_divisions_for_recip_mul): Likewise.
18101 (aarch64_reassociation_width): Likewise.
18102 (aarch64_get_condition_code_1): Likewise.
18103 (aarch64_simd_emit_reg_reg_move): Likewise.
18104 (aarch64_simd_attr_length_rglist): Likewise.
18105 (aarch64_reverse_mask): Likewise.
18106 (aarch64_operands_ok_for_ldpstp): Likewise.
18107 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18108 (aarch64_gen_adjusted_ldpstp): Likewise.
18109 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18111 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18112 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18113 (arm_lookup_simd_builtin_type): Likewise.
18114 (arm_simd_builtin_type): Likewise.
18115 (arm_init_simd_builtin_types): Likewise.
18116 (arm_expand_builtin_args): Likewise.
18117 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18118 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18119 (ft32_setup_incoming_varargs): Likewise.
18120 (ft32_function_arg): Likewise.
18121 (ft32_function_arg_advance): Likewise.
18122 (ft32_pass_by_reference): Likewise.
18123 (ft32_arg_partial_bytes): Likewise.
18124 (ft32_valid_pointer_mode): Likewise.
18125 (ft32_addr_space_pointer_mode): Likewise.
18126 (ft32_addr_space_legitimate_address_p): Likewise.
18127 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18129 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18130 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18131 (iamcu_alignment): Likewise.
18132 (canonicalize_vector_int_perm): Likewise.
18133 (ix86_noce_conversion_profitable_p): Likewise.
18134 (ix86_mpx_bound_mode): Likewise.
18135 (ix86_operands_ok_for_move_multiple): Likewise.
18136 * config/microblaze/microblaze-protos.h
18137 (microblaze_expand_conditional_branch_reg): Likewise.
18138 * config/microblaze/microblaze.c
18139 (microblaze_expand_conditional_branch_reg): Likewise.
18140 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18142 (rs6000_reassociation_width): Likewise.
18143 (rs6000_invalid_binary_op): Likewise.
18144 (fusion_p9_p): Likewise.
18145 (emit_fusion_p9_load): Likewise.
18146 (emit_fusion_p9_store): Likewise.
18147 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18149 (riscv_hard_regno_mode_ok_p): Likewise.
18150 (riscv_address_insns): Likewise.
18151 (riscv_split_symbol): Likewise.
18152 (riscv_legitimize_move): Likewise.
18153 (riscv_function_value): Likewise.
18154 (riscv_hard_regno_nregs): Likewise.
18155 (riscv_expand_builtin): Likewise.
18156 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18157 (riscv_build_integer): Likewise.
18158 (riscv_split_integer): Likewise.
18159 (riscv_legitimate_constant_p): Likewise.
18160 (riscv_cannot_force_const_mem): Likewise.
18161 (riscv_regno_mode_ok_for_base_p): Likewise.
18162 (riscv_valid_base_register_p): Likewise.
18163 (riscv_valid_offset_p): Likewise.
18164 (riscv_valid_lo_sum_p): Likewise.
18165 (riscv_classify_address): Likewise.
18166 (riscv_legitimate_address_p): Likewise.
18167 (riscv_address_insns): Likewise.
18168 (riscv_load_store_insns): Likewise.
18169 (riscv_force_binary): Likewise.
18170 (riscv_split_symbol): Likewise.
18171 (riscv_force_address): Likewise.
18172 (riscv_legitimize_address): Likewise.
18173 (riscv_move_integer): Likewise.
18174 (riscv_legitimize_const_move): Likewise.
18175 (riscv_legitimize_move): Likewise.
18176 (riscv_address_cost): Likewise.
18177 (riscv_subword): Likewise.
18178 (riscv_output_move): Likewise.
18179 (riscv_canonicalize_int_order_test): Likewise.
18180 (riscv_emit_int_order_test): Likewise.
18181 (riscv_function_arg_boundary): Likewise.
18182 (riscv_pass_mode_in_fpr_p): Likewise.
18183 (riscv_pass_fpr_single): Likewise.
18184 (riscv_pass_fpr_pair): Likewise.
18185 (riscv_get_arg_info): Likewise.
18186 (riscv_function_arg): Likewise.
18187 (riscv_function_arg_advance): Likewise.
18188 (riscv_arg_partial_bytes): Likewise.
18189 (riscv_function_value): Likewise.
18190 (riscv_pass_by_reference): Likewise.
18191 (riscv_setup_incoming_varargs): Likewise.
18192 (riscv_print_operand): Likewise.
18193 (riscv_elf_select_rtx_section): Likewise.
18194 (riscv_save_restore_reg): Likewise.
18195 (riscv_for_each_saved_reg): Likewise.
18196 (riscv_register_move_cost): Likewise.
18197 (riscv_hard_regno_mode_ok_p): Likewise.
18198 (riscv_hard_regno_nregs): Likewise.
18199 (riscv_class_max_nregs): Likewise.
18200 (riscv_memory_move_cost): Likewise.
18201 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18202 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18203 (rl78_addr_space_address_mode): Likewise.
18204 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18206 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18207 (rs6000_reassociation_width): Likewise.
18208 (rs6000_invalid_binary_op): Likewise.
18209 (fusion_p9_p): Likewise.
18210 (emit_fusion_p9_load): Likewise.
18211 (emit_fusion_p9_store): Likewise.
18212 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18213 (ok_for_simple_move_operands): Likewise.
18214 (ok_for_simple_move_strict_operands): Likewise.
18215 (ok_for_simple_arith_logic_operands): Likewise.
18216 (visium_legitimize_reload_address): Likewise.
18217 (visium_select_cc_mode): Likewise.
18218 (output_cbranch): Likewise.
18219 (visium_split_double_move): Likewise.
18220 (visium_expand_copysign): Likewise.
18221 (visium_expand_int_cstore): Likewise.
18222 (visium_expand_fp_cstore): Likewise.
18223 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18224 (visium_function_arg): Likewise.
18225 (visium_function_arg_advance): Likewise.
18226 (visium_libcall_value): Likewise.
18227 (visium_setup_incoming_varargs): Likewise.
18228 (visium_legitimate_constant_p): Likewise.
18229 (visium_legitimate_address_p): Likewise.
18230 (visium_legitimize_address): Likewise.
18231 (visium_secondary_reload): Likewise.
18232 (visium_register_move_cost): Likewise.
18233 (visium_memory_move_cost): Likewise.
18234 (prepare_move_operands): Likewise.
18235 (ok_for_simple_move_operands): Likewise.
18236 (ok_for_simple_move_strict_operands): Likewise.
18237 (ok_for_simple_arith_logic_operands): Likewise.
18238 (visium_function_value_1): Likewise.
18239 (rtx_ok_for_offset_p): Likewise.
18240 (visium_legitimize_reload_address): Likewise.
18241 (visium_split_double_move): Likewise.
18242 (visium_expand_copysign): Likewise.
18243 (visium_expand_int_cstore): Likewise.
18244 (visium_expand_fp_cstore): Likewise.
18245 (visium_split_cstore): Likewise.
18246 (visium_select_cc_mode): Likewise.
18247 (visium_split_cbranch): Likewise.
18248 (output_cbranch): Likewise.
18249 (visium_print_operand_address): Likewise.
18250 * expmed.c (flip_storage_order): Likewise.
18251 * expmed.h (emit_cstore): Likewise.
18252 (flip_storage_order): Likewise.
18253 * genrecog.c (validate_pattern): Likewise.
18254 * hsa-gen.c (gen_hsa_addr): Likewise.
18255 * internal-fn.c (expand_arith_overflow): Likewise.
18256 * ira-color.c (allocno_copy_cost_saving): Likewise.
18257 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18258 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18259 (process_invariant_for_inheritance): Likewise.
18260 * lra-eliminations.c (move_plus_up): Likewise.
18261 * omp-low.c (lower_oacc_reductions): Likewise.
18262 * simplify-rtx.c (simplify_subreg): Likewise.
18263 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18264 (TARGET_CHKP_BOUND_MODE): Likewise..
18265 * targhooks.c (default_chkp_bound_mode): Likewise.
18266 (default_setup_incoming_vararg_bounds): Likewise.
18267 * targhooks.h (default_chkp_bound_mode): Likewise.
18268 (default_setup_incoming_vararg_bounds): Likewise.
18269 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18270 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18271 (have_whole_vector_shift): Likewise.
18272 * tree-vect-stmts.c (vectorizable_load): Likewise.
18273 * doc/tm.texi: Regenerate.
18275 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18277 Graceful degrade if Binutils PR21472 is not available.
18280 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18281 .rodata in flash test fails.
18282 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18283 * confgure: Regenerate.
18284 * config.in: Regenerate.
18285 * config/avr/avr.c (avr_asm_named_section)
18286 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18287 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18288 (avr_asm_init_sections): Same.
18290 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18292 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18293 (fma<VH:mode>4_intrinsic): Likewise.
18294 (*fmsub<VCVTF:mode>4): Likewise.
18295 (*fmsub<VH:mode>4_intrinsic): Likewise.
18297 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18300 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18301 Don't depend on "optimize > 0".
18302 (out_movhi_r_mr, out_movqi_mr_r): Same.
18303 (out_movhi_mr_r, out_movqi_r_mr): Same.
18304 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18305 io_address_operand on "optimize > 0".
18307 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18309 * tree-loop-distribution.c: Add general explanantion on the pass.
18310 (generate_loops_for_partition): Mark distributed loop.
18311 (pg_add_dependence_edges): New parameter. Handle alias data
18312 dependence specially and record it in the parameter if asked.
18313 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18314 (init_partition_graph_vertices, add_partition_graph_edge): New.
18315 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18316 (free_partition_graph_vdata, build_partition_graph): New.
18317 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18318 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18319 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18320 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18321 (version_for_distribution_p, finalize_partitions): New.
18322 (distribute_loop): Handle alias data dependence specially. Factor
18323 out loop fusion code as functions and call these functions.
18325 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18327 * tree-loop-distribution.c (classify_partition): New parameter and
18328 better handle reduction statement.
18329 (rdg_build_partitions): Revise comment.
18330 (distribute_loop): Compute statements in all partitions and pass it
18331 to classify_partition.
18333 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18335 * tree-loop-distribution.c (enum partition_type): New.
18336 (struct partition): New field type.
18337 (partition_merge_into): Add parameter. Update partition type.
18338 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18339 (build_rdg_partition_for_vertex): Compute partition type.
18340 (rdg_build_partitions): Dump partition type.
18341 (distribute_loop): Update calls to partition_merge_into.
18343 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18345 * tree-loop-distribution.c (struct ddr_hasher): New.
18346 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18348 (classify_partition): Call get_data_dependence.
18349 (pg_add_dependence_edges): Ditto.
18350 (distribute_loop): Release data dependence hash table.
18352 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18354 * tree-loop-distribution.c (ref_base_address): Delete.
18355 (similar_memory_accesses): Rename ...
18356 (share_memory_accesses): ... to this. Check if partitions access
18357 the same memory reference.
18358 (distribute_loop): Call share_memory_accesses.
18360 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18362 * tree-loop-distribution.c (struct partition): New field recording
18363 its data reference.
18364 (partition_alloc, partition_free): Init and release data refs.
18365 (partition_merge_into): Merge data refs.
18366 (build_rdg_partition_for_vertex): Collect data refs for partition.
18367 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18369 (distribute_loop): Remve data refs from vertice data of partition
18372 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18374 * tree-loop-distribution.c (params.h): Include header file.
18375 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18376 (datarefs_vec): New global var.
18377 (create_rdg_vertices): Use datarefs_vec directly.
18378 (free_rdg): Don't free data references.
18379 (build_rdg): Update use. Don't free data references.
18380 (distribute_loop): Compute global variable for data references.
18381 Bail out if there are too many data references.
18383 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18385 * tree-loop-distribution.c (loop_nest): New global var.
18386 (build_rdg): Use loop directly, rather than loop nest.
18387 (pg_add_dependence_edges): Remove loop nest parameter. Use global
18389 (distribute_loop): Compute global variable loop nest. Update use.
18391 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18393 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
18394 (partition_merge_into): New parameter. Dump reason for fusion.
18395 (distribute_loop): Update use of partition_merge_into.
18397 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18399 * tree-loop-distribution.c (bb_top_order_index): New.
18400 (bb_top_order_index_size, bb_top_order_cmp): New.
18401 (stmts_from_loop): Use topological order.
18402 (pass_loop_distribution::execute): Compute and release topological
18403 order for basic blocks.
18405 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18407 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
18410 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18412 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
18413 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
18414 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
18415 * internal-fn.def (LOOP_DIST_ALIAS): New.
18416 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
18417 (fold_loop_internal_call): ... this.
18418 (vect_loop_dist_alias_call): New function.
18419 (set_uid_loop_bbs): Call fold_loop_internal_call.
18420 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
18423 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18426 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
18427 Require dead FLAGS_REG at the beginning of a peephole.
18429 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18432 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
18433 arguments in the call to __builtin_ia32_sbb_u32.
18434 (_subborrow_u64): Swap _X and _Y arguments in the call to
18435 __builtin_ia32_sbb_u64.
18437 2017-07-04 Jakub Jelinek <jakub@redhat.com>
18440 * tree-vrp.c (compare_assert_loc): Turn into a function template
18441 with stable template parameter. Only test if a->e is NULL,
18442 !a->e == !b->e has been verified already. Use e == NULL or
18443 e != NULL instead of e or ! e tests. If stable is true, don't use
18444 iterative_hash_expr, on the other side allow a or b or both NULL
18445 and sort the NULLs last.
18446 (process_assert_insertions): Sort using compare_assert_loc<false>
18447 instead of compare_assert_loc, later sort using
18448 compare_assert_loc<true> before calling process_assert_insertions_for
18449 in a loop. Use break instead of continue once seen NULL pointer.
18451 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18453 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18454 Cortex-R7 and Cortex-R8 processors.
18456 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18458 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
18459 uninitialized while src is not.
18461 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
18463 * common/config/arm/arm-common.c: Adjust include path for
18465 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
18466 (arm-cpu.h): Create in build directory. Adjust dependency rules.
18467 (arm-cpu-data.h): Likewise.
18468 (arm-cpu-cdata.h): Likewise.
18469 * config/arm/arm-cpu.h: Delete.
18470 * config/arm/arm-cpu-cdata.h: Delete.
18471 * config/arm/arm-cpu-data.h: Delete.
18473 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
18475 * config/arm/arm-cpus.in (cortex-a55): New.
18476 (cortex-a75): Likewise.
18477 (cortex-a75.cortex-a55): Likewise.
18478 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
18480 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
18481 * config/arm/arm-cpu-cdata.h: Regenerate.
18482 * config/arm/arm-cpu-data.h: Regenerate.
18483 * config/arm/arm-cpu.h: Regenerate.
18484 * config/arm/arm-tables.opt: Regenerate.
18485 * config/arm/arm-tune.md: Regenerate.
18487 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18489 * haifa-sched.c (sched_create_recovery_edges): Update profile.
18491 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18493 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
18496 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
18498 PR tree-optimization/81292
18499 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
18500 full_string_p, also call adjust_related_strinfos if the adjustment
18501 is simple, otherwise invalidate related strinfos.
18503 2017-07-04 Martin Liska <mliska@suse.cz>
18506 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
18507 newly created variable as DECL_IGNORED_P.
18509 2017-07-04 Martin Liska <mliska@suse.cz>
18512 * ipa-inline.c (inline_small_functions):
18513 Use xstrdup_for_dump.
18515 2017-07-04 Tom de Vries <tom@codesourcery.com>
18517 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
18519 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18522 * config/darwin.c (darwin_function_switched_text_sections):
18525 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
18527 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
18529 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18531 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
18533 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18535 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
18536 min_profitable_iters, and th as inclusive lower bounds.
18537 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
18538 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
18539 for min_profitable_iters and min_profitable_estimate.
18540 (vect_transform_loop): Treat th as an inclusive lower bound.
18541 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
18543 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18546 * config/darwin.c (darwin_function_switched_text_sections):
18547 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
18548 in two pieces, and suppress the use of buf.
18550 2017-07-03 Nathan Sidwell <nathan@acm.org>
18552 * hash-table.h (hash_table_mod1): Fix indentation.
18554 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18556 PR middle-end/81290
18557 * predict.c (force_edge_cold): Be more careful about propagation
18559 * profile-count.h (profile_probability::guessed,
18560 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
18562 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
18564 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
18566 * doc/invoke.texi (rcpc architecture extension): Document it.
18568 2017-07-03 Richard Biener <rguenther@suse.de>
18570 PR tree-optimization/60510
18571 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
18572 the scalar reduction PHI and use it.
18573 (vectorizable_reduction): Properly guard the single_defuse_cycle
18574 path for non-SLP reduction chains where we cannot use it.
18575 Rework reduc_def/index and vector type deduction. Rework
18576 vector operand gathering during reduction op code-gen.
18577 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
18578 chains dissolve the chain and leave it to non-SLP reduction
18581 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18583 * tree-data-ref.h (dr_alignment): Declare.
18584 * tree-data-ref.c (dr_alignment): New function.
18585 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
18586 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
18588 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
18590 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18592 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
18593 and base_misalignment fields.
18594 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
18595 * tree-data-ref.c: Include builtins.h.
18596 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
18597 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
18598 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
18599 * tree-vect-data-refs.c: Include tree-cfg.h.
18600 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
18601 fields instead of calculating an alignment here.
18602 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
18603 innermost_loop_behavior fields.
18605 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18607 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
18609 (DR_STEP_ALIGNMENT): New macro.
18610 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
18611 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
18612 (create_data_ref): Print it.
18613 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
18614 to tell whether the step preserves vector (mis)alignment.
18615 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18616 Move the check for an integer step and generalise to all INTEGER_CST.
18617 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
18618 Print the outer step alignment.
18620 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18622 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
18623 with offset_alignment.
18624 (DR_ALIGNED_TO): Delete.
18625 (DR_OFFSET_ALIGNMENT): New macro.
18626 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
18627 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
18628 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
18629 (create_data_ref): Likewise.
18630 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18631 (vect_analyze_data_refs): Likewise.
18632 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
18633 creating dummy innermost behavior.
18635 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18637 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
18638 with a "innermost_loop_behavior *" and refeence tree.
18639 * tree-data-ref.c (dr_analyze_innermost): Likewise.
18640 (create_data_ref): Update call accordingly.
18641 * tree-predcom.c (find_looparound_phi): Likewise.
18643 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18645 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
18646 fields with dr_wrt_vec_loop.
18647 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
18648 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
18649 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
18650 (vect_dr_behavior): New function.
18651 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18652 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
18653 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
18654 track whether the step preserves the misalignment.
18655 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
18656 Use vect_dr_behavior.
18657 (vect_setup_realignment): Update call accordingly.
18658 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
18659 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
18660 call to vect_create_addr_base_for_vector_ref.
18661 (vect_create_cond_for_align_checks): Likewise.
18662 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
18663 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
18664 (vect_recog_mask_conversion_pattern): Likewise.
18665 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
18666 (new_stmt_vec_info): Remove redundant zeroing.
18668 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18670 * common/config/arm/arm-common.c (arm_be8_option): New function.
18671 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
18672 (ISA_ARMv6): Add isa_bit_be8.
18673 * config/arm/arm.h (arm_be8_option): Add prototype.
18674 (BE8_SPEC_FUNCTION): New define.
18675 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
18676 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
18677 (mlittle-endian): Similarly.
18678 (mbe8, mbe32): New options.
18679 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
18680 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
18682 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18684 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
18686 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18688 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
18689 (cleanup_tree_cfg_bb): Use it.
18690 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
18692 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
18694 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18697 * loop-doloop.c (add_test): Update profile.
18699 2017-07-03 Martin Liska <mliska@suse.cz>
18702 * sanopt.c (rewrite_usage_of_param): New function.
18703 (sanitize_rewrite_addressable_params): Likewise.
18704 (pass_sanopt::execute): Call rewrite_usage_of_param.
18706 2017-07-03 Richard Biener <rguenther@suse.de>
18708 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
18709 back to using VIEW_CONVERT_EXPR.
18711 2017-07-03 Martin Liska <mliska@suse.cz>
18714 * doc/extend.texi: Document when a resolver function is
18715 generated for target_clones.
18717 2017-07-03 Martin Liska <mliska@suse.cz>
18719 * asan.c (asan_emit_stack_protection): Unpoison just red zones
18720 and shadow memory of auto variables which are subject of
18721 use-after-scope sanitization.
18722 (asan_expand_mark_ifn): Add do set only when is_poison.
18724 2016-07-03 Richard Biener <rguenther@suse.de>
18726 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
18728 (vect_force_simple_reduction): Record reduction def -> phi mapping.
18729 (vectorizable_reduction): Perform reduction PHI creation when
18730 visiting a reduction PHI and adjust and simplify code generation
18731 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
18732 (vect_transform_loop): Visit reduction PHIs.
18733 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
18734 defs into the SLP tree.
18735 (vect_build_slp_tree): Reduction defs terminate the recursion.
18736 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
18738 (vect_get_vec_defs_for_stmt_copy): Export.
18739 (vect_get_vec_defs): Likewise.
18740 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
18742 (vect_get_vec_defs_for_stmt_copy): Declare.
18743 (vect_get_vec_defs): Likewise.
18745 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18747 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
18748 parameter with a "loop" parameter and use it instead of the
18749 loop containing DR_STMT. Don't check simple_iv when doing
18750 BB analysis. Describe the two analysis modes in the comment.
18752 2017-07-03 Tom de Vries <tom@codesourcery.com>
18754 PR tree-optimization/69468
18755 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
18756 (find_same_succ_bb): Handle ignore_edge_flags.
18758 2017-07-03 Tom de Vries <tom@codesourcery.com>
18760 PR tree-optimization/81192
18761 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
18763 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
18765 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
18767 2017-07-03 Tom de Vries <tom@codesourcery.com>
18769 PR tree-optimization/81192
18770 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
18771 BB_SAME_SUCC (bb) == NULL.
18773 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18775 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
18778 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18780 * dumpfile.c: Include profile-count.h
18781 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
18783 (insert_cond_bb): Update profile.
18784 * tree-cfg.h (insert_cond_bb): Update prototype.
18785 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
18786 * tree-dump.c: Do not include tree-cfg.
18788 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18790 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
18792 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18794 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
18797 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18799 * tree-complex.c (expand_complex_div_wide): update profile.
18801 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18802 Alan Hayward <alan.hayward@arm.com>
18803 David Sherwood <david.sherwood@arm.com>
18805 * Makefile.in (MACHMODE_H): Remove insn-modes.h
18806 (CORETYPES_H): New define.
18807 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
18808 (insn-modes-inline.h, s-modes-inline-h): New rules.
18809 (generated_files): Add insn-modes-inline.h.
18810 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
18811 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
18812 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
18813 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
18814 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
18815 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
18816 (build/gencodes.o, build/genconditions.o): Likewise.
18817 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
18818 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
18819 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
18820 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
18821 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
18822 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
18823 * coretypes.h: Include everything up to real.h for generators.
18824 Include insn-modes.h first. Include wide-int-print.h after
18825 wide-int.h. Include insn-modes-inline.h and then machmode.h.
18826 * machmode.h: Don't include insn-modes.h here.
18827 * function-tests.c: Remove includes of signop.h, machmode.h,
18828 double-int.h and wide-int.h.
18830 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18832 * optc-save-gen.awk: Likewise.
18833 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18834 * godump.c: Remove include of wide-int-print.h.
18835 * pretty-print.h: Likewise.
18836 * wide-int-print.cc: Likewise.
18837 * wide-int.cc: Likewise.
18838 * hash-map-tests.c: Remove include of signop.h.
18839 * hash-set-tests.c: Likewise.
18840 * rtl-tests.c: Likewise.
18841 * mkconfig.sh: Remove include of machmode.h.
18842 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18844 (emit_insn_modes_inline_h): ...this new function. Emit the code
18845 into an insn-modes-inline.h header file, adding appropriate
18846 include guards and end comments.
18847 (emit_insn_modes_c_header): Remove include of machmode.h.
18848 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18850 (main): Handle -i flag and call emit_insn_modes_inline_h when
18853 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18855 * tree-ssa-strlen.c (strinfo): Rename the length field to
18856 nonzero_chars. Add a full_string_p field.
18857 (compare_nonzero_chars, zero_length_string_p): New functions.
18858 (get_addr_stridx): Add an offset_out parameter.
18859 Use compare_nonzero_chars.
18860 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18861 (new_strinfo): Update after above changes to strinfo.
18862 (set_endptr_and_length): Set full_string_p.
18863 (get_string_length): Update after above changes to strinfo.
18864 (unshare_strinfo): Update call to new_strinfo.
18865 (maybe_invalidate): Likewise.
18866 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18867 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18868 as a uhwi instead of an shwi. Update after above changes to
18869 strinfo and new_strinfo.
18870 (zero_length_string): Assert that chainsi contains full strings.
18871 Use zero_length_string_p. Update call to new_strinfo.
18872 (adjust_related_strinfos): Update after above changes to strinfo.
18873 Copy full_string_p from origsi.
18874 (adjust_last_stmt): Use zero_length_string_p.
18875 (handle_builtin_strlen): Update after above changes to strinfo and
18876 new_strinfo. Install the lhs as the string length if the previous
18877 entry didn't describe a full string.
18878 (handle_builtin_strchr): Update after above changes to strinfo
18880 (handle_builtin_strcpy): Likewise.
18881 (handle_builtin_strcat): Likewise.
18882 (handle_builtin_malloc): Likewise.
18883 (handle_pointer_plus): Likewise.
18884 (handle_builtin_memcpy): Likewise. Track nonzero characters
18885 that aren't necessarily followed by a nul terminator.
18886 (handle_char_store): Likewise.
18888 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18890 PR tree-optimization/80769
18891 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
18892 for malloc and calloc. Document the new invariant that all related
18893 strinfos have delayed lengths or none do.
18894 (verify_related_strinfos): Move earlier in file.
18895 (set_endptr_and_length): New function, split out from...
18896 (get_string_length): ...here. Also set the lengths of related
18898 (zero_length_string): Assert that chainsi has known (rather than
18900 (adjust_related_strinfos): Likewise.
18902 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18904 PR tree-optimization/81136
18905 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18906 assert that two references with the same misalignment have the same
18907 compile-time misalignment if those compile-time misalignments
18910 2017-07-01 Andi Kleen <ak@linux.intel.com>
18912 * print-tree.c (print_node): Print all attributes.
18914 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18916 * cfg.c (scale_bbs_frequencies): New function.
18917 * cfg.h (scale_bbs_frequencies): Declare it.
18918 * cfgloopanal.c (single_likely_exit): Cleanup.
18919 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18921 (scale_loop_profile): Likewise.
18922 (loop_version): Likewise.
18923 (create_empty_loop_on_edge): Update.
18924 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18925 scale_loop_frequencies, scale_loop_profile, loopify,
18926 loop_version): Update prototypes.
18927 * modulo-sched.c (sms_schedule): Update.
18928 * predict.c (unlikely_executed_edge_p): Also check probability.
18929 (probably_never_executed_edge_p): Fix typo.
18930 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18931 * tree-parloops.c (gen_parallel_loop): Update.
18932 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18933 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18934 * tree-ssa-loop-split.c (split_loop): Update.
18935 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18936 * tree-vect-loop-manip.c (vect_do_peeling): Update.
18937 (vect_loop_versioning): Update.
18938 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18940 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18942 * trans-mem.c (split_bb_make_tm_edge): Update profile.
18944 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18946 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
18947 to keep profile consistent.
18949 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18951 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
18952 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
18953 * profile-count.h (max_safe_multiplier): Make unsigned.
18954 (profile_count::guessed_zero): New.
18956 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18958 * bb-reorder.c (fix_up_crossing_landing_pad,
18959 fix_crossing_conditional_branches): Use make_single_succ_edge
18960 to keep profile consistent.
18962 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18964 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
18967 2017-07-01 Jakub Jelinek <jakub@redhat.com>
18970 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
18971 the right scopes, make sure cond_jump isn't preserved between multiple
18972 iterations. Search for fallthru edge whenever there are 3+ edges and
18973 use find_fallthru_edge for it.
18975 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18977 Patch by Alexander Monakov <amonakov@ispras.ru>
18978 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
18979 probabilities consistently.
18981 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18983 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18984 emit_cmp_and_jump_insns.
18986 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18989 * tree-inline.c (expand_call_inline): Combine profile statuses.
18991 2017-06-30 Andrew Pinski <apinski@cavium.com>
18993 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
18994 fold_stmt returned true.
18996 2017-06-30 Nathan Sidwell <nathan@acm.org>
18998 * ggc.h (empty_string): Delete.
18999 * cfgexpand.c (expand_asm_stmt): Use plain "".
19000 * optabs.c (expand_asm_memory_barrier): Likewise.
19001 * stringpool.c (empty_string): Delete.
19002 (digit_vector, digit_string): Delete.
19003 (ggc_alloc_string): Use plain "", don't optimize single digit
19004 strings. Use ggc_alloc_atomic.
19006 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19008 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19009 comparison set and one other set, use the cost of the non-comparison
19012 2017-06-30 Nathan Sidwell <nathan@acm.org>
19014 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19017 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19019 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19020 loops. Remove now unneeded calls to gimple_switch_set_label() that
19021 just set removed labels to NULL_TREE.
19023 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19025 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19026 (set_range_info): ...here. Only call set_range_info_raw if domain
19028 (set_nonzero_bits): Call set_range_info_raw.
19029 * tree-ssanames.h (set_range_info_raw): New.
19031 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19034 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19035 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19036 of nonimmediate_operand and <store_mask_constraint> instead of m
19037 for the input operand. For V8FI iterator, always split if input
19038 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19039 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19040 <store_mask_predicate> instead of register_operand and
19041 <store_mask_constraint> instead of v for the input operand. Make
19042 sure both operands aren't MEMs for if not <mask_applied>.
19044 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19046 * lto-wrapper.c (copy_file) Close both file descriptors before
19049 2017-06-30 Martin Liska <mliska@suse.cz>
19052 * multiple_target.c (create_dispatcher_calls): Make ifunc
19053 also for function that don't have calls or are not referenced.
19055 2017-06-30 Richard Biener <rguenther@suse.de>
19057 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19058 analyze the first scalar stmt. Move vector type computation
19059 for the BB case here from ...
19060 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19061 live operation processing in the SLP case properly.
19063 2017-06-30 Richard Biener <rguenther@suse.de>
19065 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19067 2017-06-30 Martin Liska <mliska@suse.cz>
19070 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19071 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19073 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19075 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19078 2017-06-30 Richard Biener <rguenther@suse.de>
19080 PR tree-optimization/81249
19081 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19082 condition reduction result to original scalar type.
19084 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19086 * profile-count.h (enum profile_quality): Fix typos and whitespace
19089 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19091 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19092 type for branch probabilities.
19094 2017-06-29 Julian Brown <julian@codesourcery.com>
19095 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19097 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19098 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19099 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19100 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19102 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19104 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19105 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19106 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19107 CC usage from generic code to here.
19108 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19109 CC usage into the target macros.
19111 2017-06-29 Maya Rashish <coypu@sdf.org>
19113 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19116 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19118 * arm/arm-builtins.c: Include profile-count.h
19119 * except.c (sjlj_emit_function_enter): Use
19120 profile_probability::unlikely.
19122 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19124 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19125 and tocrel_offset be pointer args rather than implicitly using
19127 (legitimate_constant_pool_address_p, rs6000_emit_move,
19128 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19129 tocrel_offset and use in toc_relative_expr_p call.
19130 (print_operand, print_operand_address): Use static tocrel_base_oac
19131 and tocrel_offset_oac.
19132 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19135 2017-06-29 Maya Rashish <coypu@sdf.org>
19137 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19139 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19141 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19142 objects, take into account only the alignment of 'op0' and 'mode1' if
19145 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19147 * ccmp.c (ccmp_tree_comparison_p): New function.
19148 (ccmp_candidate_p): Update to use above function.
19149 (get_compare_parts): New function.
19150 (expand_ccmp_next): Update to use new functions.
19151 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19153 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19154 take mode as argument.
19155 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19156 * expr.c (expand_expr_real_1): Pass mode as argument.
19158 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19160 * combine.c (combine_instructions): Print insns to dump_file, together
19163 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19165 * asan.c (asan_emit_stack_protection): Update.
19166 (create_cond_insert_point): Update.
19167 * auto-profile.c (afdo_propagate_circuit): Update.
19168 * basic-block.h (struct edge_def): Turn probability to
19169 profile_probability.
19170 (EDGE_FREQUENCY): Update.
19171 * bb-reorder.c (find_traces_1_round): Update.
19172 (better_edge_p): Update.
19173 (sanitize_hot_paths): Update.
19174 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19175 (make_single_succ_edge): Update.
19176 (check_bb_profile): Update.
19177 (dump_edge_info): Update.
19178 (update_bb_profile_for_threading): Update.
19179 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19180 probabilitycount to 0.
19181 * cfgbuild.c (compute_outgoing_frequencies): Update.
19182 * cfgcleanup.c (try_forward_edges): Update.
19183 (outgoing_edges_match): Update.
19184 (try_crossjump_to_edge): Update.
19185 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19186 (expand_gimple_tailcall): Update.
19187 (construct_init_block): Use make_single_succ_edge.
19188 (construct_exit_block): Use make_single_succ_edge.
19189 * cfghooks.c (verify_flow_info): Update.
19190 (redirect_edge_succ_nodup): Update.
19191 (split_edge): Update.
19192 (account_profile_record): Update.
19193 * cfgloopanal.c (single_likely_exit): Update.
19194 * cfgloopmanip.c (scale_loop_profile): Update.
19195 (set_zero_probability): Remove.
19196 (duplicate_loop_to_header_edge): Update.
19197 * cfgloopmanip.h (loop_version): Update prototype.
19198 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19199 (force_nonfallthru_and_redirect): Update.
19200 (update_br_prob_note): Update.
19201 (rtl_verify_edges): Update.
19202 (purge_dead_edges): Update.
19203 (rtl_lv_add_condition_to_bb): Update.
19204 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19205 * cgraphunit.c (init_lowered_empty_function): Update.
19206 (cgraph_node::expand_thunk): Update.
19207 * cilk-common.c: Include profile-count.h
19208 * dojump.c (inv): Remove.
19209 (jumpifnot): Update.
19210 (jumpifnot_1): Update.
19211 (do_jump_1): Update.
19213 (do_jump_by_parts_greater_rtx): Update.
19214 (do_compare_rtx_and_jump): Update.
19215 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19216 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19217 * dwarf2cfi.c: Include profile-count.h
19218 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19219 (sjlj_emit_dispatch_table): Likewise.
19220 * explow.c: Include profile-count.h
19221 * expmed.c (emit_store_flag_force): Update.
19222 (do_cmp_and_jump): Update.
19223 * expr.c (compare_by_pieces_d::generate): Update.
19224 (compare_by_pieces_d::finish_mode): Update.
19225 (emit_block_move_via_loop): Update.
19226 (store_expr_with_bounds): Update.
19227 (store_constructor): Update.
19228 (expand_expr_real_2): Update.
19229 (expand_expr_real_1): Update.
19230 * expr.h (try_casesi, try_tablejump): Update prototypes.
19231 * gimple-pretty-print.c (dump_probability): Update.
19232 (dump_profile): New.
19233 (dump_gimple_label): Update.
19234 (dump_gimple_bb_header): Update.
19235 * graph.c (draw_cfg_node_succ_edges): Update.
19236 * hsa-gen.c (convert_switch_statements): Update.
19237 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19238 (find_if_case_1): Update.
19239 (find_if_case_2): Update.
19240 * internal-fn.c (expand_arith_overflow_result_store): Update.
19241 (expand_addsub_overflow): Update.
19242 (expand_neg_overflow): Update.
19243 (expand_mul_overflow): Update.
19244 (expand_vector_ubsan_overflow): Update.
19245 * ipa-cp.c (good_cloning_opportunity_p): Update.
19246 * ipa-split.c (split_function): Use make_single_succ_edge.
19247 * ipa-utils.c (ipa_merge_profiles): Update.
19248 * loop-doloop.c (add_test): Update.
19249 (doloop_modify): Update.
19250 * loop-unroll.c (compare_and_jump_seq): Update.
19251 (unroll_loop_runtime_iterations): Update.
19252 * lra-constraints.c (lra_inheritance): Update.
19253 * lto-streamer-in.c (input_cfg): Update.
19254 * lto-streamer-out.c (output_cfg): Update.
19255 * mcf.c (adjust_cfg_counts): Update.
19256 * modulo-sched.c (sms_schedule): Update.
19257 * omp-expand.c (expand_omp_for_init_counts): Update.
19258 (extract_omp_for_update_vars): Update.
19259 (expand_omp_ordered_sink): Update.
19260 (expand_omp_for_ordered_loops): Update.
19261 (expand_omp_for_generic): Update.
19262 (expand_omp_for_static_nochunk): Update.
19263 (expand_omp_for_static_chunk): Update.
19264 (expand_cilk_for): Update.
19265 (expand_omp_simd): Update.
19266 (expand_omp_taskloop_for_outer): Update.
19267 (expand_omp_taskloop_for_inner): Update.
19268 * omp-simd-clone.c (simd_clone_adjust): Update.
19269 * optabs.c (expand_doubleword_shift): Update.
19270 (expand_abs): Update.
19271 (emit_cmp_and_jump_insn_1): Update.
19272 (expand_compare_and_swap_loop): Update.
19273 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19274 * predict.c (predictable_edge_p): Update.
19275 (edge_probability_reliable_p): Update.
19276 (set_even_probabilities): Update.
19277 (combine_predictions_for_insn): Update.
19278 (combine_predictions_for_bb): Update.
19279 (propagate_freq): Update.
19280 (estimate_bb_frequencies): Update.
19281 (force_edge_cold): Update.
19282 * profile-count.c (profile_count::dump): Add missing space into dump.
19283 (profile_count::debug): Add newline.
19284 (profile_count::differs_from_p): Explicitly convert to unsigned.
19285 (profile_count::stream_in): Update.
19286 (profile_probability::dump): New member function.
19287 (profile_probability::debug): New member function.
19288 (profile_probability::differs_from_p): New member function.
19289 (profile_probability::differs_lot_from_p): New member function.
19290 (profile_probability::stream_in): New member function.
19291 (profile_probability::stream_out): New member function.
19292 * profile-count.h (profile_count_quality): Rename to ...
19293 (profile_quality): ... this one.
19294 (profile_probability): New.
19295 (profile_count): Update.
19296 * profile.c (compute_branch_probabilities): Update.
19297 * recog.c (peep2_attempt): Update.
19298 * sched-ebb.c (schedule_ebbs): Update.
19299 * sched-rgn.c (find_single_block_region): Update.
19300 (compute_dom_prob_ps): Update.
19301 (schedule_region): Update.
19302 * sel-sched-ir.c (compute_succs_info): Update.
19303 * stmt.c (struct case_node): Update.
19304 (do_jump_if_equal): Update.
19305 (get_outgoing_edge_probs): Update.
19306 (conditional_probability): Update.
19307 (emit_case_dispatch_table): Update.
19308 (expand_case): Update.
19309 (expand_sjlj_dispatch_table): Update.
19310 (emit_case_nodes): Update.
19311 * targhooks.c: Update.
19312 * tracer.c (better_p): Update.
19313 (find_best_successor): Update.
19314 * trans-mem.c (expand_transaction): Update.
19315 * tree-call-cdce.c: Update.
19316 * tree-cfg.c (gimple_split_edge): Upate.
19317 (move_sese_region_to_fn): Upate.
19318 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19319 * tree-eh.c (lower_resx): Upate.
19320 (cleanup_empty_eh_move_lp): Upate.
19321 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19322 * tree-inline.c (copy_edges_for_bb): Update.
19323 (copy_cfg_body): Update.
19324 * tree-parloops.c (gen_parallel_loop): Update.
19325 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19326 (gimple_gen_time_profiler): Update.
19327 * tree-ssa-dce.c (remove_dead_stmt): Update.
19328 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19329 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19330 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19331 (unloop_loops): Update.
19332 (try_peel_loop): Update.
19333 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19334 * tree-ssa-loop-split.c (connect_loops): Update.
19335 (split_loop): Update.
19336 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19337 (hoist_guard): Update.
19338 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19339 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19340 (value_replacement): Update.
19341 * tree-ssa-reassoc.c (branch_fixup): Update.
19342 * tree-ssa-tail-merge.c (replace_block_by): Update.
19343 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19344 (create_edge_and_update_destination_phis): Update.
19345 (compute_path_counts): Update.
19346 (recompute_probabilities): Update.
19347 (update_joiner_offpath_counts): Update.
19348 (freqs_to_counts_path): Update.
19349 (duplicate_thread_path): Update.
19350 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19351 (struct switch_conv_info): Update.
19352 (gen_inbound_check): Update.
19353 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19354 (vect_do_peeling): Update.
19355 (vect_loop_versioning): Update.
19356 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19357 (optimize_mask_stores): Update.
19358 * ubsan.c (ubsan_expand_null_ifn): Update.
19359 * value-prof.c (gimple_divmod_fixed_value): Update.
19360 (gimple_divmod_fixed_value_transform): Update.
19361 (gimple_mod_pow2): Update.
19362 (gimple_mod_pow2_value_transform): Update.
19363 (gimple_mod_subtract): Update.
19364 (gimple_mod_subtract_transform): Update.
19365 (gimple_ic): Update.
19366 (gimple_stringop_fixed_value): Update.
19367 (gimple_stringops_transform): Update.
19368 * value-prof.h: Update.
19370 2017-06-29 Carl Love <cel@us.ibm.com>
19372 * config/rs6000/rs6000-c.c: Add support for built-in functions
19373 vector signed int vec_signed (vector float);
19374 vector signed long long vec_signed (vector double);
19375 vector signed int vec_signed2 (vector double, vector double);
19376 vector signed int vec_signede (vector double);
19377 vector signed int vec_signedo (vector double);
19378 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19379 instruction generator.
19380 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19381 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19382 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
19384 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
19385 vunsignede_v2df): Add define_expands.
19386 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
19387 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
19388 VEC_UNSIGNEDO): Add definitions.
19389 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19390 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
19391 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
19392 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
19393 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
19394 * config/rs6000/altivec.h (vec_signed, vec_signed2,
19395 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
19396 vec_unsignede, vec_unsignedo): Add builtin defines.
19397 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
19399 * doc/extend.texi: Update the built-in documentation file for the
19400 new built-in functions.
19402 2017-06-29 Richard Biener <rguenther@suse.de>
19404 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
19405 reduction chains to LOOP_VINFO_REDUCTIONS.
19406 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
19407 SLP reductions after processing reduction chains.
19409 2017-06-29 Nathan Sidwell <nathan@acm.org>
19411 * builtins.c (fold_builtin_FUNCTION): Use
19412 lang_hooks.decl_printable_name.
19414 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
19416 PR middle-end/81194
19417 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
19418 with only one label.
19419 * stmt.c (expand_case): Assert NCASES is greater than one.
19421 2017-06-29 Richard Biener <rguenther@suse.de>
19423 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
19425 (group_case_labels): Likewise.
19426 (find_taken_edge): Push sanity checking on val to workers...
19427 (find_taken_edge_cond_expr): ... here
19428 (find_taken_edge_switch_expr): ... and here, handle cases
19429 with just a default label.
19430 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
19431 (group_case_labels): Likewise.
19432 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
19433 group_case_labels does anything cleanup the CFG again.
19435 2017-06-29 Bin Cheng <bin.cheng@arm.com>
19437 PR tree-optimization/81196
19438 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
19439 exit condition comparing two IVs.
19441 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
19443 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
19444 profile to the dummy entry at the end of the list of architectures.
19445 * config/arm/arm-cpu-cdata.h: Regenerated.
19447 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19448 Michael Collison <michael.collison@arm.com>
19451 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
19453 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
19454 (*aarch64_reg_<mode>3_minus_mask): New pattern.
19455 (*aarch64_<optab>_reg_di3_mask2): New pattern.
19456 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
19457 of shift when the shift amount is masked with constant equal to
19458 the size of the mode.
19459 * config/aarch64/predicates.md (subreg_lowpart_operator): New
19462 2017-06-29 Martin Liska <mliska@suse.cz>
19464 * config/i386/i386.opt: Change range from [1,5] to [0,5].
19466 2017-06-29 Yury Gribov <tetra2005@gmail.com>
19469 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
19472 (edge_growth_cache_entry::edge_growth_cache_entry): New
19474 (reset_edge_growth_cache): Update to use constructor.
19476 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19478 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
19479 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19480 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19482 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
19484 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
19485 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
19487 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
19489 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
19490 (*-linux-uclibc*): Add t-uclibc tmake_file.
19491 * config/t-musl: New.
19492 * config/t-uclibc: New.
19494 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
19496 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
19498 (gen_comm_data): Emit architectural setting of arch_prof.
19499 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
19501 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
19502 (armv8-m.base, armv8-m.main): Likewise.
19503 * arm-protos.h (arm_build_target): Add profile field.
19504 (arch_option): Likewise.
19505 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
19507 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
19508 arm_active_target.profile.
19510 2017-06-28 Richard Biener <rguenther@suse.de>
19512 PR middle-end/81227
19513 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
19514 TYPE_OVERFLOW_WRAPS.
19515 * match.pd (negate_expr_p): Likewise.
19516 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
19517 fold_build2, not fold_binary.
19519 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19521 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
19522 Convert memory address to Pmode.
19523 (aarch64_print_operand): Assert MEM operands are always Pmode.
19525 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19528 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
19529 Remove redundant if.
19530 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
19531 * config/arm/aarch-common-protos.h
19532 (aarch_forward_to_shift_is_not_shifted_re): Remove.
19533 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
19535 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
19538 * multiple_target.c (create_dispatcher_calls): Set the default
19539 clone to be static, not public.
19541 2017-06-28 Richard Biener <rguenther@suse.de>
19543 * tree-vect-loop.c (vectorizable_reduction): Move special
19544 cond reduction IV var creation ...
19545 (vect_create_epilog_for_reduction): ... here. Remove induction_index
19546 parameter. Use STMT_VINFO_VECTYPE.
19547 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
19550 2017-06-28 Martin Liska <mliska@suse.cz>
19553 * ipa-visibility.c (non_local_p): Handle visibility.
19555 2017-06-28 Martin Liska <mliska@suse.cz>
19558 * common.opt: Add IntegerRange to various options.
19559 * opt-functions.awk (integer_range_info): New function.
19560 * optc-gen.awk: Add integer_range_info to cl_options struct.
19561 * opts-common.c (decode_cmdline_option): Handle
19562 CL_ERR_INT_RANGE_ARG.
19563 (cmdline_handle_error): Likewise.
19564 * opts.c (print_filtered_help): Show valid interval in
19565 when --help is provided.
19566 * opts.h (struct cl_option): Add range_min and range_max fields.
19567 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
19569 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
19571 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
19572 (x * C EQ/NE y * C): New transformation.
19574 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
19576 * genmultilib (combination_space): Accept '+' in option names.
19578 2017-06-28 Martin Liska <mliska@suse.cz>
19581 * asan.c (instrument_derefs): Bail out inner references
19582 that are hard register variables.
19584 2017-06-28 Jakub Jelinek <jakub@redhat.com>
19587 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
19588 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
19590 2017-06-28 Richard Biener <rguenther@suse.de>
19592 * tree-vectorizer.h (vect_get_vec_defs): Remove.
19593 (vect_get_slp_defs): Adjust.
19594 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
19596 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
19598 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
19599 get_initial_defs_for_reduction instead of vect_get_vec_defs.
19600 (vectorizable_reduction): Adjust.
19601 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
19603 (vect_get_slp_defs): Likewise.
19604 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
19605 (vectorizable_bswap): Adjust.
19606 (vectorizable_call): Likewise.
19607 (vectorizable_conversion): Likewise.
19608 (vectorizable_assignment): Likewise.
19609 (vectorizable_shift): Likewise.
19610 (vectorizable_operation): Likewise.
19611 (vectorizable_store): Likewise.
19612 (vectorizable_condition): Likewise.
19613 (vectorizable_comparison): Likewise.
19615 2017-06-28 Michael Collison <michael.collison@arm.com>
19618 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
19620 (arm_gen_movmemqi): Removed unused variable 'i'.
19621 Convert 'for' loop into 'while' loop.
19622 (arm_expand_prologue): Remove last unnecessary set of insn.
19623 (thumb_pop): Remove unused variable 'pushed_words'.
19624 (thumb_exit): Remove last unnecessary set of regs_to_pop.
19626 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19628 * config/s390/predicates.md: Use s390_rel_address_ok_p.
19629 * config/s390/s390-protos.h: Add prototype of
19630 s390_rel_address_ok_p.
19631 * config/s390/s390.c (s390_got_symbol): New function.
19632 (s390_rel_address_ok_p): New function.
19633 (legitimize_pic_address): Use s390_rel_address_ok_p.
19634 (s390_load_got): Use s390_got_symbol.
19635 (s390_option_override): Issue error if
19636 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
19637 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
19639 * config/s390/s390.opt: New option mpic-data-is-text-relative.
19641 2017-06-27 Andrew Pinski <apinski@cavium.com>
19643 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
19644 (X * copysign (1.0, X)): New pattern.
19645 (X * copysign (1.0, -X)): New pattern.
19646 (copysign (-1.0, CST)): New pattern.
19648 2017-06-27 Joseph Myers <joseph@codesourcery.com>
19650 * genmultilib (combination_space): Remove variable.
19651 Validate reuse rules against regular expression for any sequence
19652 of multilib options in any order.
19654 2017-06-27 Michael Collison <michael.collison@arm.com>
19656 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
19657 call aarch64_split_simd_combine.
19658 * (aarch64_combine_internal<mode>): Delete pattern.
19659 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
19660 Allow register and subreg operands.
19662 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19664 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
19665 specific need, just fallback on defaults.
19666 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
19668 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19669 Olivier Hainque <hainque@adacore.com>
19671 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
19673 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
19674 targets. Pick a default if no particular attempt applied.
19675 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
19678 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19680 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
19681 (x86_64-wrs-vxworks7): Likewise.
19683 2017-06-27 Marek Polacek <polacek@redhat.com>
19686 * ubsan.c (instrument_null): Check get_base_address's result for null.
19688 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19690 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
19692 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
19694 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
19695 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
19696 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
19697 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
19698 New function types.
19699 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
19700 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
19701 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
19702 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
19703 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
19704 BUILT_IN_FEUPDATEENV): New builtins.
19705 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
19706 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
19707 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
19708 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
19710 (builtin_structptr_types): Adjust size.
19711 * tree.c (builtin_structptr_types): Add four entries.
19713 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19714 Olivier Hainque <hainque@adacore.com>
19716 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
19717 (TLS_SYM): New local macro, forcing reference to __tls__ on
19718 link command lines for VxWorks 7 RTPs, triggering initialization
19720 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
19721 OS features TLS support, true for RTPs on VxWorks 7.
19722 * config/vxworks.c (vxworks_override_options): Setup emutls
19725 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19727 * predict.c (test_prediction_value_range): Use -1U instead of -1
19728 to avoid narrowing conversion warning.
19729 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
19730 to avoid narrowing conversion warning.
19731 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
19733 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
19735 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
19737 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
19738 64bit configurations.
19739 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
19740 (SIZE_TYPE): Likewise.
19741 * config/vxworks.c (vxworks_emutls_var_fields): Use
19742 long_unsigned_type_node instead of unsigned_type_node as the offset
19743 field type, which is "pointer" mode in emutls.c.
19745 2017-06-27 Jakub Jelinek <jakub@redhat.com>
19748 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
19750 PR middle-end/81207
19751 * gimple-fold.c (replace_call_with_call_and_fold): Handle
19752 gimple_vuse copying separately from gimple_vdef copying.
19754 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19756 * value-prof.c (free_hist): Remove call to memset and the enclosing if
19759 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19760 Olivier Hainque <hainque@adacore.com>
19762 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
19763 for all vxworks7 targets.
19764 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
19765 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
19766 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
19767 variations for VX6/VX7 and 32/64bits later on in ...
19768 (VXWORKS_LIB_SPEC): Leverage new macros.
19769 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
19770 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
19772 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
19774 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
19775 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
19777 2017-06-26 Carl Love <cel@us.ibm.com>
19779 * config/rs6000/rs6000-c.c: Add support for built-in functions
19780 vector bool char vec_reve (vector bool char);
19781 vector signed char vec_reve (vector signed char);
19782 vector unsigned char vec_reve (vector unsigned char);
19783 vector bool int vec_reve (vector bool int);
19784 vector signed int vec_reve (vector signed int);
19785 vector unsigned int vec_reve (vector unsigned int);
19786 vector bool long long vec_reve (vector bool long long);
19787 vector signed long long vec_reve (vector signed long long);
19788 vector unsigned long long vec_reve (vector unsigned long long);
19789 vector bool short vec_reve (vector bool short);
19790 vector signed short vec_reve (vector signed short);
19791 vector double vec_reve (vector double);
19792 vector float vec_reve (vector float);
19793 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
19794 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
19795 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
19796 (altivec_vreve): New pattern.
19797 * config/rs6000/altivec.h (vec_reve): New define.
19798 * doc/extend.texi (vec_rev): Update the built-in documentation file
19799 for the new built-in functions.
19801 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19803 PR tree-optimization/71815
19804 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
19806 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
19808 (slsr_process_phi): Likewise.
19809 (replace_uncond_cands_and_profitable_phis): Don't replace a
19810 multiply candidate with a stride of 1 (copy or cast).
19811 (phi_incr_cost): Call uses_consumed_by_stmt rather than
19813 (lowest_cost_path): Likewise.
19814 (total_savings): Likewise.
19816 2017-06-26 Richard Biener <rguenther@suse.de>
19819 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
19820 Use def_builtin_pure for all gather builtins.
19822 2017-06-26 Richard Biener <rguenther@suse.de>
19824 PR tree-optimization/81203
19825 * tree-tailcall.c (find_tail_calls): Do not move stmts into
19826 non-dominating BBs.
19828 2017-06-26 Marek Polacek <polacek@redhat.com>
19831 * doc/invoke.texi: Document -Wmultistatement-macros.
19833 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19835 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19836 arm_neon_ok_no_float_abi effective target.
19838 2017-06-26 Richard Biener <rguenther@suse.de>
19840 PR tree-optimization/80928
19841 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19842 (copy_bbs): Set BB_DUPLICATED flag early.
19843 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19845 (execute_on_shrinking_pred): Likewise.
19846 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19847 BB_DUPLICATED blocks.
19848 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19849 iterate over all PHIs considering removal of *gsi.
19851 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19853 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19856 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19858 * config/rs6000/rs6000-string.c: (expand_block_clear,
19859 do_load_for_compare, select_block_compare_mode,
19860 compute_current_alignment, expand_block_compare,
19861 expand_strncmp_align_check, expand_strn_compare,
19862 expand_block_move, rs6000_output_load_multiple)
19863 Move functions related to string/block move/compare
19864 to a separate file.
19865 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19866 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19867 for this function which is now used in two files.
19868 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19869 * config.gcc: Add rs6000-string.o to extra_objs for
19870 targets powerpc*-*-* and rs6000*-*-*.
19872 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19875 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19876 32-bit, since indexed is not valid for DImode.
19877 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19878 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19879 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19880 (define_peephole2 for Altivec d-form store): Likewise.
19883 * multiple_target.c (create_dispatcher_calls): Only create the
19884 dispatcher call if the function is the default clone of a
19885 versioned function.
19887 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
19889 PR middle-end/80902
19890 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
19891 a call, force the call to not be a tail call.
19893 2017-06-23 Jeff Law <law@redhat.com>
19895 * doc/contrib.texi: Add entry for Steven Pemberton's work on
19898 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
19900 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19901 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19902 handling for early expansion of vector shifts (sl,sr,sra,rl).
19903 (builtin_function_type): Add vector shift right instructions
19904 to the unsigned argument list.
19906 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19908 rtl-optimizatoin/79286
19909 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19910 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19911 trap. PIC register plus a const unspec without offset can never trap.
19913 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19915 * tree.h (builtin_structptr_type): New type.
19916 (builtin_structptr_types): Declare new array.
19917 * tree.c (builtin_structptr_types): New array.
19918 (free_lang_data, build_common_tree_nodes): Use it.
19920 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19923 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19926 2017-06-22 Matt Turner <mattst88@gmail.com>
19928 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19929 Lake models to skylake case. Assume skylake for unknown
19930 models with clflushopt.
19932 2017-06-22 Jeff Law <law@redhat.com>
19934 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
19935 frame sizes that do not satisfy aarch64_uimm12_shift.
19937 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
19939 * profile-count.h (apply_probability,
19940 apply_scale, probability_in): Fix checks for zero.
19942 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19944 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
19945 * doc/cppdiropts.texi (-I @var{dir}): Document it.
19947 2016-06-22 Richard Biener <rguenther@suse.de>
19949 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19950 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
19951 REDUC_MAX_EXPR support.
19952 (vectorizable_reduction): Likewise.
19953 (vect_create_epilog_for_reduction): Likewise.
19955 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
19957 * match.pd (A / (1 << B) -> A >> B): New.
19958 * generic-match-head.c: Include optabs-tree.h.
19959 * gimple-match-head.c: Likewise.
19960 * optabs-tree.h (target_supports_op_p): New.
19961 * optabs-tree.c (target_supports_op_p): New.
19963 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19965 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
19966 $gcc_cv_ld --help output.
19967 (gcc_cv_ld_demangle): Likewise.
19968 (gcc_cv_ld_eh_frame_hdr): Likewise.
19969 (gcc_cv_ld_pie): Likewise.
19970 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
19971 (gcc_cv_ld_buildid): Likewise.
19972 (gcc_cv_ld_sysroot): Likewise.
19973 (ld_bndplt_support): Likewise.
19974 (ld_pushpopstate_support): Likewise.
19975 * configure: Regenerate.
19976 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
19978 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19981 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19982 operands indexes to avoid gap between operands and match_dups.
19984 2017-06-21 Andrew Pinski <apinski@cavium.com>
19986 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
19987 Increment Arith_shift and Arith_shift_reg by 1.
19988 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
19990 * config/aarch64/aarch64.c (thunderx_tunings): Enable
19991 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
19992 (aarch64_strip_extend): Add new argument and test for it.
19993 (aarch64_cheap_mult_shift_p): New function.
19994 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
19995 add a cost if it is true.
19996 Update calls to aarch64_strip_extend.
19997 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
19999 2017-06-21 Andrew Pinski <apinski@cavium.com>
20001 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20003 (thunderxt88): Likewise.
20004 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20005 (thunderx_prefetch_tune): New variable.
20006 (thunderx2t99_prefetch_tune): Update for the correct values.
20007 (thunderxt88_tunings): New variable.
20008 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20009 generic_prefetch_tune.
20010 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20012 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20014 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20015 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20016 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20017 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20018 (aarch64_atomic_cas<mode>, GPI): Likewise.
20020 2017-06-21 Martin Liska <mliska@suse.cz>
20022 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20023 statements on cold and hot labels.
20024 * predict.c (tree_estimate_probability_bb): Remove the
20025 prediction from this place.
20027 2017-06-21 Martin Liska <mliska@suse.cz>
20029 PR tree-optimization/79489
20030 * gimplify.c (maybe_add_early_return_predict_stmt): New
20032 (gimplify_return_expr): Call the function.
20033 * predict.c (tree_estimate_probability_bb): Remove handling
20035 * predict.def: Update comment about early return predictor.
20036 * gimple-predict.h (is_gimple_predict): New function.
20037 * predict.def: Change default value of early return to 66.
20038 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20040 * passes.def: Put pass_strip_predict_hints to the beginning of
20043 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20045 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20046 FUNCTION_DECL declarations.
20047 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20049 (dwaf2out_decl): Likewise.
20050 * godump.c (go_early_global_decl): Skip call to the real debug hook
20051 for FUNCTION_DECL declarations.
20052 * passes.c (rest_of_decl_compilation): Skip call to the
20053 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20054 -fdump-go-spec is passed.
20056 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20058 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20059 Reorder for compactness.
20060 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20061 (def_builtin_pure, def_builtin_pure2): New functions.
20062 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20064 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20066 * match.pd (nop_convert): New predicate.
20067 ((A +- CST1) +- CST2): Allow some NOP conversions.
20069 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20072 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20073 with ctors/dtors if GOVD_SHARED is set.
20075 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20077 * config/aarch64/aarch64.md (movti_aarch64):
20078 Emit mov rather than orr.
20079 (movtf_aarch64): Likewise.
20080 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20081 Emit mov rather than orr.
20083 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20085 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20086 Swap alternatives, make integer dup more expensive.
20088 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20090 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20091 Return true for non-tls symbols.
20093 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20095 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20096 (cortex-a75): Likewise.
20097 (cortex-a75.cortex-a55): Likewise.
20098 * config/aarch64/aarch64-tune.md: Regenerate.
20099 * doc/invoke.texi (-mtune): Document new values for -mtune.
20101 2017-06-21 Tom de Vries <tom@codesourcery.com>
20103 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20104 stack_size feature.
20105 (Effective-Target Keywords, Other attributes): Suggest using
20106 dg-add-options stack_size feature to get stack limit in stack_size
20107 effective target documentation.
20109 2017-06-21 Julian Brown <julian@codesourcery.com>
20110 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20112 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20113 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20114 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20116 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20117 attribute type list for neon_multiply.
20118 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20119 attribute type list for neon_multiply.
20120 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20121 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20122 attribute type list for neon_multiply.
20123 * config/arm/types.md (crypto_pmull): Add.
20124 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20125 attribute type list.
20127 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20129 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20132 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20134 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20135 to make sure not to dereference a NULL cost_classes_ptr pointer.
20137 2017-06-20 Carl Love <cel@us.ibm.com>
20139 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20140 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20141 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20142 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20143 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20144 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20145 VMULOSW): New enum "unspec" values.
20146 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20147 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20148 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20149 altivec_vmulosw): New patterns.
20150 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20151 VMULOSW): Add definitions.
20153 2017-06-20 Julia Koval <julia.koval@intel.com>
20155 * config/i386/i386.c: Fix rounding expand for new pattern.
20156 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20158 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20160 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20161 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20163 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20165 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20168 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20170 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20171 architecture, then by implementer ID.
20172 * config/aarch64/aarch64-tune.md: Regenerate.
20174 2017-06-20 Richard Biener <rguenther@suse.de>
20176 PR middle-end/81097
20177 * fold-const.c (split_tree): Fold to type before negating.
20179 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20181 * diagnostic-show-locus.c
20182 (selftest::test_fixit_deletion_affecting_newline): New function.
20183 (selftest::diagnostic_show_locus_c_tests): Call it.
20185 2017-06-20 Andreas Schwab <schwab@suse.de>
20188 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20191 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20193 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20194 __ARM_FEATURE_COPROC according to support.
20196 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20198 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20199 Rewritten to avoid overflow for > 32-bit pointers.
20202 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20203 by removing enum keyword.
20204 (ubsan_type_descriptor): Likewise. Formatting fix.
20207 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20208 splitter): Require TARGET_SSE2 in the condition.
20210 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20213 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20214 for doing vector set of SFmode on ISA 3.0.
20215 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20216 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20218 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20219 SFmode value into a V4SF variable that was extracted from another
20220 V4SF variable without converting the element to double precision
20221 and back to single precision vector format.
20222 (vsx_insert_extract_v4sf_p9_2): Likewise.
20224 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20226 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20227 in UWHI to avoid undefined overflow.
20230 * ubsan.h (enum ubsan_encode_value_phase): New.
20231 (ubsan_encode_value): Change second argument to
20232 enum ubsan_encode_value_phase with default value of
20233 UBSAN_ENCODE_VALUE_GENERIC.
20234 * ubsan.c (ubsan_encode_value): Change second argument to
20235 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20236 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20237 create_tmp_var_raw instead of create_tmp_var and use a
20239 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20240 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20241 ubsan_encode_value callers.
20244 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20245 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20246 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20248 2017-06-19 Richard Biener <rguenther@suse.de>
20250 PR middle-end/81118
20251 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20252 estimates if we changed anything.
20254 2017-06-19 Richard Biener <rguenther@suse.de>
20256 PR tree-optimization/80887
20257 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20258 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20259 simplified lookups, then reset mprts_hook.
20260 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20262 (try_to_simplify): Likewise.
20264 2017-06-19 Martin Liska <mliska@suse.cz>
20267 * gimplify.c (gimplify_switch_expr):
20268 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20270 2017-06-19 Martin Liska <mliska@suse.cz>
20272 * doc/install.texi: Document that PGO runs in 4 stages.
20274 2017-06-19 Martin Liska <mliska@suse.cz>
20277 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20278 to dispatcher function name.
20279 * multiple_target.c (replace_function_decl): New function.
20280 (create_dispatcher_calls): Redirect both edges and references.
20282 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20284 * profile-count.c (profile_count::dump): Dump quality.
20285 (profile_count::differs_from_p): Update for unsigned val.
20286 * profile-count.h (profile_count_quality): New enum.
20287 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20289 2017-06-19 Richard Biener <rguenther@suse.de>
20291 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20292 struct function as arg.
20293 (estimate_numbers_of_iterations): Export overload with loop arg.
20294 (free_numbers_of_iterations_estimates_loop): Use an overload of
20295 free_numbers_of_iterations_estimates instead.
20296 * tree-cfg.c (remove_bb): Adjust.
20297 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20298 * tree-parloops.c (gen_parallel_loop): Likewise.
20299 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20301 (tree_unroll_loops_completely): Likewise.
20302 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20303 Use an overload instead and export.
20304 (estimated_loop_iterations): Adjust.
20305 (max_loop_iterations): Likewise.
20306 (likely_max_loop_iterations): Likewise.
20307 (estimate_numbers_of_iterations): Take struct function as arg
20309 (loop_exits_before_overflow): Adjust.
20310 (free_numbers_of_iterations_estimates_loop): Use an overload.
20311 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20312 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20314 2017-06-19 Richard Biener <rguenther@suse.de>
20317 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20318 RANGE_EXPR conservatively.
20320 2017-06-16 Carl Love <cel@us.ibm.com>
20322 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20323 definitions for vec_float, vec_float2, vec_floato,
20324 vec_floate built-ins.
20325 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20326 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20328 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20329 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20330 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20331 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20332 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20333 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20334 vec_floato): Add builtin defines.
20335 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20336 Update the built-in documentation file for the new built-in
20339 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20341 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20342 (mthumb): Mark as the negative of -marm.
20344 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20346 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20348 (ARM Options, -mtune): Document that this accepts the same
20349 extension options as -mcpu.
20350 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20352 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20354 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20355 permitted extensions.
20357 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20359 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20360 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20361 (armv8-m.main): Add option +nodsp.
20362 * config/arm/arm-cpu-cdata.h: Regenerated.
20364 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20366 * config/arm/t-fuchsia: New file.
20367 * config.gcc (arm*-*-fuchsia*): Use it.
20369 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20371 * config/arm/t-symbian: Rewrite for new option infrastructure.
20373 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20375 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20376 (MULTILIB_REQUIRED): Likewise.
20378 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20380 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20381 (MULTILIB_RESUE): Likewise.
20382 (MULTILIB_MATCHES): Likewise.
20383 (MULTLIB_REQUIRED): Likewise.
20385 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20387 * config/arm/t-rtems: Rewrite for new option framework.
20389 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20391 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
20392 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
20393 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
20394 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
20395 * config/arm/t-multilib: ... here.
20396 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
20397 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
20398 armv7-a and armv8*-a when A-profile libraries have not been built.
20399 * config/arm/t-rmprofile: Rewrite.
20401 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20403 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
20404 with a backslash. Remove the backslash after substituting unescaped
20406 * doc/fragments.texi (MULTILIB_REUSE): Document it.
20408 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20410 * config.gcc: (arm*-*-*): When building a-profile libraries, force
20411 the driver to pass through the default setting of -mfloat-abi.
20412 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
20414 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
20415 (all_feat_combs): New rule.
20416 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
20418 * config/arm/t-aprofile: Rewrite.
20420 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20422 * config/arm/arm.h (FPUTYPE_AUTO): Define.
20423 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
20424 fpu is not specified by the user/command-line.
20425 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
20426 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
20427 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
20428 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
20429 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
20430 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
20432 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20434 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
20435 * config/arm/t-arm-elf: Rewritten.
20437 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20439 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
20440 have some floating-point instructions.
20441 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
20442 (TARGET_MAYBE_HARD_FLOAT): New macro.
20443 * config/arm/arm-builtins.c (arm_init_builtins): Use
20444 TARGET_MAYBE_HARD_FLOAT.
20445 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
20447 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20449 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
20450 (configargs.h): Include it.
20451 (arm_print_hint_for_fpu_option): New function.
20452 (arm_parse_fpu_option): New function.
20453 (candidate_extension): New class.
20454 (arm_canon_for_multilib): New function.
20455 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
20456 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
20457 (ARCH_CANONICAL_SPECS): New macro.
20458 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
20460 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20462 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
20463 are set after handling multilib fragments. Set target_cpu_default2
20466 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20468 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
20470 (arm*-*-*): Set target_cpu_default2 to a quoted string.
20471 * config/arm/parsecpu.awk (check_cpu): Validate any extension
20473 (check_arch): Likewise.
20474 * config/arm/arm.c (arm_configure_build_target): Handle
20475 TARGET_CPU_DEFAULT being a string constant. Scan any feature
20476 options in the default.
20478 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20480 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
20481 when an option is an alias of another.
20482 * config/arm/parsecpu.awk (optalias): New parser token.
20483 (gen_comm_data): Mark non-alias options as such. Emit entries
20484 for extension aliases.
20485 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
20486 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
20487 (armv6kz, armv6zk, armv6t2): Likewise.
20488 (armv7): Make vfpv3-d16 an alias.
20489 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
20491 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
20492 Sort in canonical order.
20493 (armv8-a): Sort in canonical order.
20494 (armv8.1-a, armv8.2-a): Likewise.
20495 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
20497 (cortex-a9): Sort in canonical order.
20498 * config/arm/arm.c (selftests.h): Include it.
20499 (arm_test_cpu_arch_data): New function.
20500 (arm_run_self_tests): New function.
20501 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
20502 (targetm): Move declaration to the end of the file.
20503 * arm-cpu-cdata.h: Regenerated.
20505 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20507 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
20508 call to target_mode_check describing the type of option passed.
20509 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
20510 (arm_target_thumb_only): Use arm_parse_arch_option_name or
20511 arm_parse_cpu_option_name to match parameters against list of
20513 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
20514 arm_arch_core_flags data structure.
20515 * config/arm/arm-cpu_cdata.h: Regenerated.
20517 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20519 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
20521 (arm_print_hint_for_cpu_option): Likewise.
20522 (arm_print_hint_for_arch_option): Likewise.
20523 (arm_parse_cpu_option_name): Likewise.
20524 (arm_parse_arch_option_name): Likewise.
20525 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
20526 of entries in the all_fpus list.
20527 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
20528 (arm_parse_cpu_option_name): Declare.
20529 (arm_parse_arch_option_name): Declare.
20530 (arm_parse_option_features): Declare.
20531 (arm_intialize_isa): Declare.
20532 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
20534 (gen_comm_data): ... here. Make definitions non-static.
20535 * config/arm/arm-cpu-data.h: Regenerated.
20536 * config/arm/arm-cpu-cdata.h: Regenerated.
20538 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20540 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
20541 (cpu_arch_extension): New structure.
20542 (cpu_arch_option, arch_option, cpu_option): New structures.
20543 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
20544 architecture types.
20545 (gen_data): Generate new format data tables.
20546 * config/arm/arm.c (cpu_tune): New structure.
20547 (cpu_option, processors): Delete.
20548 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
20549 (arm_print_hint_for_cpu_option): ... this and ...
20550 (arm_print_hint_for_arch_option): ... this.
20551 (arm_parse_arch_cpu_name): Delete. Replace with ...
20552 (arm_parse_cpu_option_name): ... this and ...
20553 (arm_parse_arch_option_name): ... this.
20554 (arm_unrecognized_feature): Change type of target parameter to
20556 (arm_parse_arch_cpu_features): Delete. Replace with ...
20557 (arm_parse_option_features): ... this.
20558 (arm_configure_build_target): Rework to use new configuration data
20560 (arm_print_tune_info): Rework for new configuration data tables.
20561 * config/arm/arm-cpu-data.h: Regenerated.
20562 * config/arm/arm-cpu.h: Regenerated.
20564 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20566 * Makefile.in (OBJS): Move sbitmap.o from here ...
20567 (OBJS-libcommon): ... to here.
20569 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20571 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
20572 (ISA_ALL_CRYPTO): New macro.
20573 (ISA_ALL_SIMD): New macro
20574 (ISA_ALL_FP): New macro.
20575 * config/arm/arm.c (fpu_bitlist): Update initializer.
20576 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
20578 (arm9e): Add fpu. Add option for nofp
20579 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
20580 (arm926ej-s, arm1026ej-s): Likewise.
20581 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
20582 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
20583 neon-fp16, neon-vfpv4, nofp and nosimd.
20584 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
20585 (cortex-a8): Add fpu. Add option for nofp.
20586 (cortex-a9): Add fpu. Add options for nosimd and nofp.
20587 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
20588 (cortex-r4f): Add fpu.
20589 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
20590 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
20592 (cortex-r8): Likewise.
20593 (cortex-m4): Add fpu. Add option for nofp.
20594 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
20595 (cortex-a17.cortex-a7): Likewise.
20596 (cortex-a32): Add fpu. Add options for crypto and nofp.
20597 (cortex-a35, cortex-a53): Likewise.
20598 (cortex-a57): Add fpu. Add option for crypto.
20599 (cortex-a72, cortex-a73): Likewise.
20600 (exynos-m1): Likewise.
20601 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
20602 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
20603 (cortex-m33): Add fpu. Add option for nofp.
20604 * config/arm/arm-cpu-cdata.h: Regenerated
20605 * config/arm/arm-cpu-data.h: Regenerated.
20607 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20609 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
20610 (armv5te, armv5tej): Likewise.
20611 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
20612 (armv7): Add options fp and vfpv3-d16.
20613 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
20614 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
20616 (armv7ve): Likewise.
20617 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
20618 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
20619 (armv8-a): Add nocrypto option.
20620 (armv8.1-a, armv8.2-a): Likewise.
20621 (armv8-m.main): add options fp, fp.dp and nofp.
20623 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20625 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
20627 (armv8-a+crc): Delete.
20628 (armv8.1-a): Add options simd, crypto and nofp.
20629 (armv8.2-a): Add options fp16, simd, crypto and nofp.
20630 (armv8.2-a+fp16): Delete.
20631 (armv8-m.main): Add option dsp.
20632 (armv8-m.main+dsp): Delete.
20633 (cortex-a8): Add fpu. Add nofp option.
20634 (cortex-a9): Add fpu. Add nofp and nosimd options.
20635 * config/arm/parsecpu.awk (gen_data): Generate option tables and
20636 link to main cpu and architecture data structures.
20637 (gen_comm_data): Only put isa attributes from the main architecture
20639 (option): New statement for architecture and CPU entries.
20640 * arm.c (struct cpu_option): New structure.
20641 (struct processors): Add entry for options.
20642 (arm_unrecognized_feature): New function.
20643 (arm_parse_arch_cpu_name): Ignore any characters after the first
20645 (arm_parse_arch_cpu_feature): New function.
20646 (arm_configure_build_target): Separate out any CPU and architecture
20647 features and parse separately. Don't error out if -mfpu=auto is
20648 used with only an architecture string.
20649 (arm_print_asm_arch_directives): New function.
20650 (arm_file_start): Call it.
20651 * config/arm/arm-cpu-cdata.h: Regenerated.
20652 * config/arm/arm-cpu-data.h: Likewise.
20653 * config/arm/arm-tables.opt: Likewise.
20655 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20657 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
20658 assembler when it is not -mfpu=auto.
20660 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20662 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
20663 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
20664 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
20665 (ASM_CPU_SPEC): Rewrite.
20666 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
20667 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
20668 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
20669 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
20670 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
20671 copied string is NUL-terminated. Also strip any characters prefixed
20673 (arm_rewrite_selected_arch): New function.
20674 (arm_rewrite_march): New function.
20676 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20678 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
20679 (x_arm_cpu_string, x_arm_tune_string): Likewise.
20680 (march, mcpu, mtune): Convert to string-based options.
20681 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
20682 (arm_parse_arch_cpu_name): New function.
20683 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
20684 identify selected architecture or CPU.
20685 (arm_option_save): New function.
20686 (TARGET_OPTION_SAVE): Redefine.
20687 (arm_option_restore): Restore string options.
20688 (arm_option_print): Print string options.
20690 2017-06-16 Martin Sebor <msebor@redhat.com>
20692 PR tree-optimization/80933
20693 PR tree-optimization/80934
20694 * builtins.c (fold_builtin_3): Do not handle bcmp here.
20695 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
20696 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
20697 (gimple_fold_builtin): Call them.
20699 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20701 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
20702 as unlikely; update profile.
20704 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20706 * predict.c (force_edge_cold): Handle declaring edges impossible
20709 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20711 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
20713 (try_unroll_loop_completely): Fix reporting.
20715 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20717 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
20719 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
20722 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
20723 if given a non-constant argument for an intrinsic which requires a
20726 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20728 * profile.c (compare_freqs): New function.
20729 (branch_prob): Sort edge list.
20730 (find_spanning_tree): Assume that the list is priority sorted.
20732 2017-06-16 Richard Biener <rguenther@suse.de>
20734 PR tree-optimization/81090
20735 * passes.def (pass_record_bounds): Remove.
20736 * tree-pass.h (make_pass_record_bounds): Likewise.
20737 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
20738 make_pass_record_bounds): Likewise.
20739 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
20740 not free niter estimates at the beginning but at the end.
20741 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
20743 2017-06-16 Richard Biener <rguenther@suse.de>
20745 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
20746 initializer to workaround ICE in host GCC 4.8.
20748 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
20750 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
20752 (clone_inlined_nodes): Update.
20754 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20756 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
20757 prefetch settings, and enable prefetching by default at -O3.
20759 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20761 * config/aarch64/aarch64.c (aarch64_override_options_internal):
20762 Set flag_prefetch_loop_arrays according to tuning data.
20764 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
20766 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
20767 New tune structure.
20768 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
20769 [Unrelated to main purpose of the patch] Place the pointer field last
20770 to enable type checking errors when tune structure are wrongly merged.
20771 * config/aarch64/aarch64.c (generic_prefetch_tune,)
20772 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
20773 (thunderx2t99_prefetch_tune): New tune constants.
20774 (tune_params *_tunings): Update all tunings (no functional change).
20775 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
20776 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
20777 from tunings structures.
20779 2017-06-16 Jakub Jelinek <jakub@redhat.com>
20782 * ubsan.c (instrument_null): Add T argument, use it instead
20783 of computing it based on IS_LHS.
20784 (instrument_object_size): Likewise.
20785 (pass_ubsan::execute): Adjust instrument_null and
20786 instrument_object_size callers to pass gimple_get_lhs or
20787 gimple_assign_rhs1 result to it. Use instrument_null instead of
20788 calling get_base_address and instrument_mem_ref. Handle
20789 aggregate call arguments for object-size sanitization.
20791 2017-06-16 Yury Gribov <tetra2005@gmail.com>
20793 PR tree-optimization/81089
20794 * tree-vrp.c (is_masked_range_test): Validate operands of
20797 2017-06-15 Martin Sebor <msebor@redhat.com>
20800 * dumpfile.c (dump_register): Avoid calling memset to initialize
20801 a class with a default ctor.
20802 * gcc.c (struct compiler): Remove const qualification.
20803 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
20804 * hash-table.h: Ditto.
20805 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
20807 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
20808 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
20810 * params.h (struct param_info): Make struct members non-const.
20811 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
20812 with default initialization.
20813 * vec.h (vec_copy_construct, vec_default_construct): New helper
20815 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
20816 with vec_copy_construct.
20817 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
20818 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
20819 * doc/invoke.texi (-Wclass-memaccess): Document.
20821 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20823 * emit-rtl.h (is_leaf): Update comment about local
20824 register allocator.
20826 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20829 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
20830 for a variable to have a section before checking if the section has a
20832 Set section to.persistent if persistent attribute is set.
20833 Warn if .persistent attribute is used on an automatic variable.
20835 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20837 PR rtl-optimization/80474
20838 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20840 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20842 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20845 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20847 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20848 (casesi): Emit bounds checking as RTL.
20849 (casesi_internal_mips16_<mode>): Remove bounds checking.
20851 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20853 * config/xtensa/xtensa.c (xtensa_option_override): Append
20854 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20855 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20856 xtensa_doloop_hooks): Define unconditionally.
20857 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20859 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20860 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20861 for it in xtensa_option_override.
20862 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20863 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20865 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20867 * doc/cppopts.texi: Document '-' special value to -MF.
20869 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20871 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20872 (cortex_a53_fconst): Likewise.
20873 (cortex_a53_fpmul): Likewise.
20874 (cortex_a53_f_load_64): Likewise.
20875 (cortex_a53_f_load_many): Likewise.
20876 (cortex_a53_advsimd_alu): Likewise.
20877 (cortex_a53_advsimd_alu_q): Likewise.
20878 (cortex_a53_advsimd_mul): Likewise.
20879 (cortex_a53_advsimd_mul_q): Likewise.
20880 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20881 Add missing fmul, r2f_cvt and fconst cases.
20883 2017-06-14 Richard Biener <rguenther@suse.de>
20885 PR middle-end/81088
20886 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
20888 (fold_binary_loc): When associating do not treat pre-existing
20889 TREE_OVERFLOW on literal constants as a reason to allow
20890 TREE_OVERFLOW on associated literal constants.
20892 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20894 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
20895 (MASK_FEATURES): New macro.
20896 * config/sparc/sparc.c (sparc_option_override): Remove the special
20897 handling of -mfpu and generalize it to all MASK_FEATURES switches.
20899 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20901 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20902 a division of 0 if non-call exceptions are enabled.
20904 2017-06-14 Andrew Pinski <apinski@cavium.com>
20905 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20908 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20909 Improve vector initialization code gen for only variable case.
20911 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20913 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20915 2017-06-14 Richard Biener <rguenther@suse.de>
20917 PR tree-optimization/81083
20918 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20921 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20923 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20924 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20925 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20926 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20927 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20928 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20930 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20932 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20933 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
20935 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20937 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
20939 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20941 * config/rs6000/t-rtems: Don't handle SPE.
20943 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20945 * config/rs6000/t-linux: Don't handle SPE.
20947 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20949 * config/rs6000/eabispe.h: Delete file.
20951 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20953 * config/rs6000/t-spe: Delete file.
20955 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20957 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
20958 (rs6000_legitimate_offset_address_p): Return false for anything in
20959 V2SImode or V2SFmode.
20961 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20963 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
20964 except V2SF and V2SI. Rearrange the vector modes, and add comments.
20965 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
20967 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
20968 (rs6000_legitimate_offset_address_p): Ditto.
20969 (rs6000_emit_move): Ditto.
20970 (rs6000_init_builtins): Remove V4HI_type_node.
20972 2017-06-13 Martin Liska <mliska@suse.cz>
20975 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
20976 (gate_asan): Likewise.
20977 * asan.h (asan_no_sanitize_address_p): Remove the function.
20978 (sanitize_flags_p): New function.
20979 * builtins.def: Fix coding style.
20980 * common.opt: Use renamed enum value.
20981 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20982 * doc/extend.texi: Document no_sanitize attribute.
20983 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20984 to SANITIZE_UNDEFINED_NONDEFAULT.
20985 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20986 * gimple-fold.c (optimize_atomic_compare_exchange_p):
20987 Use sanitize_flags_p.
20988 * gimplify.c (gimplify_function_tree): Likewise.
20989 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
20990 * opts.c (parse_no_sanitize_attribute): New function.
20991 (common_handle_option): Use renamed enum value.
20992 * opts.h (parse_no_sanitize_attribute): Declare.
20993 * tree.c (sanitize_flags_p): New function.
20994 * tree.h: Declared here.
20995 * tsan.c: Use sanitize_flags_p.
20996 * ubsan.c (ubsan_expand_null_ifn): Likewise.
20997 (instrument_mem_ref): Likewise.
20998 (instrument_bool_enum_load): Likewise.
20999 (do_ubsan_in_current_function): Remove the function.
21000 (pass_ubsan::execute): Use sanitize_flags_p.
21001 * ubsan.h: Remove do_ubsan_in_current_function
21002 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21003 (dump_function_to_file): Use it here.
21005 2017-06-13 Martin Jambor <mjambor@suse.cz>
21007 PR tree-optimization/80803
21008 PR tree-optimization/81063
21009 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21010 (propagate_subaccesses_across_link): Enqueue subtree whenever
21011 necessary instead of relying on the caller.
21013 2017-06-13 Martin Jambor <mjambor@suse.cz>
21015 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21016 that have a first_link.
21017 (sort_and_splice_var_accesses): Do not check first_link before
21019 (subtree_mark_written_and_enqueue): Likewise.
21020 (propagate_all_subaccesses): Likewise and do not stop at first
21021 parent with a first_link.
21023 2017-06-13 Martin Jambor <mjambor@suse.cz>
21025 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21028 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21030 * match.pd: New pattern.
21032 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21034 * tree-vrp.c (is_masked_range_test): New function.
21035 (register_edge_assert_for): Determine ranges for
21038 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21040 PR tree-optimization/67328
21041 * fold-const.c (maskable_range_p): New function.
21042 (build_range_check): Generate bittests if possible.
21044 2017-06-13 Martin Liska <mliska@suse.cz>
21046 * gimple-pretty-print.c (dump_probability): Add new argument.
21047 (dump_edge_probability): Dump both probability and count.
21048 (dump_gimple_label): Likewise.
21049 (dump_gimple_bb_header): Likewise.
21051 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21054 * config/avr/avr-devices.c: Fix indentation.
21055 * config/avr/gen-avr-mmcu-specs.c: Dito.
21057 2017-06-13 Richard Biener <rguenther@suse.de>
21059 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21060 instead get vector type from stmt_info.
21061 (vectorizable_reduction): Adjust. Remove dead code.
21063 2017-06-13 Richard Biener <rguenther@suse.de>
21065 PR middle-end/81065
21066 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21067 case of C * (x * C2 + C3).
21068 (fold_addr_of_array_ref_difference): Properly fold index difference.
21070 2017-06-12 David S. Miller <davem@davemloft.net>
21073 * config/sparc/sparc.md (return expander): Emit frame blockage if
21074 function uses alloca.
21076 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21078 * combine.c (make_field_assignment): Check len rather than the mode
21079 precision when calling force_to_mode.
21081 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21083 Support multilibs and devices that see flash in RAM address range.
21086 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21087 (avr_mcu_t) <flash_pm_offset>: New field.
21088 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21089 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21090 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21091 (AVR_TINY_PM_OFFSET): Remove macro.
21092 * config/avr/avr.opt (-mshort-calls): New option.
21093 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21094 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21095 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21096 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21097 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21098 instead of avr_arch->have_jmp_call.
21099 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21100 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21101 avr_arch->flash_pm_offset to define.
21102 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21103 new field flash_pm_offset. Add entry for avrxmega3.
21104 (avr_texinfo): Add entry for avrxmega3.
21105 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21106 attiny212, attiny214,
21107 attiny412, attiny414, attiny416, attiny417,
21108 attiny814, attiny816, attiny817,
21109 attiny1614, attiny1616, attiny1617,
21110 attiny3214, attiny3216, attiny3217.
21111 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21112 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21113 (avr_print_operand_address) [AVR_TINY]: Same.
21114 (avr_asm_init_sections) <readonly_data_section>: Only patch
21115 callback if avr_arch->flash_pm_offset = 0.
21116 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21117 for rodata if avr_arch->flash_pm_offset != 0.
21118 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21119 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21120 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21121 (m_options): Append opt_rcall.
21122 (m_dirnames): Append dir_rcall.
21123 * config/avr/t-multilib: Regenerate.
21125 * configure.ac [target=avr]: Check whether avrxmega3 default
21126 linker description file works as needed.
21127 * configure: Regenerate.
21128 * doc/avr-mmcu.texi: Regenerate.
21129 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21130 <__AVR_ARCH__>: Document avrxmega3 and 103.
21131 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21132 <__AVR_SHORT_CALLS__>: Document it.
21133 <__AVR_PM_BASE_ADDRESS__>: Document it.
21134 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21135 (AVR Variable Attributes) <progmem>: Document this is
21136 not needed for avrxmega3.
21137 (AVR Named Address Spaces) <__flash>: Dito.
21139 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21141 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21143 2017-06-12 Doug Rupp <rupp@adacore.com>
21145 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21146 Append vxworks-stdint.h to the tm_file list.
21147 * config/vxworks-stdint.h: New file.
21149 2017-06-12 Martin Liska <mliska@suse.cz>
21151 PR tree-optimization/81041
21152 * tree-profile.c (gimple_gen_ic_func_profiler):
21153 Create an extra BB in profile-generate
21154 (gimple_gen_time_profiler): Likewise.
21156 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21158 PR tree-optimization/81003
21159 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21160 (update_range_test): Use it instead of force_gimple_operand_gsi.
21162 2017-06-12 Richard Biener <rguenther@suse.de>
21164 PR tree-optimization/81053
21165 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21166 with backedge value not defined in loop. Simplify def stmt
21169 2017-06-11 Tom de Vries <tom@codesourcery.com>
21172 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21174 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21175 nvptx_cannot_force_const_mem.
21177 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21179 * opts.c (finish_options): Move test for flag_split_stack after
21180 it has been initialized.
21182 2017-06-11 Jason Merrill <jason@redhat.com>
21184 * tree.h (id_equal): New.
21185 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21186 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21187 instead of strcmp of IDENTIFIER_POINTER.
21189 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21191 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21192 (mark_all_inlined_calls_cdtor): Fix formating.
21193 (inline_transform): Rescale profile before inlining.
21195 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21197 * cgraph.h (cgraph_edge::clone): Update prototype.
21198 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21199 (cgraph_node::create_clone): Update.
21200 (cgraph_node::create_version_clone): Update.
21201 * tree-inline.c (copy_bb): Update.
21202 (expand_call_inline): Update.
21204 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21206 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21207 factored out from ...
21208 (rs6000_emit_prologue): ... here.
21210 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21212 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21213 factored out from ...
21214 (rs6000_emit_prologue): ... here.
21216 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21218 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21220 (handle_missing_profiles): Fix computation of tp_first_run.
21221 (counts_to_freqs): Do not touch freqs when count is 0.
21223 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21225 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21228 2017-06-10 Tom de Vries <tom@codesourcery.com>
21230 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21231 attributes): Document signal effective target.
21233 2017-06-10 Tom de Vries <tom@codesourcery.com>
21235 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21236 Document effective target stack_size.
21238 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21240 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21241 to the edit_context if they can be auto-applied.
21243 2017-06-9 Ian Lance Taylor <iant@golang.org>
21245 * opts.c (finish_options): If -fsplit-stack, disable implicit
21246 -forder-blocks-and-partition.
21247 * doc/invoke.texi (Optimize Options): Document that when using
21248 -fsplit-stack -forder-blocks-and-partition is not implicitly
21251 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21253 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21254 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21255 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21256 * builtins.def (abort, trap, unreachable): Declare cold.
21257 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21258 * tree-core.h (ECF_COLD): New.
21259 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21260 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21262 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21264 * predict.c (unlikely_executed_stmt_p): Cleanup.
21266 2017-06-09 Richard Biener <rguenther@suse.de>
21268 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21269 model if the ref is always written to.
21271 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21273 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21275 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21277 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21280 2017-06-09 Tom de Vries <tom@codesourcery.com>
21283 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21284 "target cannot support label values" when encountering LABEL_REF.
21286 2017-06-09 Martin Liska <mliska@suse.cz>
21288 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21289 (gimple_gen_ic_func_profiler): Emit direct comparison
21290 of __gcov_indirect_call_callee with NULL.
21291 (gimple_gen_time_profiler): Change probability from
21292 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21294 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21296 * profile.c (edge_gcov_counts): Turn to pointer.
21297 (compute_branch_probabilities, compute_branch_probabilities): Update.
21298 (branch_prob): Do not clear edge_gcov_count.
21299 * profile.h (edge_gcov_counts): Turn to pointer.
21300 (edge_gcov_count): Update.
21302 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21304 * gimple.h (gimple_check_failed): Mark cold.
21306 2017-06-09 Richard Biener <rguenther@suse.de>
21308 PR tree-optimization/66623
21309 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21310 refactor check_reduction into two parts, properly computing
21311 whether we have to check reduction validity for outer loop
21314 2017-06-09 Richard Biener <rguenther@suse.de>
21316 PR tree-optimization/79483
21317 * graphite-scop-detection.c (order): New global.
21318 (get_order): Compute bb to order mapping that satisfies code
21319 generation constraints.
21320 (cmp_pbbs): New helper.
21321 (build_scops): Start domwalk at entry block, sort generated
21324 2017-06-09 Richard Biener <rguenther@suse.de>
21326 PR middle-end/81007
21327 * ipa-polymorphic-call.c
21328 (ipa_polymorphic_call_context::restrict_to_inner_class):
21329 Skip FIELD_DECLs with error_mark_node type.
21330 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21333 2017-06-09 Martin Liska <mliska@suse.cz>
21335 * predict.c (struct branch_predictor): New struct.
21336 (test_prediction_value_range): New test.
21337 (predict_c_tests): New function.
21338 * selftest-run-tests.c (selftest::run_tests): Run the function.
21339 * selftest.h: Declare new tests.
21341 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21344 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21345 gen_add3_insn did not fail.
21346 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21347 r0, construct that number in a temporary reg and add that reg to r0.
21348 If asked to put the result in r0 as well, fail.
21350 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21352 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21353 for early expansion of vec_eqv.
21355 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21357 PR middle-end/81005
21358 * ubsan.c (instrument_null): Avoid pointless code temporary.
21359 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21361 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21365 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21368 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21371 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21373 * predict.c (unlikely_executed_edge_p): Move ahead.
21374 (probably_never_executed_edge_p): Use it.
21376 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21378 PR middle-end/79988
21379 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21380 gimple_call_builtin_p call.
21382 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21384 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
21385 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
21386 rtl_check_failed_type2, rtl_check_failed_code1,
21387 rtl_check_failed_code2, rtl_check_failed_code_mode,
21388 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
21389 rtvec_check_failed_bounds, rtl_check_failed_flag,
21390 _fatal_insn_not_found, _fatal_insn): Likewise.
21391 * tree.h (tree_contains_struct_check_failed,
21392 tree_check_failed, tree_not_check_failed,
21393 tree_class_check_failed, tree_range_check_failed,
21394 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
21395 tree_vec_elt_check_failed, phi_node_elt_check_failed,
21396 tree_operand_check_failed, omp_clause_check_failed,
21397 omp_clause_operand_check_failed, omp_clause_range_check_failed):
21400 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21402 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
21403 flag_branch_probabilities.
21404 * ipa-inline.c (edge_badness): Likewise.
21405 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
21406 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
21407 * predict.c (maybe_hot_frequency_p): Likewise.
21408 (probably_never_executed): Likewise.
21409 * sched-ebb.c (schedule_ebbs): Likewise.
21410 * sched-rgn.c (find_single_block_region): Likewise.
21411 * tracer.c (tail_duplicate): Likewise.
21413 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21415 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
21416 longer requires x_flag_profile_use.
21418 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21420 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
21421 instead of flag_reorder_blocks_and_partition.
21422 * dbxout.c (dbxout_function_end): Likewise.
21423 * dwarf2out.c (gen_subprogram_die): Likewise.
21424 * haifa-sched.c (sched_create_recovery_edges): Likewise.
21425 * hw-doloop.c (reorg_loops): Likewise.
21426 * varasm.c (assemble_start_function,
21427 assemble_end_function): Likewise.
21428 (decide_function_section): Do not check for
21429 flag_reorder_blocks_and_partition.
21431 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21433 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21435 (chkp_get_hard_register_fake_addr_expr): Ditto.
21436 (chkp_build_addr_expr): Add check for hard reg case.
21437 (chkp_parse_array_and_component_ref): Ditto.
21438 (chkp_find_bounds_1): Ditto.
21439 (chkp_process_stmt): Don't generate bounds store for
21442 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21444 * predict.c (maybe_hot_bb_p): Do not check profile status.
21445 (maybe_hot_edge_p): Likewise.
21446 (probably_never_executed): Check for zero counts even if profile
21448 (unlikely_executed_edge_p): New function.
21449 (unlikely_executed_stmt_p): New function.
21450 (unlikely_executed_bb_p): New function.
21451 (set_even_probabilities): Use unlikely predicates.
21452 (combine_predictions_for_bb): Likewise.
21453 (predict_paths_for_bb): Likewise.
21454 (predict_paths_leading_to_edge): Likewise.
21455 (determine_unlikely_bbs): New function.
21456 (estimate_bb_frequencies): Use it.
21457 (compute_function_frequency): Use zero counts even if profile is
21459 * profile-count.h: Fix typo.
21461 2017-08-08 Julia Koval <julia.koval@intel.com>
21463 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
21464 _mm512_mask_cvtsepi16_storeu_epi8,
21465 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
21466 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
21467 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
21468 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
21469 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
21470 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
21471 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
21472 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
21473 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
21474 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
21475 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
21476 __builtin_ia32_pmovuswb256mem_mask,
21477 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
21478 __builtin_ia32_pmovwb512mem_mask): New builtins.
21480 2017-08-08 Julia Koval <julia.koval@intel.com>
21482 PR target/73350,80862
21483 * config/i386/subst.md (round): Fix round pattern.
21484 * config/i386/i386.c (ix86_erase_embedded_rounding):
21485 Fix erasing rounding for the fixed pattern.
21487 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21489 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
21491 2017-06-08 Martin Liska <mliska@suse.cz>
21493 PR gcov-profile/80911
21494 * gcov.c (block_info::block_info): New constructor.
21496 2017-06-07 Carl Love <cel@us.ibm.com>
21498 * config/rs6000/rs6000-c: The return type of the following
21499 built-in functions was implemented as int not long long. Fix sign
21500 of return value for the unsigned version of vec_mulo and vec_mule.
21501 vector unsigned long long vec_bperm (vector unsigned long long,
21502 vector unsigned char)
21503 vector signed long long vec_mule (vector signed int,
21505 vector unsigned long long vec_mule (vector unsigned int,
21506 vector unsigned int)
21507 vector signed long long vec_mulo (vector signed int,
21509 vector unsigned long long vec_mulo (vector unsigned int,
21510 vector unsigned int)
21511 * doc/extend.texi: Fix the documentation for the built-in
21514 2017-06-07 Carl Love <cel@us.ibm.com>
21517 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
21520 2017-06-07 Carl Love <cel@us.ibm.com>
21522 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
21523 support, Generate doublehv for signed int/float for BE case only.
21525 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
21527 * doc/invoke.texi (mcx16): Rewrite.
21529 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21531 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
21532 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
21533 *mov<mode>_softfloat, and an anonymous splitter): Use
21534 nonimmediate_operand instead of rs6000_nonimmediate_operand.
21536 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21538 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
21540 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21541 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
21542 (rs6000_debug_reg_global): Adjust.
21543 (rs6000_init_hard_regno_mode_ok): Adjust.
21544 (rs6000_dbx_register_number): Adjust.
21545 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
21546 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21547 Remove SPE_ACC and SPEFSCR.
21548 (REG_ALLOC_ORDER): Ditto.
21549 (FRAME_POINTER_REGNUM): Change to 111.
21550 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
21551 (REG_CLASS_NAMES): Ditto.
21552 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
21553 (REGISTER_NAMES): Ditto.
21554 (ADDITIONAL_REG_NAMES): Ditto.
21555 (rs6000_reg_names): Ditto.
21556 * config/rs6000/rs6000.md: Renumber some register number
21559 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21561 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
21563 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21564 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
21566 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
21567 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21568 Delete the SPE high registers.
21569 (REG_ALLOC_ORDER): Ditto.
21570 (enum reg_class): Remove SPE_HIGH_REGS.
21571 (REG_CLASS_NAMES): Ditto.
21572 (REG_CLASS_CONTENTS): Delete the SPE high registers.
21573 (REGISTER_NAMES): Ditto.
21574 (rs6000_reg_names): Ditto.
21575 * doc/tm.texi.in: Remove SPE as example.
21576 * doc/tm.texi: Regenerate.
21578 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21580 * config/rs6000/8540.md (ppc8540_brinc): Delete.
21581 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
21582 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
21583 * config/rs6000/rs6000.md (type): Remove "brinc".
21585 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21587 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
21588 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
21589 * config/rs6000/linuxspe.h: Delete file.
21590 * config/rs6000/rs6000.md: Don't include spe.md.
21591 * config/rs6000/spe.h: Delete file.
21592 * config/rs6000/spe.md: Delete file.
21593 * config/rs6000/t-rs6000: Remove spe.md.
21595 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21597 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
21598 (reg_or_none500mem_operand): Delete.
21599 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
21600 instead of reg_or_none500mem_operand.
21602 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21604 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21605 handling of SPE flags.
21606 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
21608 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21610 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
21612 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
21613 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
21614 paired_divv2sf3): Similar.
21615 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
21616 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
21617 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
21619 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
21620 Rename the paired_* instruction patterns.
21621 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
21623 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
21624 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
21625 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
21626 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
21627 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
21628 PAIRED_VECTOR_MODE.
21629 (struct machine_function): Delete field spe_insn_chain_scanned_p.
21630 (spe_func_has_64bit_regs_p): Delete.
21631 (spe_expand_predicate_builtin): Delete.
21632 (spe_expand_evsel_builtin): Delete.
21633 (TARGET_DWARF_REGISTER_SPAN): Do not define.
21634 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
21635 (invalid_e500_subreg): Delete.
21636 (rs6000_legitimize_address): Always force_reg op2 as well, for
21637 paired single memory accesses.
21638 (rs6000_member_type_forces_blk): Delete.
21639 (rs6000_spe_function_arg): Delete.
21640 (rs6000_expand_unop_builtin): Delete SPE handling.
21641 (rs6000_expand_binop_builtin): Ditto.
21642 (spe_expand_stv_builtin): Delete.
21643 (bdesc_2arg_spe): Delete.
21644 (spe_expand_builtin): Delete.
21645 (spe_expand_predicate_builtin): Delete.
21646 (spe_expand_evsel_builtin): Delete.
21647 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
21648 (spe_init_builtins): Delete.
21649 (spe_func_has_64bit_regs_p): Delete.
21650 (savres_routine_name): Delete "info" parameter. Adjust callers.
21651 (rs6000_emit_stack_reset): Ditto.
21652 (rs6000_dwarf_register_span): Delete.
21653 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
21654 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
21655 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
21657 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
21659 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
21660 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
21661 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
21662 mulv2sf3, divv2sf3): Delete expanders.
21664 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21666 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
21668 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21670 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
21671 * config/rs6000/rs6000.c: Ditto.
21673 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21675 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
21676 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
21677 comparison_operator.
21679 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21681 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
21682 * config/rs6000/rs6000.opt: Ditto.
21683 * config/rs6000/t-rtems: Ditto.
21685 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21687 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
21688 TARGET_E500_SINGLE by 0, simplify.
21689 * config/rs6000/rs6000.c: Ditto.
21690 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
21691 (spe_build_register_parallel): Delete.
21692 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
21693 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
21694 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
21695 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
21696 (E500_CONVERT): Delete.
21697 * config/rs6000/spe.md: Remove many patterns and all define_constants.
21699 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21701 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
21702 * config/rs6000/dfp.md: Ditto.
21703 (negdd2, *negdd2_fpr): Merge.
21704 (absdd2, *absdd2_fpr): Merge.
21705 (negtd2, *negtd2_fpr): Merge.
21706 (abstd2, *abstd2_fpr): Merge.
21707 * config/rs6000/e500.h: Delete file.
21708 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
21709 TARGET_FPRS by 1 and simplify.
21710 * config/rs6000/rs6000-c.c: Ditto.
21711 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
21712 TARGET_DF_SPE by 0.
21713 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
21715 * config/rs6000/rs6000.md: Ditto.
21716 (floatdidf2, *floatdidf2_fpr): Merge.
21717 (move_from_CR_gt_bit): Delete.
21718 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
21719 (E500_CR_IOR_COMPARE): Delete.
21720 (All patterns that require !TARGET_FPRS): Delete.
21721 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
21723 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21725 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
21727 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21729 * graphds.c (add_edge): Intitialize edge's attached data.
21730 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
21731 pointer parameter. Call pointed function on each edge during
21732 graph traversing. Skip traversing the edge when the function
21734 (graphds_dfs, graphds_scc): Ditto.
21735 (for_each_edge): New parameter. Pass the new parameter to callback
21737 * graphds.h (skip_edge_callback): New function pointer type.
21738 (graphds_dfs, graphds_scc): New function pointer parameter.
21739 (graphds_edge_callback, for_each_edge): New parameter.
21741 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21743 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
21744 out code checking if runtime alias check is possible to below ...
21745 Call the new function.
21746 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
21747 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
21749 2017-06-07 Marek Polacek <polacek@redhat.com>
21752 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
21753 TYPE_OVERFLOW_WRAPS check.
21755 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21757 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
21758 if versioning is required.
21759 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
21760 peeling with the check for versioning.
21762 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21764 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
21765 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
21766 Set true to new parameter if new ssa variable is defined.
21767 (vect_gen_vector_loop_niters): Refactor. Set range information
21768 for the new vector loop bound variable.
21769 (vect_do_peeling): Ditto.
21771 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21773 * tree-affine.c (ssa.h): Include header file.
21774 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
21775 has wrapping overflow behavior.
21777 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21779 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
21781 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21783 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
21784 (tree_to_aff_combination): ... here.
21786 2017-06-07 Bin Cheng <bin.cheng@arm.com>
21788 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
21789 reg_pressure model function.
21790 (ivopts_global_cost_for_size): Delete.
21791 (determine_set_costs, iv_ca_recount_cost): Call new model function
21792 ivopts_estimate_reg_pressure.
21794 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21796 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
21797 expensive than udiv. Remove floating point cases from mod.
21799 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21801 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
21802 Increase idiv cost.
21804 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21806 * config/aarch64/aarch64.md
21807 (copysignsf3): Fix mask generation.
21809 2017-06-07 Jakub Jelinek <jakub@redhat.com>
21811 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
21813 (class dump_manager): Add register_dumps method.
21814 * dumpfile.c: Include langhooks.h.
21815 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
21816 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
21817 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
21818 (dump_manager::dump_register): Start with 512 entries instead of 32.
21819 (dump_manager::register_dumps): New method.
21820 * toplev.c (general_init): Instead of invoking register_dumps
21821 langhook, invoke register_dumps method on the dump manager.
21822 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
21825 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
21827 * doc/md.texi: Clarify the restrictions on a define_insn condition.
21828 Say that # requires an associated define_split to exist, and that
21829 the define_split must be suitable for use after register allocation.
21831 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21833 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21834 (compute_outgoing_frequencies): Also initialize zero counts.
21835 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21836 around loops; preserve more of profile when nothing changes.
21838 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21840 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21842 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21843 * config/arm/arm-cpu-cdata.h: Regenerate.
21844 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21845 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21846 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21847 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21848 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21850 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21852 * config/arm/t-rmprofile: Likewise.
21853 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21855 2017-06-06 David S. Miller <davem@davemloft.net>
21858 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21859 blockage if function uses alloca.
21861 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21863 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21864 New "uid" fields to hold pretty-print IDs of group and ref.
21865 Memory references are now identified as <group_id>:<ref_id>
21866 instead of using [random] addresses.
21867 (dump_mem_details): Simplify, no functional change.
21868 (dump_mem_ref): Simplify and make output more concise.
21869 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21870 (find_or_create_group): Initialize group uid.
21871 (record_ref): Initialize ref uid. Improve debug output.
21872 (prune_group_by_reuse, should_issue_prefetch_p,)
21873 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21874 (mark_nontemporal_store, determine_loop_nest_reuse):
21875 Improve debug output.
21877 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21879 * dbgcnt.def (prefetch): New debug counter.
21880 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21881 (schedule_prefetches): Stop issueing prefetches if debug counter
21884 2017-06-06 Tom de Vries <tom@codesourcery.com>
21886 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21887 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
21889 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21891 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
21892 Use aarch64_reg_or_zero predicate for operand 4.
21893 (aarch64_compare_and_swap<mode> define_insn_and_split):
21894 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
21895 (aarch64_store_exclusive<mode>): Likewise for operand 2.
21897 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21899 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
21900 (arm_compute_save_core_reg_mask): This.
21901 (thumb1_compute_save_reg_mask): Rename into ...
21902 (thumb1_compute_save_core_reg_mask): This.
21903 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21904 (arm_compute_frame_layout): Likewise.
21906 2017-06-06 Richard Biener <rguenther@suse.de>
21908 PR tree-optimization/80974
21909 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21910 keep or clear leaders SSA info.
21912 2017-06-06 Tom de Vries <tom@codesourcery.com>
21914 * config/nvptx/nvptx.c (split_mode_p): New function.
21915 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21917 2017-06-06 Tom de Vries <tom@codesourcery.com>
21919 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21921 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21924 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21927 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21929 * shrink-wrap.c (handle_simple_exit): Update profile.
21930 (try_shrink_wrapping): Upate profile.
21932 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21934 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
21935 (tree_guess_outgoing_edge_probabilities): New.
21936 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
21937 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
21939 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21941 * ipa-split.c (split_function): Initialize return bb profile.
21943 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21945 * profile.c (compute_branch_probabilities): Also initialize
21946 EXIT_BLOCK profile.
21948 2017-06-06 Richard Biener <rguenther@suse.de>
21950 PR tree-optimization/80928
21951 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
21952 (vect_analyze_loop_operations): Properly guard analysis for
21954 (vect_transform_loop): Likewise.
21955 (vect_analyze_loop_2): Also reset SLP type on PHIs.
21956 (vect_model_induction_cost): Do not cost for pure SLP.
21957 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
21958 of induction in inner loop vectorization.
21959 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
21960 (vect_get_and_check_slp_defs): Handle vect_induction_def.
21961 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
21963 (vect_analyze_slp_cost_1): Cost induction.
21964 (vect_detect_hybrid_slp_stmts): Handle PHIs.
21965 (vect_get_slp_vect_defs): Likewise.
21966 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
21967 (vect_transform_stmt): Handle SLP reductions.
21968 * tree-vectorizer.h (vectorizable_induction): Adjust.
21970 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21972 * config/rs6000/rs6000.c (make_resolver_func): Update
21973 init_lowered_empty_function call.
21975 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
21977 * doc/invoke.texi: Document the -fprofile-abs-path option.
21978 * common.opt (fprofile-abs-path): New option.
21979 * gcov-io.h (gcov_write_filename): Declare.
21980 * gcov-io.c (gcov_write_filename): New function.
21981 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21982 * profile.c (output_location): Likewise.
21984 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21986 * shring-wrap.c: Revert accidental commit.
21988 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
21990 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
21992 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21994 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
21996 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
21997 profile in callgraph edge.
21998 * profile-count.h (apply_probability): If THIS is 0, then result is 0
21999 (apply_scale): Likewise.
22000 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22001 Also scale profile when inlining function with zero profile.
22002 (initialize_cfun): Update exit block profile even when it is zero.
22003 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22004 when profile is read.
22006 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22008 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22009 (CLONE_*): New constants to define the processors we can generate
22010 code for with the target_clone attribute.
22011 (rs6000_clone_map): New array to identify which clone processors
22012 the current program is running on.
22013 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22014 target_clone attribute.
22015 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22016 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22017 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22018 (cpu_expand_builtin): Add support for target_clone attribute.
22019 (rs6000_valid_attribute_p): Allow "default" attribute.
22020 (get_decl_name): New debug function to simplify printing the
22021 current function name in debugging statements.
22022 (rs6000_clone_priority): New functions to support the target_clone
22023 attribute, and be able to generate code to switch between ISA 2.05
22024 through ISA 3.0 (power6 through power9).
22025 (rs6000_compare_version_priority): Likewise.
22026 (rs6000_get_function_versions_dispatcher): Likewise.
22027 (make_resolver_func): Likewise.
22028 (add_condition_to_bb): Likewise.
22029 (dispatch_function_versions): Likewise.
22030 (rs6000_generate_version_dispatcher_body): Likewise.
22031 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22032 (fusion_gpr_load_p): Fix a spacing issue.
22033 * doc/extend.texi (Common Function Attributes): Document that the
22034 PowerPC supports the target_clone attribute.
22036 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22038 * config/arm/arm.h: explain F symbol found in description of ARM
22039 register allocation in its legend.
22041 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22043 * config/mips/frame-header-opt.c: Include profile-count.h.
22044 * config/riscv/riscv.c: Include profile-count.h
22046 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22048 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22050 (sm_set_flag_if_changed): Add bbs field.
22051 (execute_sm_if_changed_flag_set): Pass BBS.
22052 (execute_sm): Update.
22054 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22056 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22059 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22061 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22062 (peephole2): New peephole2 to emit the above.
22063 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22065 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22067 * config/aarch64/aarch64.c (define_peephole2 above
22068 *sub_<shift>_<mode>): New peephole.
22070 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22072 * config/i386/i386.c (make_resolver_func): Update.
22073 * Makefile.in: Add profile-count.h and profile-count.o
22074 * auto-profile.c (afdo_indirect_call): Update to new API.
22075 (afdo_set_bb_count): Update.
22076 (afdo_propagate_edge): Update.
22077 (afdo_propagate_circuit): Update.
22078 (afdo_calculate_branch_prob): Update.
22079 (afdo_annotate_cfg): Update.
22080 * basic-block.h: Include profile-count.h
22081 (struct edge_def): Turn count to profile_count.
22082 (struct basic_block_def): Likewie.
22083 (REG_BR_PROB_BASE): Move to profile-count.h
22084 (RDIV): Move to profile-count.h
22085 * bb-reorder.c (max_entry_count): Turn to profile_count.
22086 (find_traces): Update.
22087 (rotate_loop):Update.
22088 (connect_traces):Update.
22089 (sanitize_hot_paths):Update.
22090 * bt-load.c (migrate_btr_defs): Update.
22091 * cfg.c (RDIV): Remove.
22092 (init_flow): Use alloc_block.
22093 (alloc_block): Uninitialize count.
22094 (unchecked_make_edge): Uninitialize count.
22095 (check_bb_profile): Update.
22096 (dump_edge_info): Update.
22097 (dump_bb_info): Update.
22098 (update_bb_profile_for_threading): Update.
22099 (scale_bbs_frequencies_int): Update.
22100 (scale_bbs_frequencies_gcov_type): Update.
22101 (scale_bbs_frequencies_profile_count): New.
22102 * cfg.h (update_bb_profile_for_threading): Update.
22103 (scale_bbs_frequencies_profile_count): Declare.
22104 * cfgbuild.c (compute_outgoing_frequencies): Update.
22105 (find_many_sub_basic_blocks): Update.
22106 * cfgcleanup.c (try_forward_edges): Update.
22107 (try_crossjump_to_edge): Update.
22108 * cfgexpand.c (expand_gimple_tailcall): Update.
22109 (construct_exit_block): Update.
22110 * cfghooks.c (verify_flow_info): Update.
22111 (dump_bb_for_graph): Update.
22112 (split_edge): Update.
22113 (make_forwarder_block): Update.
22114 (duplicate_block): Update.
22115 (account_profile_record): Update.
22116 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22117 (get_estimated_loop_iterations): Update.
22118 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22119 (single_likely_exit): Update.
22120 * cfgloopmanip.c (scale_loop_profile): Update.
22122 (set_zero_probability): Update.
22123 (lv_adjust_loop_entry_edge): Update.
22124 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22125 (purge_dead_edges): Update.
22126 (rtl_account_profile_record): Update.
22127 * cgraph.c (cgraph_node::create): Uninitialize count.
22128 (symbol_table::create_edge): Uninitialize count.
22129 (cgraph_update_edges_for_call_stmt_node): Update.
22130 (cgraph_edge::dump_edge_flags): Update.
22131 (cgraph_node::dump): Update.
22132 (cgraph_edge::maybe_hot_p): Update.
22133 * cgraph.h: Include profile-count.h
22134 (create_clone), create_edge, create_indirect_edge): Update.
22135 (cgraph_node): Turn count to profile_count.
22136 (cgraph_edge0: Likewise.
22137 (make_speculative, clone): Update.
22138 (create_edge): Update.
22139 (init_lowered_empty_function): Update.
22140 * cgraphclones.c (cgraph_edge::clone): Update.
22141 (duplicate_thunk_for_node): Update.
22142 (cgraph_node::create_clone): Update.
22143 * cgraphunit.c (cgraph_node::analyze): Update.
22144 (cgraph_node::expand_thunk): Update.
22145 * final.c (dump_basic_block_info): Update.
22146 * gimple-streamer-in.c (input_bb): Update.
22147 * gimple-streamer-out.c (output_bb): Update.
22148 * graphite.c (print_global_statistics): Update.
22149 (print_graphite_scop_statistics): Update.
22150 * hsa-brig.c: Include basic-block.h.
22151 * hsa-dump.c: Include basic-block.h.
22152 * hsa-gen.c (T sum_slice): Update.
22153 (convert_switch_statements):Update.
22154 * hsa-regalloc.c: Include basic-block.h.
22155 * ipa-chkp.c (chkp_produce_thunks): Update.
22156 * ipa-cp.c (struct caller_statistics): Update.
22157 (init_caller_stats): Update.
22158 (gather_caller_stats): Update.
22159 (ipcp_cloning_candidate_p): Update.
22160 (good_cloning_opportunity_p): Update.
22161 (get_info_about_necessary_edges): Update.
22162 (dump_profile_updates): Update.
22163 (update_profiling_info): Update.
22164 (update_specialized_profile): Update.
22165 (perhaps_add_new_callers): Update.
22166 (decide_about_value): Update.
22167 (ipa_cp_c_finalize): Update.
22168 * ipa-devirt.c (struct odr_type_warn_count): Update.
22169 (struct decl_warn_count): Update.
22170 (struct final_warning_record): Update.
22171 (possible_polymorphic_call_targets): Update.
22172 (ipa_devirt): Update.
22173 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22174 * ipa-icf.c (sem_function::merge): Update.
22175 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22176 * ipa-inline.c (compute_uninlined_call_time): Update.
22177 (compute_inlined_call_time): Update.
22178 (want_inline_small_function_p): Update.
22179 (want_inline_self_recursive_call_p): Update.
22180 (edge_badness): Update.
22181 (lookup_recursive_calls): Update.
22182 (recursive_inlining): Update.
22183 (inline_small_functions): Update.
22184 (dump_overall_stats): Update.
22185 (dump_inline_stats): Update.
22186 * ipa-profile.c (ipa_profile_generate_summary): Update.
22187 (ipa_propagate_frequency): Update.
22188 (ipa_profile): Update.
22189 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22190 * ipa-utils.c (ipa_merge_profiles): Update.
22191 * loop-doloop.c (doloop_modify): Update.
22192 * loop-unroll.c (report_unroll): Update.
22193 (unroll_loop_runtime_iterations): Update.
22194 * lto-cgraph.c (lto_output_edge): Update.
22195 (lto_output_node): Update.
22196 (input_node): Update.
22197 (input_edge): Update.
22198 (merge_profile_summaries): Update.
22199 * lto-streamer-in.c (input_cfg): Update.
22200 * lto-streamer-out.c (output_cfg): Update.
22201 * mcf.c (create_fixup_graph): Update.
22202 (adjust_cfg_counts): Update.
22203 (sum_edge_counts): Update.
22204 * modulo-sched.c (sms_schedule): Update.
22205 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22206 * predict.c (maybe_hot_count_p): Update.
22207 (probably_never_executed): Update.
22208 (dump_prediction): Update.
22209 (combine_predictions_for_bb): Update.
22210 (propagate_freq): Update.
22211 (handle_missing_profiles): Update.
22212 (counts_to_freqs): Update.
22213 (rebuild_frequencies): Update.
22214 (force_edge_cold): Update.
22215 * predict.h: Include profile-count.h
22216 (maybe_hot_count_p, counts_to_freqs): UPdate.
22217 * print-rtl-function.c: Do not include cfg.h
22218 * print-rtl.c: Include basic-block.h
22219 * profile-count.c: New file.
22220 * profile-count.h: New file.
22221 * profile.c (is_edge_inconsistent): Update.
22222 (correct_negative_edge_counts): Update.
22223 (is_inconsistent): Update.
22224 (set_bb_counts): Update.
22225 (read_profile_edge_counts): Update.
22226 (compute_frequency_overlap): Update.
22227 (compute_branch_probabilities): Update; Initialize and deinitialize
22229 (branch_prob): Update.
22230 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22231 (edge_gcov_count): New.
22232 (bb_gcov_count): New.
22233 * shrink-wrap.c (try_shrink_wrapping): Update.
22234 * tracer.c (better_p): Update.
22235 * trans-mem.c (expand_transaction): Update.
22236 (ipa_tm_insert_irr_call): Update.
22237 (ipa_tm_insert_gettmclone_call): Update.
22238 * tree-call-cdce.c: Update.
22239 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22240 (gimple_duplicate_sese_tail): Update.
22241 (gimple_account_profile_record): Update.
22242 (execute_fixup_cfg): Update.
22243 * tree-inline.c (copy_bb): Update.
22244 (copy_edges_for_bb): Update.
22245 (initialize_cfun): Update.
22246 (freqs_to_counts): Update.
22247 (copy_cfg_body): Update.
22248 (expand_call_inline): Update.
22249 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22250 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22251 (try_unroll_loop_completely): Update.
22252 (try_peel_loop): Update.
22253 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22254 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22255 * tree-ssa-loop-split.c (connect_loops): Update.
22256 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22257 * tree-ssa-reassoc.c (branch_fixup): Update.
22258 * tree-ssa-tail-merge.c (replace_block_by): Update.
22259 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22260 (compute_path_counts): Update.
22261 (update_profile): Update.
22262 (recompute_probabilities): Update.
22263 (update_joiner_offpath_counts): Update.
22264 (estimated_freqs_path): Update.
22265 (freqs_to_counts_path): Update.
22266 (clear_counts_path): Update.
22267 (ssa_fix_duplicate_block_edges): Update.
22268 (duplicate_thread_path): Update.
22269 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22270 (struct switch_conv_info): Update.
22271 * tree-tailcall.c (decrease_profile): Update.
22272 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22273 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22274 * value-prof.c (check_counter): Update.
22275 (gimple_divmod_fixed_value): Update.
22276 (gimple_mod_pow2): Update.
22277 (gimple_mod_subtract): Update.
22278 (gimple_ic_transform): Update.
22279 (gimple_stringop_fixed_value): Update.
22280 * value-prof.h (gimple_ic): Update.
22282 2017-06-02 Carl Love <cel@us.ibm.com>
22284 * config/rs6000/rs6000-c: Add support for built-in functions
22285 vector double vec_doublee (vector signed int);
22286 vector double vec_doublee (vector unsigned int);
22287 vector double vec_doublee (vector float);
22288 vector double vec_doubleh (vector signed int);
22289 vector double vec_doubleh (vector unsigned int);
22290 vector double vec_doubleh (vector float);
22291 vector double vec_doublel (vector signed int);
22292 vector double vec_doublel (vector unsigned int);
22293 vector double vec_doublel (vector float);
22294 vector double vec_doubleo (vector signed int);
22295 vector double vec_doubleo (vector unsigned int);
22296 vector double vec_doubleo (vector float);.
22297 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22298 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22300 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22301 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22302 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22304 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22305 vec_doublel, vec_doubleh.
22306 * doc/extend.texi: Update the built-in documentation file for the
22307 new built-in functions.
22309 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22312 * ipa-inline-analysis.c (free_growth_caches): Set
22313 edge_removal_hook_holder to NULL after removing it.
22315 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22317 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22318 comparision with zero.
22320 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22321 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22322 for early expansion of vec_min and vec_max builtins.
22323 (builtin_function_type): Add min/max unsigned variants to those
22324 identified as having unsigned arguments.
22326 2017-06-02 Olivier Hainque <hainque@adacore.com>
22328 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22330 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22332 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22333 Use VALL_F16 iterator rather than VALL.
22335 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22337 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22338 Emit CBNZ inside loop when doing a strong exchange and comparing
22339 against zero. Generate the CC flags after the loop.
22341 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22343 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22344 (dl_section_ref): New.
22345 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22346 On AIX, append an expression to subtract the size of the
22347 section length to dl_section_ref.
22349 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22351 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22352 for early expansion of vector absolute builtins.
22354 2017-06-02 Richard Biener <rguenther@suse.de>
22356 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22357 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22359 2017-06-02 Richard Biener <rguenther@suse.de>
22361 PR tree-optimization/80948
22362 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22363 stmt order as well.
22365 2017-06-02 Richard Biener <rguenther@suse.de>
22367 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22369 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22370 for inductions as relevant.
22372 2017-06-02 Richard Biener <rguenther@suse.de>
22374 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22375 (vectorizable_induction): ... this. Remove dead code.
22377 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22379 * builtins. (expand_builtin_alloca): Remove second parameter and
22380 infer its value from the first parameter instead.
22381 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22383 2017-06-02 Jakub Jelinek <jakub@redhat.com>
22385 PR rtl-optimization/80903
22386 * loop-doloop.c (add_test): Unshare sequence.
22388 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
22390 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
22392 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
22394 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
22396 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
22397 xlogue_layout::get_instance, logue_layout::xlogue_layout,
22398 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
22399 (xlogue_layout::get_stub_rtx): Make static.
22400 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
22401 (xlogue_layout::compute_stub_managed_regs): Rename to...
22402 (xlogue_layout::count_stub_managed_regs): ...this.
22403 (xlogue_layout::is_stub_managed_reg): New function.
22404 (xlogue_layout::m_stub_names): Rename to...
22405 (xlogue_layout::s_stub_names): ...this, make static.
22406 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
22407 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
22408 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
22409 xlogue_layout::s_stub_names): Instantiate statics.
22410 (stub_managed_regs): Remove.
22411 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
22412 (disable_call_ms2sysv_xlogues): Rename to...
22413 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
22414 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
22416 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
22417 change after reload_completed.
22418 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
22420 (ix86_expand_prologue): Likewise.
22421 (ix86_expand_epilogue): Likewise.
22422 (ix86_expand_split_stack_prologue): Likewise.
22423 (ix86_compute_frame_layout): Remove frame parameter ...
22424 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
22425 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
22427 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
22428 (ix86_frame): Move from here ...
22429 * config/i386/i386.h (ix86_frame): ... to here.
22430 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
22431 complete ix86_frame data structure instead. Remove some_ld_name.
22433 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22435 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
22436 symbols that hold a DECL_VALUE_EXPR.
22438 2017-06-01 Martin Jambor <mjambor@suse.cz>
22440 PR tree-optimization/80898
22441 * tree-sra.c (process_subtree_disqualification): Removed.
22442 (disqualify_candidate): Do not acll
22443 process_subtree_disqualification.
22444 (subtree_mark_written_and_enqueue): New function.
22445 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
22446 RHS has been disqualified and re-queue LHS if necessary. Apart
22447 from that, ignore disqualified RHS.
22449 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22451 * config/s390/s390.c (s390_emit_epilogue): Disable early return
22452 address fetch for z10 or later.
22454 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22456 * config/arc/arc.md (tst_movb): Add guard when splitting.
22458 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22460 * config/arc/arc.c (arc_can_eliminate): Test against
22461 arc_frame_pointer_needed.
22463 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22465 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
22466 to prevent store reordering.
22467 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
22468 (type): Add block type.
22469 (stack_tie): Define special instruction to be used in
22472 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22474 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
22475 constraint. It is not valid for the pattern.
22476 (noncommutative_binary_comparison): Likewise.
22478 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22480 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
22483 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22485 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
22486 be used by the reg-alloc.
22488 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22490 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
22491 reg-alloc when having mul64 or mul32x16 instructions.
22492 (mulsidi3): Likewise.
22493 (umulsidi3): Likewise.
22494 (mulsi32x16): New pattern.
22495 (mulsi64): Likewise.
22496 (mulsidi64): Likewise.
22497 (umulsidi64): Likewise.
22498 (MUL32x16_REG): Define.
22499 (mul64_600): Use MUL32x16_REG.
22500 (mac64_600): Likewise.
22501 (umul64_600): Likewise.
22502 (umac64_600): Likewise.
22504 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22506 * config/arc/arc.md (mulsi3_700): Make it commutative.
22508 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
22510 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
22512 (*sign_extendsidi2_insn): Likewise for movstosw.
22514 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22516 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
22517 the type of the input discriminant value. Convert the
22518 discriminant value of signedness vary.
22520 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
22522 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
22523 Add to -Wall section.
22525 2017-06-01 Richard Biener <rguenther@suse.de>
22527 PR middle-end/66313
22528 * fold-const.c (fold_plusminus_mult_expr): If the factored
22529 factor may be zero use a wrapping type for the inner operation.
22530 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
22531 and handle moved defs.
22532 (process_assignment): Properly guard the unary op case. Return a
22533 tri-state indicating that moving the stmt before the call may allow
22534 to continue. Pass through to_move.
22535 (find_tail_calls): Handle moving unrelated defs before
22538 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
22541 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
22542 splitter result in the canonical way.
22544 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
22546 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
22547 also for 32bit target. Update insn attributes.
22548 (zero-extendsidi2 splitter): Allow all registers for operand 1.
22550 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
22552 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
22553 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
22554 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
22555 (_mm_maskz_min_ss): New intrinsics.
22557 2017-05-31 Martin Liska <mliska@suse.cz>
22559 * tree-vect-loop.c (vect_create_epilog_for_reduction):
22560 Change comment style to one we normally use.
22561 (vectorizable_reduction): Likewise.
22562 (vectorizable_induction): Likewise.
22563 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
22564 (vectorizable_call): Likewise.
22565 (vectorizable_simd_clone_call): Likewise.
22566 (vectorizable_conversion): Likewise.
22567 (vectorizable_assignment): Likewise.
22568 (vectorizable_shift): Likewise.
22569 (vectorizable_operation): Likewise.
22570 (vectorizable_store): Likewise.
22571 (vectorizable_load): Likewise.
22572 * tree-vectorizer.h: Likewise.
22574 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
22576 * passes.c (emergency_dump_function): New.
22577 * tree-pass.h (emergency_dump_function): Declare.
22578 * plugin.c (plugins_internal_error_function): Remove.
22579 * plugin.h (plugins_internal_error_function): Remove declaration.
22580 * toplev.c (internal_error_function): New static function. Use it...
22581 (general_init): ...here.
22583 2017-05-31 Graham Markall <graham.markall@embecosm.com>
22585 * config/arc/arc.c (arc_print_operand): Handle constant operands.
22586 (arc_rtx_costs): Add costs for new patterns.
22587 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
22588 * config/arc/predicates.md: Add _1_2_3_operand predicate.
22590 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22592 * tree-ssa-strlen.c (get_next_strinfo): New function.
22593 (get_stridx_plus_constant): Use it.
22594 (zero_length_string): Likewise.
22595 (adjust_related_strinfos): Likewise.
22596 (adjust_last_stmt): Likewise.
22598 2017-05-31 Richard Biener <rguenther@suse.de>
22601 * config/i386/i386.c (ix86_expand_builtin): Remove assert
22602 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
22604 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22606 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
22607 loop_vinfo argument and use of dependence distance vectors.
22608 Check instead whether the two references differ only in their
22609 initial value and assume that they have the same alignment if the
22610 difference is a multiple of the vector alignment.
22611 (vect_analyze_data_refs_alignment): Update call accordingly.
22613 2017-05-31 Martin Liska <mliska@suse.cz>
22616 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
22618 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22620 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
22621 (create_intersect_range_checks): Move from ...
22622 * tree-data-ref.c (create_intersect_range_checks_index)
22623 (create_intersect_range_checks): ... to here.
22624 (create_runtime_alias_checks): New function factored from ...
22625 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
22626 here. Call above function.
22627 * tree-data-ref.h (create_runtime_alias_checks): New function.
22629 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22631 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
22632 segment length for dr_b and compute it in wide_int.
22634 2017-05-31 Richard Biener <rguenther@suse.de>
22636 PR tree-optimization/80906
22637 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
22638 and pass through iv_map.
22639 (copy_bb_and_scalar_dependences): Adjust.
22640 (translate_pending_phi_nodes): Likewise.
22641 (copy_loop_close_phi_args): Handle code-generating IVs instead
22644 2017-05-30 David Malcolm <dmalcolm@redhat.com>
22646 * diagnostic-color.c (color_dict): Add "type-diff".
22647 (parse_gcc_colors): Update comment.
22648 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
22649 -fdiagnostics-show-template-tree and -fno-elide-type.
22650 (GCC_COLORS): Add type-diff to example.
22652 (-fdiagnostics-show-template-tree): New.
22653 (-fno-elide-type): New.
22654 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
22655 the pp_format_decoder callback. Call any m_format_postprocessor's
22657 (pretty_printer::pretty_printer): Initialize
22658 m_format_postprocessor.
22659 (pretty_printer::~pretty_printer): Delete any
22660 m_format_postprocessor.
22661 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
22662 (class format_postprocessor): New class.
22663 (struct pretty_printer::format_decoder): Document the new parameters.
22664 (struct pretty_printer::m_format_postprocessor): New field.
22665 * tree-diagnostic.c (default_tree_printer): Update for new
22666 bool and const char ** params.
22667 * tree-diagnostic.h (default_tree_printer): Likewise.
22669 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
22671 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
22672 (lwa_operand): Delete rs6000_gen_cell_microcode test.
22673 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22674 rs6000_gen_cell_microcode code.
22675 (rs6000_final_prescan_insn): Delete.
22676 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
22677 "warn-cell-microcode" entries.
22678 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
22679 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
22680 throughout. Change cc_reg_not_micro_cr0_operand to
22681 cc_reg_not_cr0_operand throughout.
22682 (*extendhi<mode>2_noload): Delete.
22683 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
22684 (mwarn-cell-microcode): Delete.
22685 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
22686 -mgen-cell-microcode and -mwarn-cell-microcode.
22688 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
22691 * config/i386/constraints.md (Yd): New constraint.
22693 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
22694 and (?Yd, r) alternatives. Update insn attributes.
22695 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
22696 and (?*Yd, r) alternatives. Update insn attributes.
22697 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
22699 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
22701 * gimplify.c (gimplify_modify_expr): Don't create a
22702 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
22705 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
22707 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
22709 2017-05-30 Richard Biener <rguenther@suse.de>
22711 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
22712 and reduc_def fields.
22713 (STMT_VINFO_REDUC_TYPE): New define.
22714 (STMT_VINFO_REDUC_DEF): Likewise.
22715 (vect_force_simple_reduction): Adjust prototype.
22716 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
22717 (vect_is_simple_reduction): Remove check_reduction argument.
22718 (vect_force_simple_reduction): Adjust and set
22719 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22720 (vectorizable_reduction): Do not re-do reduction analysis
22721 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
22722 * tree-parloops.c (gather_scalar_reductions): Adjust.
22724 2017-05-30 Richard Biener <rguenther@suse.de>
22726 PR middle-end/80901
22727 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
22730 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
22732 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
22733 Introduce unknown_misalignment parameter and remove vf.
22734 (vect_peeling_hash_get_lowest_cost):
22735 Pass unknown_misalignment parameter.
22736 (vect_enhance_data_refs_alignment):
22737 Fix unsupportable data ref treatment.
22739 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22741 * tree-vect-data-refs.c (vect_get_data_access_cost):
22742 Workaround for SLP handling.
22743 (vect_enhance_data_refs_alignment):
22744 Compute costs for doing no peeling at all, compare to the best
22745 peeling costs so far and avoid peeling if cheaper.
22747 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22749 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
22750 Return peeling info and set costs to zero for unlimited cost
22752 (vect_enhance_data_refs_alignment): Also inspect all datarefs
22753 with unknown misalignment. Compute and costs for unknown
22754 misalignment, compare them to the costs for known misalignment
22755 and choose the cheapest for peeling.
22757 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22759 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
22760 (vect_get_peeling_costs_all_drs): Create function.
22761 (vect_peeling_hash_get_lowest_cost):
22762 Use vect_get_peeling_costs_all_drs.
22763 (vect_peeling_supportable): Create function.
22764 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
22766 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
22768 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
22769 DR_HAS_NEGATIVE_STEP.
22770 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
22771 (vect_enhance_data_refs_alignment): Use.
22772 (vect_duplicate_ssa_name_ptr_info): Use.
22773 * tree-vectorizer.h (dr_misalignment): Use.
22774 (known_alignment_for_access_p): Use.
22776 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
22779 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
22781 (has_section_name): New function.
22783 2017-05-30 Martin Liska <mliska@suse.cz>
22786 * auto-profile.c (get_function_decl_from_block): Fix
22789 2017-05-30 Richard Biener <rguenther@suse.de>
22791 PR middle-end/80876
22792 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
22794 2017-05-30 Martin Liska <mliska@suse.cz>
22796 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
22797 * dumpfile.h (struct dump_file_info): Remove ctors.
22799 2017-05-30 Martin Liska <mliska@suse.cz>
22801 * predict.def: Fix GNU coding style.
22803 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22805 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22806 Mark 'to' argument with ATTRIBUTE_UNUSED.
22808 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22810 * config/xtensa/xtensa.c (xtensa_emit_call): Use
22811 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
22812 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
22815 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
22817 * doc/install.texi (Options specification): Restore entry of
22818 --enable-sjlj-exceptions.
22820 2017-05-27 Michael Eager <eager@eagercon.com>
22823 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
22825 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
22827 * config/microblaze/microblaze.h
22828 (FIXED_REGISTERS): Update in macro.
22829 (CALL_USED_REGISTERS): Update in macro.
22831 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22833 * doc/install.texi: Add links to macOS binary distributions.
22835 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22839 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22841 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22843 2017-05-26 Martin Liska <mliska@suse.cz>
22845 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22847 2017-05-26 Martin Liska <mliska@suse.cz>
22849 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22850 always leading ';; '.
22851 (dump_bb_info): Likewise.
22852 (brief_dump_cfg): Likewise.
22853 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22854 * dumpfile.c: Remove usage of TDF_VERBOSE.
22855 * dumpfile.h (enum dump_kind): Likewise.
22856 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22857 * print-tree.c (debug_verbose): Remove.
22858 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22859 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22860 * tree-diagnostic.c (default_tree_printer): Replace
22861 TDF_DIAGNOSTIC with TDF_SLIM.
22863 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22865 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22866 in parameter loop, rather than loop_vinfo.
22867 (create_intersect_range_checks): Ditto.
22868 (vect_create_cond_for_alias_checks): Update call to above functions.
22870 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22872 PR tree-optimization/80815
22873 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22874 for merging runtime alias checks. Handle negative DR_STEPs.
22876 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22878 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22880 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22881 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22882 out code pruning runtime alias checks.
22883 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22884 factored out from above.
22885 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22887 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22889 (prune_runtime_alias_test_list): New decalaration.
22891 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22893 * tree-vect-data-refs.c (compare_tree): Rename and move ...
22894 * tree-data-ref.c (data_ref_compare_tree): ... to here.
22895 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
22896 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
22897 (operator==, comp_dr_with_seg_len_pair): Ditto.
22898 (vect_prune_runtime_alias_test_list): Ditto.
22900 2017-05-26 Martin Liska <mliska@suse.cz>
22903 * params.def: Bound partial-inlining-entry-probability param.
22905 2017-05-26 Marek Polacek <polacek@redhat.com>
22908 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22911 2017-05-26 Richard Biener <rguenther@suse.de>
22913 PR tree-optimization/80842
22914 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22917 2017-05-26 Richard Biener <rguenther@suse.de>
22919 PR tree-optimization/80844
22920 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22922 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22924 * doc/md.texi (Machine Constraints): Update x86 family
22925 machine constraints section to match 'config/i386/constraints.md'.
22927 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22929 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22931 2017-05-25 Nathan Sidwell <nathan@acm.org>
22933 * doc/invoke.texi (--enable-languages): Update documentation.
22935 2017-05-25 Martin Liska <mliska@suse.cz>
22937 * dumpfile.c: Add TDF_FOLDING.
22938 * dumpfile.h (enum dump_kind): Likewise.
22939 * genmatch.c (dt_simplify::gen_1): Use it.
22941 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22943 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
22945 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22947 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22948 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
22950 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22952 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
22953 * match.pd (X == C): Rewrite it here.
22954 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
22955 with_certain_nonzero_bits2): New predicates.
22956 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
22958 2017-05-24 Nathan Sidwell <nathan@acm.org>
22960 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
22963 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
22966 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22968 * config/powerpcspe: New port. Files are copied from the rs6000
22969 port, with "rs6000" in filenames replaced by "powerpcspe".
22971 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
22973 PR rtl-optimization/80754
22974 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
22976 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
22978 * config/sparc/sparc.md (length): Return the correct value for -mflat
22979 sibcalls to match output_sibcall.
22981 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22985 * config/rs6000/rs6000.c (struct machine_function): Add new field
22987 (rs6000_get_separate_components): Init that field, use it.
22988 (rs6000_components_for_bb): Use the field.
22990 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
22992 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
22994 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
22996 PR middle-end/80823
22997 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
22999 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23002 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23003 address against address_operand predicate.
23004 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23006 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23008 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23009 parameters passed indirectly.
23011 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23013 * config/i386/i386.md (*movdi_internal): Remove SSE4
23014 alternative 18 (?r, *v). Update insn attributes.
23015 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23016 Update insn attributes.
23017 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23018 Update insn attributes.
23019 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23020 alternative 1 (r, v). Remove isa attribute.
23021 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23022 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23023 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23025 2017-05-23 Tom de Vries <tom@codesourcery.com>
23027 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23030 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23032 * cgraphunit.c (symbol_table::process_new_functions): Update.
23033 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23034 (inline_generate_summary): Rename to ...
23035 (ipa_fn_summary_generate): ... this one.
23036 (inline_read_summary): Rename to ...
23037 (ipa_fn_summary_read): ... this one.
23038 (inline_write_summary): Rename to ...
23039 (ipa_fn_summary_write): ... this one.
23040 (inline_free_summary): Rename to ...
23041 (ipa_free_fn_summary): ... this one.
23042 (pass_data_local_fn_summary, pass_local_fn_summary,
23043 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23044 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23045 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23046 make_pass_ipa_fn_summary): New.
23047 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23048 inline_write_summary, inline_free_summary): Remove.
23049 (ipa_free_fn_summary) : New.
23050 * ipa-inline.c (ipa_inline): Update.
23051 (pass_ipa_inline): Do not generate summaries.
23052 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23054 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23055 and add pass_ipa_fn_summary.
23056 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23058 (make_pass_inline_parameters): Remove.
23060 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23062 * omp-low.c (struct omp_context): Remove "default_kind" member.
23065 * omp-offload.c (execute_oacc_device_lower): Remove the
23066 parallelism dimensions function attributes for unparallelized
23067 OpenACC kernels constructs.
23069 2017-05-23 Martin Liska <mliska@suse.cz>
23071 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23073 (cgraph_edge::make_speculative): Likewise.
23074 (cgraph_edge::resolve_speculation): Likewise.
23075 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23076 (cgraph_node::dump): Likewise.
23077 * cgraph.h: Likewise.
23078 * cgraphunit.c (analyze_functions): Likewise.
23079 (symbol_table::compile): Likewise.
23080 * ipa-cp.c (print_all_lattices): Likewise.
23081 (determine_versionability): Likewise.
23082 (initialize_node_lattices): Likewise.
23083 (ipcp_verify_propagated_values): Likewise.
23084 (estimate_local_effects): Likewise.
23085 (update_profiling_info): Likewise.
23086 (create_specialized_node): Likewise.
23087 (perhaps_add_new_callers): Likewise.
23088 (decide_about_value): Likewise.
23089 (decide_whether_version_node): Likewise.
23090 (identify_dead_nodes): Likewise.
23091 (ipcp_store_bits_results): Likewise.
23092 * ipa-devirt.c (dump_targets): Likewise.
23093 (ipa_devirt): Likewise.
23094 * ipa-icf.c (sem_item::dump): Likewise.
23095 (sem_function::equals): Likewise.
23096 (sem_variable::equals): Likewise.
23097 (sem_item_optimizer::read_section): Likewise.
23098 (sem_item_optimizer::execute): Likewise.
23099 (congruence_class::dump): Likewise.
23100 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23101 (dump_inline_summary): Likewise.
23102 (estimate_node_size_and_time): Likewise.
23103 (inline_analyze_function): Likewise.
23104 * ipa-inline-transform.c (inline_call): Likewise.
23105 * ipa-inline.c (report_inline_failed_reason): Likewise.
23106 (want_early_inline_function_p): Likewise.
23107 (edge_badness): Likewise.
23108 (update_edge_key): Likewise.
23109 (inline_small_functions): Likewise.
23110 * ipa-profile.c (ipa_profile): Likewise.
23111 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23112 (ipa_make_edge_direct_to_target): Likewise.
23113 (remove_described_reference): Likewise.
23114 (ipa_impossible_devirt_target): Likewise.
23115 (propagate_controlled_uses): Likewise.
23116 (ipa_print_node_params): Likewise.
23117 (ipcp_transform_function): Likewise.
23118 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23119 (propagate_pure_const): Likewise.
23120 * ipa-reference.c (generate_summary): Likewise.
23121 (read_write_all_from_decl): Likewise.
23122 (propagate): Likewise.
23123 (ipa_reference_read_optimization_summary): Likewise.
23124 * ipa-utils.c (ipa_merge_profiles): Likewise.
23125 * ipa.c (walk_polymorphic_call_targets): Likewise.
23126 (symbol_table::remove_unreachable_nodes): Likewise.
23127 (ipa_single_use): Likewise.
23128 * passes.c (execute_todo): Likewise.
23129 * predict.c (drop_profile): Likewise.
23130 * symtab.c (symtab_node::get_dump_name): New function.
23131 (symtab_node::dump_name): Likewise.
23132 (symtab_node::dump_asm_name): Likewise.
23133 (symtab_node::dump_references): Likewise.
23134 (symtab_node::dump_referring): Likewise.
23135 (symtab_node::dump_base): Likewise.
23136 (symtab_node::debug_symtab): Likewise.
23137 * tree-sra.c (convert_callers_for_node): Likewise.
23138 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23139 * value-prof.c (init_node_map): Likewise.
23141 2017-05-23 Martin Liska <mliska@suse.cz>
23143 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23144 and symtab_node::debug_symtab to symbol_table::debug.
23145 * cgraphunit.c (analyze_functions): Use the renamed function.
23146 (symbol_table::compile): Likewise.
23147 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23148 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23149 * passes.c (execute_todo): Likewise.
23150 * symtab.c (symbol_table::dump): New function.
23151 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23153 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23155 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23156 that nonconst implies exec.
23158 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23160 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23161 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23162 (inline_edge_summary_vec): Turn into ...
23163 (ipa_call_summaries): ... this one.
23164 (redirect_to_unreachable, edge_set_predicate,
23165 evaluate_properties_for_edge, inline_summary_alloc,
23166 reset_ipa_call_summary, reset_inline_summary,
23167 inline_summary_t::duplicate): Update.
23168 (inline_edge_duplication_hook): Turn to ...
23169 (ipa_call_summary_t::duplicate): ... this one.
23170 (inline_edge_removal_hook): Turn to ...
23171 (ipa_call_summary_t::remove): ... this one.
23172 (dump_inline_edge_summary): Turn to ...
23173 (dump_ipa_call_summary): ... this one.
23174 (estimate_function_body_sizes): Update.
23175 (inline_update_callee_summaries): Update.
23176 (remap_edge_change_prob): Update.
23177 (remap_edge_summaries): Update.
23178 (inline_merge_summary): Update.
23179 (do_estimate_edge_time): Update.
23180 (inline_generate_summary): Update.
23181 (inline_read_section): Update.
23182 (inline_read_summary): Update.
23183 (inline_free_summary): Update.
23184 * ipa-inline.c (can_inline_edge_p): Update.
23185 (compute_inlined_call_time): Update.
23186 (want_inline_small_function_p): Update.
23187 (edge_badness): Update.
23188 (early_inliner): Update.
23189 * ipa-inline.h (inline_edge_summary): Turn to ...
23190 (ipa_call_summary): ... this one.
23191 (ipa_call_summary_t): New class.
23192 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23193 (ipa_call_summaries): New.
23194 (inline_edge_summary): Remove.
23195 (estimate_edge_growth): Update.
23196 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23197 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23198 * ipa-split.c (execute_split_functions): Update.
23199 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23201 2017-05-23 Tom de Vries <tom@codesourcery.com>
23203 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23204 attributes): Document rdrand effective target.
23206 2017-05-23 Tom de Vries <tom@codesourcery.com>
23208 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23209 attributes): Sort alphabetically.
23211 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23213 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23215 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23218 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23219 V2DF/V2DI splat into two separate patterns, one that handles
23220 registers, and the other that only handles memory. Drop support
23221 for splatting from a GPR on ISA 2.07 and then splitting the
23222 splat into direct move and splat.
23223 (vsx_splat_<mode>_reg): Likewise.
23224 (vsx_splat_<mode>_mem): Likewise.
23226 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23228 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23230 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23232 PR middle-end/80809
23233 * omp-low.c (finish_taskreg_remap): New function.
23234 (finish_taskreg_scan): If unit size of ctx->record_type
23235 is non-constant, unshare the size expression and replace
23236 decls in it with possible outer var refs.
23238 PR middle-end/80809
23239 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23240 GOVD_SHARED rather than GOVD_PRIVATE with it.
23241 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23242 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23244 PR middle-end/80853
23245 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23246 as last argument to build_outer_var_ref for pointer bases of array
23247 section reductions.
23249 2017-05-19 Martin Sebor <msebor@redhat.com>
23251 * print-tree.c (print_node): Print DECL_READ_P flag.
23253 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23255 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23256 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23257 * cgraph.c: Likewise.
23258 * cgraphunit.c: Likewise.
23259 * gengtype.c: Likewise.
23260 * ipa-cp.c: Likewise.
23261 * ipa-devirt.c: Likewise.
23262 * ipa-icf.c: Likewise.
23263 * ipa-predicate.c: Likewise.
23264 * ipa-profile.c: Likewise.
23265 * ipa-prop.c: Likewise.
23266 * ipa-split.c: Likewise.
23268 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23269 edge_predicate_pool, dump_inline_hints,
23270 inline_summary::account_size_time, redirect_to_unreachable,
23271 edge_set_predicate, set_hint_predicate,
23272 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23273 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23274 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23275 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23276 ipa_call_summary_t::remove, initialize_growth_caches,
23277 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23278 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23279 mark_modified, unmodified_parm_1, unmodified_parm,
23280 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23281 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23282 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23283 will_be_nonconstant_predicate, record_modified_bb_info,
23284 get_minimal_bb, record_modified, param_change_prob,
23285 phi_result_unknown_predicate, predicate_for_phi_result,
23286 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23287 estimate_function_body_sizes, compute_inline_parameters,
23288 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23289 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23290 inline_update_callee_summaries, remap_edge_change_prob,
23291 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23292 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23293 inline_analyze_function, inline_summary_t::insert,
23294 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23295 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23296 inline_free_summary): Move to ipa-fnsummary.h
23297 (predicate_t): Remove.
23298 * ipa-fnsummary.c: New file.
23299 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23300 (enum inline_hints_vals, inline_hints, agg_position_info,
23301 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23302 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23303 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23304 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23305 inline_read_summary, inline_write_summary, inline_free_summary,
23306 inline_analyze_function, initialize_inline_failed,
23307 inline_merge_summary, inline_update_overall_summary,
23308 compute_inline_parameters): Move to ipa-fnsummary.h
23309 * ipa-fnsummary.h: New file.
23310 * ipa-inline-transform.h: Include ipa-inline.h.
23311 * ipa-inline.c: LIkewise.
23313 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23315 * ipa-inline.c (edge_badness): Use inlined_time instead of
23316 inline_summaries->get.
23318 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23320 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23322 2017-05-22 Nathan Sidwell <nathan@acm.org>
23324 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23325 (fdump-lang): Document 'raw' option.
23326 * dumpfile.h (TDI_tu): Delete.
23327 * dumpfile.c (dump_files): Remove translation-unit.
23328 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23330 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23332 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23333 command option from $(AWK) call.
23334 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23335 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23336 [FORMAT]: Remove handling of variable.
23337 * config/avr/t-multilib: Regenerate.
23339 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23341 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23343 (dump_inline_summary): Do not print self_time.
23344 (estimate_function_body_sizes): Do not set self_time.
23345 (compute_inline_parameters): Likewise.
23346 (inline_read_section, inline_write_summary): Do not stream self_time.
23347 * ipa-inline.h (inline_summary): Drop self_time.
23349 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23351 * ipa-inline-analysis.c (account_size_time): Rename to ...
23352 (inline_summary::account_size_time): ... this one.
23353 (reset_ipa_call_summary): Turn to ...
23354 (ipa_call_summary::reset): ... this one.
23355 (reset_inline_summary): Turn to ...
23356 (inline_summary::reset): ... this one.
23357 (inline_summary_t::remove): Update.
23358 (inline_summary_t::duplicate): Update.
23359 (ipa_call_summary_t::remove): Update.
23360 (dump_inline_summary): Update.
23361 (estimate_function_body_sizes): Update.
23362 (compute_inline_parameters): Update.
23363 (estimate_node_size_and_time): Update.
23364 (inline_merge_summary): Update.
23365 (inline_update_overall_summary): Update.
23366 (inline_read_section): Update.
23367 (inline_write_summary): Update.
23368 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23369 add account_size_time and reset member functions.
23370 (ipa_call_summary): Add reset function.
23371 * ipa-predicate.h (predicate::operator &): Constify.
23373 2017-05-22 Richard Biener <rguenther@suse.de>
23375 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23377 2017-05-19 Jason Merrill <jason@redhat.com>
23379 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23381 2017-05-19 Marek Polacek <polacek@redhat.com>
23384 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
23385 TYPE_OVERFLOW_WRAPS checks.
23387 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
23389 * tree-core.h (enum omp_clause_default_kind): Add
23390 "OMP_CLAUSE_DEFAULT_PRESENT".
23391 * tree-pretty-print.c (dump_omp_clause): Handle it.
23392 * gimplify.c (enum gimplify_omp_var_data): Add
23393 "GOVD_MAP_FORCE_PRESENT".
23394 (gimplify_adjust_omp_clauses_1): Map it to
23395 "GOMP_MAP_FORCE_PRESENT".
23396 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
23398 * gimplify.c (oacc_default_clause): Clarify.
23400 2017-05-19 Nathan Sidwell <nathan@acm.org>
23402 LANG_HOOK_REGISTER_DUMPS
23403 * toplev.c (general_init): Call register dump lang hook.
23404 * doc/invoke.texi: Document -fdump-lang option family.
23405 * dumpfile.c (dump_files): Remove class dump here.
23406 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
23407 * dumpfile.h (tree_dump_index): Remove TDI_class.
23408 * langhooks-def.h (lhd_register_dumps): Declare.
23409 (LANG_HOOKS_REGISTER_DUMPS): Define.
23410 (LANG_HOOKS_INITIALIZER): Add it.
23411 * langhooks.c (lhd_register_dumps): Define.
23412 * langhooks.h (struct lang_hooks): Add register_dumps.
23414 2017-05-19 Nathan Sidwell <nathan@acm.org>
23416 * context.h (context::set_passes): New.
23417 * context.c (context::context): Do not create pass manager.
23418 * toplev.c (general_init): Create pass manager here.
23420 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
23422 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
23423 use this splitter if two add or or instructions would also work for
23424 the constant we want to generate.
23426 2017-05-19 Richard Biener <rguenther@suse.de>
23429 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
23430 predicate evaluation.
23432 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23434 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
23436 * ipa-inline.c (want_inline_small_function_p): Do not cast to
23439 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23441 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23442 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23443 (inline_edge_summary_vec): Turn into ...
23444 (ipa_call_summaries): ... this one.
23445 (redirect_to_unreachable, edge_set_predicate,
23446 evaluate_properties_for_edge, inline_summary_alloc,
23447 reset_ipa_call_summary, reset_inline_summary,
23448 inline_summary_t::duplicate): Update.
23449 (inline_edge_duplication_hook): Turn to ...
23450 (ipa_call_summary_t::duplicate): ... this one.
23451 (inline_edge_removal_hook): Turn to ...
23452 (ipa_call_summary_t::remove): ... this one.
23453 (dump_inline_edge_summary): Turn to ...
23454 (dump_ipa_call_summary): ... this one.
23455 (estimate_function_body_sizes): Update.
23456 (inline_update_callee_summaries): Update.
23457 (remap_edge_change_prob): Update.
23458 (remap_edge_summaries): Update.
23459 (inline_merge_summary): Update.
23460 (do_estimate_edge_time): Update.
23461 (inline_generate_summary): Update.
23462 (inline_read_section): Update.
23463 (inline_read_summary): Update.
23464 (inline_free_summary): Update.
23465 * ipa-inline.c (can_inline_edge_p): Update.
23466 (compute_inlined_call_time): Update.
23467 (want_inline_small_function_p): Update.
23468 (edge_badness): Update.
23469 (early_inliner): Update.
23470 * ipa-inline.h (inline_edge_summary): Turn to ...
23471 (ipa_call_summary): ... this one.
23472 (ipa_call_summary_t): New class.
23473 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23474 (ipa_call_summaries): New.
23475 (inline_edge_summary): Remove.
23476 (estimate_edge_growth): Update.
23477 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23478 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23479 * ipa-split.c (execute_split_functions): Update.
23480 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23482 2017-05-19 Richard Biener <rguenther@suse.de>
23484 PR middle-end/80764
23485 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
23487 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
23489 * config/rs6000/rs6000.c (struct machine_function): Add field
23490 fpr_is_wrapped_separately.
23491 (rs6000_get_separate_components): Use 64 components. Handle the
23492 new FPR components.
23493 (rs6000_components_for_bb): Handle the FPR components.
23494 (rs6000_emit_prologue_components): Handle the FPR components.
23495 (rs6000_emit_epilogue_components): Handle the FPR components.
23496 (rs6000_set_handled_components): Handle the FPR components.
23497 (rs6000_emit_prologue): Don't output prologue code for those FPRs
23498 that are already separately shrink-wrapped.
23499 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
23500 that are already separately shrink-wrapped.
23502 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23505 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
23508 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
23509 (define_peephole2 for Altivec d-form load): Add peepholes to catch
23510 cases where the register allocator uses a move and an offsettable
23511 memory operation to/from a FPR register on ISA 2.06/2.07.
23512 (define_peephole2 for Altivec d-form store): Likewise.
23514 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
23517 * config/i386/mmx.md (*mov<mode>_internal): Enable
23518 alternatives 11, 12, 13 and 14 also for 32bit targets.
23519 Remove alternatives 15, 16, 17 and 18.
23520 * config/i386/sse.md (vec_concatv2di): Change
23521 alternative (!x, *y) to (x, ?!*Yn).
23523 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
23525 * dumpfile.h (enum dump_kind): Remove stray comma.
23527 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23529 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
23530 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
23531 predicate::num_conditions
23532 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
23533 (CHANGED): turn into predicate::changed.
23534 (agg_position_info): Move to ipa-predicate.h
23535 (add_condition, predicate::add_clause, predicate::operator &=,
23536 predicate::or_with, predicate::evaluate, predicate::probability,
23537 dump_condition, dump_clause, predicate::dump,
23538 predicate::remap_after_duplication, predicate::remap_after_inlining,
23539 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
23540 (evaluate_conditions_for_known_args): Update.
23541 (set_cond_stmt_execution_predicate): Update.
23542 * ipa-inline.h: Include ipa-predicate.h
23543 (condition, inline_param_summary, conditions, agg_position_info,
23544 predicate): Move to ipa-predicate.h
23545 * ipa-predicate.c: New file.
23546 * ipa-predicate.h: New file.
23548 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
23550 * final.c (leaf_function_p): Check we are not in a sequence.
23552 2017-05-18 Martin Liska <mliska@suse.cz>
23554 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
23555 * dumpfile.c (dump_register): Use new enum dump_kind.
23556 (get_dump_file_name): Likewise.
23557 (dump_enable_all): Likewise.
23558 (dump_switch_p_1): Likewise.
23559 (enable_rtl_dump_file): Remove usage of TDF_RTL.
23560 * dumpfile.h (enum dump_kind): New enum type.
23561 (struct dump_file_info): Create constructor and
23562 format fields and comments.
23563 * passes.c (pass_manager::register_one_dump_file):
23565 * statistics.c (statistics_early_init): Likewise.
23566 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
23567 TDF_TREE with TDF_SLIM.
23568 (gather_memory_references_ref): Likewise.
23570 2017-05-18 Martin Liska <mliska@suse.cz>
23572 * vec.h (struct vnull): Use it.
23574 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23576 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
23577 (true_predicate, false_predicate, true_predicate_p,
23578 false_predicate_p): Remove.
23579 (single_cond_predicate, not_inlined_predicate): Turn to member function
23581 (add_condition): Update.
23582 (add_clause): Turn to...
23583 (predicate::add_clause): ... this one; update; allow passing NULL
23585 (and_predicates): Turn to ...
23586 (predicate::operator &=): ... this one.
23587 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
23588 (or_predicates): Turn to ...
23589 (predicate::or_with): ... this one.
23590 (evaluate_predicate): Turn to ...
23591 (predicate::evaluate): ... this one.
23592 (predicate_probability): Turn to ...
23593 (predicate::probability): ... this one.
23594 (dump_condition): Update.
23595 (dump_predicate): Turn to ...
23596 (predicate::dump): ... this one.
23597 (account_size_time): Update.
23598 (edge_set_predicate): Update.
23599 (set_hint_predicate): UPdate.
23600 (evaluate_conditions_for_known_args): Update.
23601 (evaluate_properties_for_edge): Update.
23602 (remap_predicate_after_duplication): Turn to...
23603 (predicate::remap_after_duplication): ... this one.
23604 (remap_hint_predicate_after_duplication): Update.
23605 (inline_summary_t::duplicate): UPdate.
23606 (dump_inline_edge_summary): Update.
23607 (dump_inline_summary): Update.
23608 (set_cond_stmt_execution_predicate): Update.
23609 (set_switch_stmt_execution_predicate): Update.
23610 (compute_bb_predicates): Update.
23611 (will_be_nonconstant_expr_predicate): Update.
23612 (will_be_nonconstant_predicate): Update.
23613 (phi_result_unknown_predicate): Update.
23614 (predicate_for_phi_result): Update.
23615 (array_index_predicate): Update.
23616 (estimate_function_body_sizes): Update.
23617 (estimate_node_size_and_time): Update.
23618 (estimate_ipcp_clone_size_and_time): Update.
23619 (remap_predicate): Rename to ...
23620 (predicate::remap_after_inlining): ... this one.
23621 (remap_hint_predicate): Update.
23622 (inline_merge_summary): Update.
23623 (inline_update_overall_summary): Update.
23624 (estimate_size_after_inlining): Update.
23625 (read_predicate): Rename to ...
23626 (predicate::stream_in): ... this one.
23627 (read_inline_edge_summary): Update.
23628 (write_predicate): Rename to ...
23629 (predicate::stream_out): ... this one.
23630 (write_inline_edge_summary): Update.
23631 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
23632 (clause_t): Turn to uint32_t
23633 (predicate): Turn to class; implement constructor and operators
23635 (size_time_entry): Update.
23636 (inline_summary): Update.
23637 (inline_edge_summary): Update.
23639 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
23641 * fold-const.c (fold_binary_loc): Move transformation...
23642 * match.pd (C - X CMP X): ... here.
23644 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
23646 * config/sparc/sparc.c (sparc_option_override): Set function
23647 alignment for -mcpu=niagara7 to 64 to match the I$ line.
23648 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
23650 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
23652 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
23654 2017-05-18 Marek Polacek <polacek@redhat.com>
23657 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
23658 (pass_ubsan::execute): Call gimple_assign_single_p instead of
23659 gimple_assign_load_p.
23661 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
23663 PR middle-end/80692
23664 * real.c (do_compare): Give decimal_do_compare preference over
23665 comparing just the signs.
23667 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
23669 * doc/md.texi (Canonicalization of Instructions): Describe the
23670 canonical form of instructions that inherently set a condition
23673 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
23675 PR middle-end/80775
23676 * tree-cfg.c: Move deletion of unreachable case statements to after
23677 the merging of consecutive case labels.
23679 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
23681 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
23682 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
23683 restoring of callee-saved registers.
23685 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
23687 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
23688 * config/visium/visium.c (single_set_and_flags): Likewise.
23689 * config/visium/visium.md (Substitutions): Likewise.
23691 2017-05-17 Martin Liska <mliska@suse.cz>
23693 * cfg.c: Introduce dump_flags_t type and
23694 use it instead of int type.
23696 * cfghooks.c: Likewise.
23697 * cfghooks.h (struct cfg_hooks): Likewise.
23698 * cfgrtl.c: Likewise.
23699 * cfgrtl.h: Likewise.
23700 * cgraph.c (cgraph_node::get_body): Likewise.
23701 * coretypes.h: Likewise.
23702 * domwalk.c: Likewise.
23703 * domwalk.h: Likewise.
23704 * dumpfile.c (struct dump_option_value_info): Likewise.
23705 (dump_enable_all): Likewise.
23706 (dump_switch_p_1): Likewise.
23707 (opt_info_switch_p): Likewise.
23708 * dumpfile.h (enum tree_dump_index): Likewise.
23709 (struct dump_file_info): Likewise.
23710 * genemit.c: Likewise.
23711 * generic-match-head.c: Likewise.
23712 * gengtype.c (open_base_files): Likewise.
23713 * gimple-pretty-print.c: Likewise.
23714 * gimple-pretty-print.h: Likewise.
23715 * graph.c (print_graph_cfg): Likewise.
23716 * graphite-scop-detection.c (dot_all_sese): Likewise.
23717 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
23718 * loop-unroll.c (report_unroll): Likewise.
23719 * passes.c (pass_manager::register_one_dump_file): Likewise.
23720 * print-tree.c: Likewise.
23721 * statistics.c: Likewise.
23722 * tree-cfg.c: Likewise.
23723 * tree-cfg.h: Likewise.
23724 * tree-dfa.c: Likewise.
23725 * tree-dfa.h: Likewise.
23726 * tree-dump.c (dump_function): Likewise.
23727 * tree-dump.h (struct dump_info): Likewise.
23728 * tree-pretty-print.c: Likewise.
23729 * tree-pretty-print.h: Likewise.
23730 * tree-ssa-live.c: Likewise.
23731 * tree-ssa-live.h: Likewise.
23732 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
23733 * tree-vect-loop.c: Likewise.
23734 * tree-vect-slp.c: Likewise.
23736 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
23737 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23739 PR tree-optimization/80457
23740 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
23741 of all arguments to a statement as scalar_to_vec operations.
23742 (vectorizable_call): Adjust call to vect_model_simple_cost for
23744 (vectorizable_conversion): Likewise.
23745 (vectorizable_assignment): Likewise.
23746 (vectorizable_shift): Likewise.
23747 (vectorizable_operation): Likewise.
23748 (vectorizable_comparison): Likewise.
23749 (vect_is_simple_cond): Record the def types for operands.
23750 (vectorizable_condition): Likewise, call vect_model_simple_cost.
23751 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
23752 for statement argument count.
23754 2017-05-16 Carl Love <cel@us.ibm.com>
23756 * config/rs6000/rs6000-c: Add support for built-in functions
23757 vector unsigned long long vec_bperm (vector unsigned long long,
23758 vector unsigned char)
23759 vector signed long long vec_mule (vector signed int,
23761 vector unsigned long long vec_mule (vector unsigned int,
23762 vector unsigned int)
23763 vector signed long long vec_mulo (vector signed int,
23765 vector unsigned long long vec_mulo (vector unsigned int,
23766 vector unsigned int)
23767 vector signed char vec_sldw (vector signed char,
23768 vector signed char,
23770 vector unsigned char vec_sldw (vector unsigned char,
23771 vector unsigned char,
23773 vector signed short vec_sldw (vector signed short,
23774 vector signed short,
23776 vector unsigned short vec_sldw (vector unsigned short,
23777 vector unsigned short,
23779 vector signed int vec_sldw (vector signed int,
23782 vector unsigned int vec_sldw (vector unsigned int,
23783 vector unsigned int,
23785 vector signed long long vec_sldw (vector signed long long,
23786 vector signed long long,
23788 vector unsigned long long vec_sldw (vector unsigned long long,
23789 vector unsigned long long,
23791 * config/rs6000/rs6000-c: Add support for built-in functions
23792 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
23793 * config/rs6000/altivec.h: Add defintion for vec_sldw.
23794 * doc/extend.texi: Update the built-in documentation for the
23795 new built-in functions.
23797 2017-05-16 Marek Polacek <polacek@redhat.com>
23801 * tree.c (save_expr): Don't fold the expression.
23803 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
23805 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
23806 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
23807 and (?*y,m). Update insn attributes.
23809 2017-05-16 Martin Liska <mliska@suse.cz>
23811 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
23812 flags argument of print_gimple_stmt, print_gimple_expr,
23813 print_generic_stmt and print_generic_expr.
23814 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
23815 * coretypes.h: Likewise.
23816 * except.c (dump_eh_tree): Likewise.
23817 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
23818 * gimple-pretty-print.h: Likewise.
23819 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
23820 (backprop::push_to_worklist): Likewise.
23821 (backprop::pop_from_worklist): Likewise.
23822 (backprop::process_use): Likewise.
23823 (backprop::intersect_uses): Likewise.
23824 (note_replacement): Likewise.
23825 * gimple-ssa-store-merging.c
23826 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
23827 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
23828 (pass_store_merging::execute): Likewise.
23829 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
23830 (ssa_base_cand_dump_callback): Likewise.
23831 (dump_incr_vec): Likewise.
23832 (replace_refs): Likewise.
23833 (replace_mult_candidate): Likewise.
23834 (create_add_on_incoming_edge): Likewise.
23835 (create_phi_basis): Likewise.
23836 (insert_initializers): Likewise.
23837 (all_phi_incrs_profitable): Likewise.
23838 (introduce_cast_before_cand): Likewise.
23839 (replace_one_candidate): Likewise.
23840 * gimplify.c (gimplify_expr): Likewise.
23841 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23842 (set_rename): Likewise.
23843 (rename_uses): Likewise.
23844 (copy_loop_phi_nodes): Likewise.
23845 (add_close_phis_to_merge_points): Likewise.
23846 (copy_loop_close_phi_args): Likewise.
23847 (copy_cond_phi_args): Likewise.
23848 (graphite_copy_stmts_from_block): Likewise.
23849 (translate_pending_phi_nodes): Likewise.
23850 * graphite-poly.c (print_pdr): Likewise.
23851 (dump_gbb_cases): Likewise.
23852 (dump_gbb_conditions): Likewise.
23853 (print_scop_params): Likewise.
23854 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23855 (build_cross_bb_scalars_use): Likewise.
23856 (gather_bbs::before_dom_children): Likewise.
23857 * hsa-dump.c (dump_hsa_immed): Likewise.
23858 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23859 (get_replacement_map): Likewise.
23860 * ipa-inline-analysis.c (dump_condition): Likewise.
23861 (estimate_function_body_sizes): Likewise.
23862 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23863 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23864 * ipa-prop.c (ipa_dump_param): Likewise.
23865 (ipa_print_node_jump_functions_for_edge): Likewise.
23866 (ipa_modify_call_arguments): Likewise.
23867 (ipa_modify_expr): Likewise.
23868 (ipa_dump_param_adjustments): Likewise.
23869 (ipa_dump_agg_replacement_values): Likewise.
23870 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23871 * ipa-pure-const.c (check_stmt): Likewise.
23872 (pass_nothrow::execute): Likewise.
23873 * ipa-split.c (execute_split_functions): Likewise.
23874 * omp-offload.c (dump_oacc_loop_part): Likewise.
23875 (dump_oacc_loop): Likewise.
23876 * trans-mem.c (tm_log_emit): Likewise.
23877 (tm_memopt_accumulate_memops): Likewise.
23878 (dump_tm_memopt_set): Likewise.
23879 (dump_tm_memopt_transform): Likewise.
23880 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23881 (print_loop): Likewise.
23882 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23883 (chkp_gather_checks_info): Likewise.
23884 (chkp_get_check_result): Likewise.
23885 (chkp_remove_check_if_pass): Likewise.
23886 (chkp_use_outer_bounds_if_possible): Likewise.
23887 (chkp_reduce_bounds_lifetime): Likewise.
23888 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
23889 (chkp_mark_completed_bounds): Likewise.
23890 (chkp_register_incomplete_bounds): Likewise.
23891 (chkp_mark_invalid_bounds): Likewise.
23892 (chkp_maybe_copy_and_register_bounds): Likewise.
23893 (chkp_build_returned_bound): Likewise.
23894 (chkp_get_bound_for_parm): Likewise.
23895 (chkp_build_bndldx): Likewise.
23896 (chkp_get_bounds_by_definition): Likewise.
23897 (chkp_generate_extern_var_bounds): Likewise.
23898 (chkp_get_bounds_for_decl_addr): Likewise.
23899 * tree-chrec.c (chrec_apply): Likewise.
23900 * tree-data-ref.c (dump_data_reference): Likewise.
23901 (dump_subscript): Likewise.
23902 (dump_data_dependence_relation): Likewise.
23903 (analyze_overlapping_iterations): Likewise.
23904 * tree-inline.c (expand_call_inline): Likewise.
23905 (tree_function_versioning): Likewise.
23906 * tree-into-ssa.c (dump_defs_stack): Likewise.
23907 (dump_currdefs): Likewise.
23908 (dump_names_replaced_by): Likewise.
23909 (dump_update_ssa): Likewise.
23910 (update_ssa): Likewise.
23911 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23912 * tree-parloops.c (build_new_reduction): Likewise.
23913 (try_create_reduction_list): Likewise.
23914 (ref_conflicts_with_region): Likewise.
23915 (oacc_entry_exit_ok_1): Likewise.
23916 (oacc_entry_exit_single_gang): Likewise.
23917 * tree-pretty-print.h: Likewise.
23918 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23919 (get_scalar_evolution): Likewise.
23920 (add_to_evolution): Likewise.
23921 (get_loop_exit_condition): Likewise.
23922 (analyze_evolution_in_loop): Likewise.
23923 (analyze_initial_condition): Likewise.
23924 (analyze_scalar_evolution): Likewise.
23925 (instantiate_scev): Likewise.
23926 (number_of_latch_executions): Likewise.
23927 (gather_chrec_stats): Likewise.
23928 (final_value_replacement_loop): Likewise.
23929 (scev_const_prop): Likewise.
23930 * tree-sra.c (dump_access): Likewise.
23931 (disqualify_candidate): Likewise.
23932 (create_access): Likewise.
23933 (reject): Likewise.
23934 (maybe_add_sra_candidate): Likewise.
23935 (create_access_replacement): Likewise.
23936 (analyze_access_subtree): Likewise.
23937 (analyze_all_variable_accesses): Likewise.
23938 (sra_modify_assign): Likewise.
23939 (initialize_constant_pool_replacements): Likewise.
23940 (find_param_candidates): Likewise.
23941 (decide_one_param_reduction): Likewise.
23942 (replace_removed_params_ssa_names): Likewise.
23943 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
23944 * tree-ssa-copy.c (dump_copy_of): Likewise.
23945 (copy_prop_visit_cond_stmt): Likewise.
23946 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
23947 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
23948 (record_equivalences_from_stmt): Likewise.
23949 * tree-ssa-dse.c (compute_trims): Likewise.
23950 (delete_dead_call): Likewise.
23951 (delete_dead_assignment): Likewise.
23952 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
23953 (forward_propagate_into_cond): Likewise.
23954 (pass_forwprop::execute): Likewise.
23955 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23956 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
23958 (move_computations_worker): Likewise.
23959 (execute_sm): Likewise.
23960 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
23961 (remove_exits_and_undefined_stmts): Likewise.
23962 (remove_redundant_iv_tests): Likewise.
23963 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
23964 (adjust_iv_update_pos): Likewise.
23965 * tree-ssa-math-opts.c (bswap_replace): Likewise.
23966 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
23967 (value_replacement): Likewise.
23968 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
23969 * tree-ssa-pre.c (print_pre_expr): Likewise.
23970 (get_representative_for): Likewise.
23971 (create_expression_by_pieces): Likewise.
23972 (insert_into_preds_of_block): Likewise.
23973 (eliminate_insert): Likewise.
23974 (eliminate_dom_walker::before_dom_children): Likewise.
23975 (eliminate): Likewise.
23976 (remove_dead_inserted_code): Likewise.
23977 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
23978 * tree-ssa-reassoc.c (get_rank): Likewise.
23979 (eliminate_duplicate_pair): Likewise.
23980 (eliminate_plus_minus_pair): Likewise.
23981 (eliminate_not_pairs): Likewise.
23982 (undistribute_ops_list): Likewise.
23983 (eliminate_redundant_comparison): Likewise.
23984 (update_range_test): Likewise.
23985 (optimize_range_tests_var_bound): Likewise.
23986 (optimize_vec_cond_expr): Likewise.
23987 (rewrite_expr_tree): Likewise.
23988 (rewrite_expr_tree_parallel): Likewise.
23989 (linearize_expr): Likewise.
23990 (break_up_subtract): Likewise.
23991 (linearize_expr_tree): Likewise.
23992 (attempt_builtin_powi): Likewise.
23993 (attempt_builtin_copysign): Likewise.
23994 (transform_stmt_to_copy): Likewise.
23995 (transform_stmt_to_multiply): Likewise.
23996 (dump_ops_vector): Likewise.
23997 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
23998 (print_scc): Likewise.
23999 (set_ssa_val_to): Likewise.
24000 (visit_reference_op_store): Likewise.
24001 (visit_use): Likewise.
24002 (sccvn_dom_walker::before_dom_children): Likewise.
24003 (run_scc_vn): Likewise.
24004 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24006 (expr_hash_elt::print): Likewise.
24007 (const_and_copies::pop_to_marker): Likewise.
24008 (const_and_copies::record_const_or_copy_raw): Likewise.
24009 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24010 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24011 (dump_predicates): Likewise.
24012 (find_uninit_use): Likewise.
24013 (warn_uninitialized_phi): Likewise.
24014 (pass_late_warn_uninitialized::execute): Likewise.
24015 * tree-ssa.c (verify_vssa): Likewise.
24016 (verify_ssa): Likewise.
24017 (maybe_optimize_var): Likewise.
24018 * tree-vrp.c (dump_value_range): Likewise.
24019 (dump_all_value_ranges): Likewise.
24020 (dump_asserts_for): Likewise.
24021 (register_edge_assert_for_2): Likewise.
24022 (vrp_visit_cond_stmt): Likewise.
24023 (vrp_visit_switch_stmt): Likewise.
24024 (vrp_visit_stmt): Likewise.
24025 (vrp_visit_phi_node): Likewise.
24026 (simplify_cond_using_ranges_1): Likewise.
24027 (fold_predicate_in): Likewise.
24028 (evrp_dom_walker::before_dom_children): Likewise.
24029 (evrp_dom_walker::push_value_range): Likewise.
24030 (evrp_dom_walker::pop_value_range): Likewise.
24031 (execute_early_vrp): Likewise.
24033 2017-05-16 Richard Biener <rguenther@suse.de>
24035 * dwarf2out.c (loc_list_from_tree_1): Do not create
24036 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24038 2017-05-16 Richard Biener <rguenther@suse.de>
24040 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24042 (note_variable_value_in_expr): If we resolved the decl ref
24043 do not push to the stack.
24045 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24047 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24048 operations in fast-math mode.
24049 (vaddq_f16): Likewise.
24050 (vmul_f16): Likewise.
24051 (vmulq_f16): Likewise.
24052 (vsub_f16): Likewise.
24053 (vsubq_f16): Likewise.
24054 * config/arm/neon.md (add<mode>3): New.
24056 (fma:<VH:mode>3): New. Also remove outdated comment.
24059 2017-05-16 Martin Liska <mliska@suse.cz>
24063 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24064 (odr_types_equivalent_p): Likewise.
24066 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24068 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24070 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24073 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24074 non-interunit SSE move alternatives with '?'.
24075 (zero-extendsidi peephole2): New peephole to skip intermediate
24076 general register in SSE zero-extend sequence.
24078 2017-05-15 Jeff Law <law@redhat.com>
24080 * reorg.c (relax_delay_slots): Create a new variable to hold
24081 the temporary target rather than clobbering TARGET_LABEL.
24083 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24084 missing argument to extract_bit_field call.
24085 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24087 2017-05-15 Martin Liska <mliska@suse.cz>
24090 * gcc.c (process_command): Do not allow empty argument of -o option.
24092 2017-05-15 Renlin Li <renlin.li@arm.com>
24094 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24095 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24096 * config/aarch64/constraints.md (Usf): Add long call check.
24097 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24098 (call_value): Likewise.
24099 (sibcall): Likewise.
24100 (sibcall_value): Likewise.
24102 (call_value_insn): New.
24103 (sibcall_insn): Update rtx pattern.
24104 (sibcall_value_insn): Likewise.
24105 (call_internal): Remove.
24106 (call_value_internal): Likewise.
24107 (sibcall_internal): Likewise.
24108 (sibcall_value_internal): Likewise.
24109 (call_reg): Likewise.
24110 (call_symbol): Likewise.
24111 (call_value_reg): Likewise.
24112 (call_value_symbol): Likewise.
24114 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24117 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24119 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24121 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24122 compatible with CCGOCmode and with CCZmode.
24124 2017-05-14 Martin Sebor <msebor@redhat.com>
24126 PR middle-end/77671
24127 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24128 (gimple_fold_builtin_snprintf): Same.
24129 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24130 (gimple_fold_builtin_snprintf): Same.
24131 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24132 of character types.
24133 (is_call_safe): New function.
24134 (try_substitute_return_value): Call it.
24135 (try_simplify_call): New function.
24136 (pass_sprintf_length::handle_gimple_call): Call it.
24138 2017-05-14 Martin Sebor <msebor@redhat.com>
24140 PR middle-end/80669
24141 * builtins.c (expand_builtin_stpncpy): Simplify.
24143 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24145 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24146 * config/i386/i386.h
24147 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24148 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24149 (struct machine_function): Add new members call_ms2sysv,
24150 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24151 (struct machine_frame_state): New fields sp_realigned and
24152 sp_realigned_offset.
24153 * config/i386/i386.c
24154 (enum xlogue_stub): New enum.
24155 (enum xlogue_stub_sets): New enum.
24156 (class xlogue_layout): New class.
24157 (struct ix86_frame): New fields stack_realign_allocate_offset,
24158 stack_realign_offset and outlined_save_offset. Modify comments to
24159 detail stack layout when using out-of-line stubs.
24160 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24161 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24162 -mcall-ms2sysv-xlogues.
24163 (stub_managed_regs): New static variable.
24164 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24165 registers managed by out-of-line stub.
24166 (disable_call_ms2sysv_xlogues): New function.
24167 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24168 m->call_ms2sysv when appropriate and compute frame layout for
24170 (sp_valid_at, fp_valid_at): New inline functions.
24171 (choose_basereg): New function.
24172 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24174 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24175 Use align parameter of choose_baseaddr to generated aligned SSE movs
24177 (pro_epilogue_adjust_stack): Modify to track
24178 machine_frame_state::sp_realigned.
24179 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24180 (ix86_nsaved_sseregs): Likewise.
24181 (ix86_emit_save_regs): Likewise.
24182 (ix86_emit_save_regs_using_mov): Likewise.
24183 (ix86_emit_save_sse_regs_using_mov): Likewise.
24184 (get_scratch_register_on_entry): Likewise.
24185 (gen_frame_set): New function.
24186 (gen_frame_load): Likewise.
24187 (gen_frame_store): Likewise.
24188 (emit_outlined_ms2sysv_save): Likewise.
24189 (emit_outlined_ms2sysv_restore): Likewise.
24190 (ix86_expand_prologue): Modify stack re-alignment code and call
24191 emit_outlined_ms2sysv_save when appropriate.
24192 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24193 parameter rtx_insn *insn, which allows the function to be used to only
24194 generate the notes.
24195 (ix86_expand_epilogue): Modify validity checks of frame and stack
24196 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24197 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24198 * config/i386/predicates.md
24199 (save_multiple): New predicate.
24200 (restore_multiple): Likewise.
24201 * config/i386/sse.md
24202 (save_multiple<mode>): New pattern.
24203 (save_multiple_realign<mode>): Likewise.
24204 (restore_multiple<mode>): Likewise.
24205 (restore_multiple_and_return<mode>): Likewise.
24206 (restore_multiple_leave_return<mode>): Likewise.
24207 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24209 2017-05-14 Julia Koval <julia.koval@intel.com>
24211 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24212 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24213 (__builtin_ia32_xsetbv): New builtins.
24214 * config/i386/i386.c (ix86_expand_special_args_builtin):
24216 (ix86_expand_builtin): Special expand for new intrinsics.
24217 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24218 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24219 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24221 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24223 * cfganal.c (inverted_post_order_compute): Change argument type
24225 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24226 * df-core.c (rest_of_handle_df_initialize): Adjust.
24227 (rest_of_handle_df_finish): Likewise.
24228 (df_analyze_1): Likewise.
24229 (df_analyze): Likewise.
24230 (loop_inverted_post_order_compute): Change argument to be a vec *.
24231 (df_analyze_loop): Adjust.
24232 (df_get_n_blocks): Likewise.
24233 (df_get_postorder): Likewise.
24234 * df.h (struct df_d): Change field to be a vec.
24235 * lcm.c (compute_laterin): Adjust.
24236 (compute_available): Likewise.
24237 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24238 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24239 * tree-ssa-pre.c (compute_antic): Likewise.
24241 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24243 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24244 (depth_first_search::depth_first_search): Change structure init
24245 function to this constructor.
24246 (depth_first_search::add_bb): Rename function to this member.
24247 (depth_first_search::execute): Likewise.
24248 (flow_dfs_compute_reverse_finish): Adjust.
24250 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24252 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24253 (longest_simple_path): Likewise.
24254 * shrink-wrap.c (spread_components): Likewise.
24255 (disqualify_problematic_components): Likewise.
24256 (emit_common_heads_for_components): Likewise.
24257 (emit_common_tails_for_components): Likewise.
24258 (insert_prologue_epilogue_for_components): Likewise.
24260 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24262 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24265 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24267 * df-core.c (df_set_blocks): Start using auto_bitmap.
24268 (df_compact_blocks): Likewise.
24269 * df-problems.c (df_rd_confluence_n): Likewise.
24270 * df-scan.c (df_insn_rescan_all): Likewise.
24271 (df_process_deferred_rescans): Likewise.
24272 (df_update_entry_block_defs): Likewise.
24273 (df_update_exit_block_uses): Likewise.
24274 (df_entry_block_bitmap_verify): Likewise.
24275 (df_exit_block_bitmap_verify): Likewise.
24276 (df_scan_verify): Likewise.
24277 * lra-constraints.c (lra_constraints): Likewise.
24278 (undo_optional_reloads): Likewise.
24279 (lra_undo_inheritance): Likewise.
24280 * lra-remat.c (calculate_gen_cands): Likewise.
24281 (do_remat): Likewise.
24282 * lra-spills.c (assign_spill_hard_regs): Likewise.
24283 (spill_pseudos): Likewise.
24284 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24285 (bitmap_set_subtract_values): Likewise.
24287 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24289 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24290 management with auto_bitmap.
24291 (fix_inter_tick): Likewise.
24292 (fix_recovery_deps): Likewise.
24293 * ira.c (add_store_equivs): Likewise.
24294 (find_moveable_pseudos): Likewise.
24295 (split_live_ranges_for_shrink_wrap): Likewise.
24296 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24297 (rtx_reuse_manager::seen_def_p): Likewise.
24298 (rtx_reuse_manager::set_seen_def): Likewise.
24299 * print-rtl.h (class rtx_reuse_manager): Likewise.
24301 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24303 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24305 (migrate_btr_def): Likewise.
24306 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24307 * df-core.c (loop_post_order_compute): Likewise.
24308 (loop_inverted_post_order_compute): Likewise.
24309 * hsa-common.h: Likewise.
24310 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24311 * init-regs.c (initialize_uninitialized_regs): Likewise.
24312 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24313 (inline_small_functions): Likewise.
24314 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24315 * ira.c (combine_and_move_insns): Likewise.
24316 (build_insn_chain): Likewise.
24317 * loop-invariant.c (find_invariants): Likewise.
24318 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24319 * predict.c (tree_predict_by_opcode): Likewise.
24320 (predict_paths_leading_to): Likewise.
24321 (predict_paths_leading_to_edge): Likewise.
24322 (estimate_loops_at_level): Likewise.
24323 (estimate_loops): Likewise.
24324 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24325 (spread_components): Likewise.
24326 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24327 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24328 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24329 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24330 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24331 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24332 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24333 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24334 (mark_threaded_blocks): Likewise.
24335 (thread_through_all_blocks): Likewise.
24336 * tree-ssa.c (verify_ssa): Likewise.
24337 (execute_update_addresses_taken): Likewise.
24338 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24340 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24342 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24344 (post_order_compute): Likewise.
24345 (inverted_post_order_compute): Likewise.
24346 (pre_and_rev_post_order_compute_fn): Likewise.
24348 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24350 * genrecog.c (int_set::int_set): Explicitly construct our
24351 auto_vec base class.
24352 * vec.h (auto_vec::auto_vec): New constructor.
24354 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24356 * bitmap.h (class auto_bitmap): New constructor taking
24357 bitmap_obstack * argument.
24359 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24361 * bitmap.h (class auto_bitmap): Change type of m_bits to
24362 bitmap_head, and adjust ctor / dtor and member operators.
24364 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24366 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24367 when returned register mode doesn't match original mode.
24369 2017-05-12 Jeff Law <law@redhat.com>
24370 Jakub Jelinek <jakub@redhat.com>
24372 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24373 we look for cc setter after the compare-elim changes.
24374 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24375 within the vector to match what compare-elim now expects.
24376 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24377 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24379 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24380 after the compare-elim changes.
24381 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24382 the vector to match what compare-elim now expects.
24383 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
24384 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
24385 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
24386 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
24387 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
24389 * config/visium/visium.c (single_set_and_flags): Fix where
24390 we look for cc setter after the compare-elim changes.
24391 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
24392 with the vector to match what compare-elim now expects.
24393 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
24394 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
24395 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
24396 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
24397 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
24398 (neg<mode>2_insn_set_overflow): Likewise.
24400 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
24402 PR middle-end/79794
24403 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
24404 maybe_expand_insn call, set ops[0].target. If still set after call,
24405 set alt_rtl. Add extra arg to recursive calls.
24406 (extract_bit_field): Add alt_rtl argument. Pass to
24408 * expmed.h (extract_bit_field): Fix prototype.
24409 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
24410 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
24411 to extract_bit_field_calls.
24412 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
24413 Pass alt_rtl to extract_bit_field calls.
24414 * calls.c (store_unaligned_arguments_into_psuedos)
24415 load_register_parameters): Pass extra NULL to extract_bit_field calls.
24416 * optabs.c (maybe_legitimize_operand): Clear op->target when call
24418 * optabs.h (struct expand_operand): Add target bitfield.
24420 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24422 * compare-elim.c (try_eliminate_compare): Canonicalize
24423 operation with embedded compare to
24424 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
24425 (set (reg) (operation)].
24427 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
24429 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24432 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
24433 cost of adding a carry flag for ADC instruction.
24434 [case MINUS]: Ignore the cost of subtracting a carry flag
24435 for SBB instruction.
24437 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
24439 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
24441 * config/rs6000/bmiintrin.h: New file.
24442 * config/rs6000/bmi2intrin.h: New file.
24443 * config/rs6000/x86intrin.h: New file.
24445 2017-05-12 Jeff Law <law@redhat.com>
24447 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
24450 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
24452 PR middle-end/80707
24453 * tree-cfg.c: Remove cfg edges of unreachable case statements.
24455 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24457 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24458 early expansion of vector divide builtins.
24459 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
24460 builtins identified as having unsigned arguments.
24462 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24464 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
24465 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
24466 expansion of vector logical operations (and, andc, or, xor,
24469 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24471 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
24472 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
24474 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24476 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24477 early GIMPLE expansion of vector multiplies.
24479 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24481 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
24482 TARGET_HAVE_MOVT conditional.
24483 (movt splitter): Likewise.
24485 2017-05-12 Richard Biener <rguenther@suse.de>
24487 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
24488 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24489 Fold all stmts not inplace.
24491 2017-05-12 Richard Biener <rguenther@suse.de>
24493 PR tree-optimization/80713
24494 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
24495 inserted_exprs bit for not removed stmts.
24497 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
24499 PR middle-end/69921
24500 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
24501 parallelized" attribute for parallelized OpenACC kernels.
24502 * omp-offload.c (execute_oacc_device_lower): Use it.
24504 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
24505 Set "oacc kernels" attribute.
24506 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
24507 parameter. Adjust all users.
24508 (oacc_fn_attrib_kernels_p): Remove function.
24509 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
24510 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
24511 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
24512 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
24513 assert "oacc kernels" attribute is set.
24515 2017-05-11 Carl Love <cel@us.ibm.com>
24517 * config/rs6000/rs6000-c: Add support for built-in functions
24518 vector unsigned char vec_popcnt (vector signed char)
24519 vector unsigned char vec_popcnt (vector unsigned char)
24520 vector unsigned short vec_popcnt (vector signed short)
24521 vector unsigned short vec_popcnt (vector unsigned short)
24522 vector unsigned int vec_popcnt (vector signed int)
24523 vector unsigned int vec_popcnt (vector unsigned int)
24524 vector unsigned long long vec_popcnt (vector signed long long)
24525 vector unsigned long long vec_popcnt (vector unsigned long long)
24526 vector signed long long vec_slo (vector signed long long,
24527 vector signed char)
24528 vector signed long long vec_slo (vector signed long long,
24529 vector unsigned char)
24530 vector unsigned long long vec_slo (vector unsigned long long,
24531 vector signed char)
24532 vector unsigned long long vec_slo (vector unsigned long long,
24533 vector unsigned char)
24534 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
24535 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
24536 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
24537 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
24538 * doc/extend.texi: Update the built-in documentation file for the
24539 new built-in functions.
24541 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
24543 * attribs.h (sorted_attr_string): Move machine independent
24544 functions for target clone support from the i386 port to common
24545 code. Rename ix86_function_versions to common_function_versions.
24546 Rename make_name to make_unique_name.
24547 (common_function_versions): Likewise.
24548 (make_unique_name): Likewise.
24549 (make_dispatcher_decl): Likewise.
24550 (is_function_default_version): Likewise.
24551 * attribs.c (attr_strcmp): Likewise.
24552 (sorted_attr_string): Likewise.
24553 (common_function_versions): Likewise.
24554 (make_unique_name): Likewise.
24555 (make_dispatcher_decl): Likewise.
24556 (is_function_default_version): Likewise.
24557 * config/i386/i386.c (attr_strcmp): Likewise.
24558 (sorted_attr_string): Likewise.
24559 (ix86_function_versions): Likewise.
24560 (make_name): Likewise.
24561 (make_dispatcher_decl): Likewise.
24562 (is_function_default_version): Likewise.
24563 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
24565 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24568 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
24569 Account for direct move costs for vec_construct of integer
24572 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
24575 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
24576 (UNSPEC_STX_ATOMIC): Ditto.
24577 (loaddi_via_sse): New insn.
24578 (storedi_via_sse): Ditto.
24579 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
24580 Update corresponding peephole2 patterns.
24581 (atomic_storedi_fpu): Ditto.
24583 2017-05-11 Julia Koval <julia.koval@intel.com>
24585 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
24586 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
24588 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
24589 (__builtin_ia32_rsqrt14ss_mask): New builtins.
24590 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
24592 2017-05-11 Nathan Sidwell <nathan@acm.org>
24594 * graphite-poly.c: Include dumpfile.h.
24596 * dumpfle.h (dump_function): Declare here ...
24597 * tree-dump.h (dump_function): ... not here.
24598 * dumpfile.c: #include tree-cfg.h.
24599 (dump_function): Move here from ...
24600 * tree-dump.c (dump_function): ... here.
24601 * gimplify.c: #include splay-tree.h, not tree-dump.h.
24602 * graphite-poly.c: Don't include tree-dump.h.
24603 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
24604 * print-tree.c: Likewise.
24605 * stor-layout.c: Likewise.
24606 * tree-nested.c: Likewise.
24608 * dumpfile.c (dump_start): Use TDF_FLAGS.
24609 (dump_enable_all): Fix TDF_KIND check thinko.
24611 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
24613 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24614 array entries to represent two legal parameterizations of the
24615 overloaded __builtin_cmpb function, as represented by the
24616 P6_OV_BUILTIN_CMPB constant.
24617 (altivec_resolve_overloaded_builtin): Add special case handling
24618 for the __builtin_cmpb function, as represented by the
24619 P6_OV_BUILTIN_CMPB constant.
24620 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
24621 (BU_P6_64BIT_2): New macro.
24622 (BU_P6_OVERLOAD_2): New macro
24623 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
24624 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
24625 (CMPB): Add overload support to represent both 32-bit and 64-bit
24626 compare-bytes function.
24627 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
24628 support for TARGET_CMPB.
24629 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
24630 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24631 documentation of the __builtin_cmpb overloaded built-in function.
24633 2017-05-11 Richard Biener <rguenther@suse.de>
24635 PR tree-optimization/80705
24636 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
24637 bases are not vectorizable.
24639 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24641 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
24642 when counting register pressure.
24644 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24646 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
24647 (iv_ca_more_deps): Renamed to ...
24648 (iv_ca_compare_deps): ... this.
24649 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
24651 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24653 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
24655 (determine_group_iv_costs): ... here.
24656 (find_inv_vars_cb): Record inv var if it's not recorded before.
24658 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24660 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
24661 (get_shiftadd_cost): Ditto.
24663 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24665 * tree-ssa-address.c: Include header file.
24666 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
24668 (add_to_parts): Refactor.
24669 (addr_to_parts): New parameter. Update use of move_hint_to_base.
24670 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
24673 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24675 PR tree-optimization/53090
24676 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
24678 (extract_cond_operands): Detect condition with IV on both sides
24679 and return COMP_IV_EXPR_2.
24680 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
24681 (rewrite_use_compare): Simplify by removing call to function
24682 extract_cond_operands.
24684 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24686 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
24687 (extract_cond_operands): Detect condition comparing against non-
24688 invariant bound and return appropriate enum value.
24689 (find_interesting_uses_cond): Update use of extract_cond_operands.
24690 Handle its return value accordingly.
24691 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
24693 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24695 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
24696 nonlinear iv_use computation in loop invariant sensitive way.
24698 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24700 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
24701 (find_iv_candidates): Call relate_compare_use_with_all_cands.
24703 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24705 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
24706 (dump_cand): Support iv_cand.inv_exprs.
24707 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
24709 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
24712 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24714 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
24716 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
24717 as local function. Include necessary header files.
24718 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
24720 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24722 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
24724 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24726 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
24727 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
24728 RSHIFT_EXPR and BIT_NOT_EXPR.
24730 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24732 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
24733 (adjust_setup_cost): New parameter supporting round up adjustment.
24734 (struct address_cost_data): Delete.
24735 (force_expr_to_var_cost): Don't bound cost with spill_cost.
24736 (split_address_cost, ptr_difference_cost): Delete.
24737 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
24738 (struct ainc_cost_data): New struct.
24739 (get_address_cost_ainc): New function.
24740 (get_address_cost, get_computation_cost): Reimplement.
24741 (determine_group_iv_cost_address): Record inv_expr for all uses of
24743 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
24744 (iv_ca_has_deps): Reimplemented to ...
24745 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
24747 (iv_ca_extend): Call iv_ca_more_deps.
24749 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24751 * tree-ssa-address.c (struct mem_address): Move to header file.
24752 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
24753 * tree-ssa-address.h (struct mem_address): Move from C file.
24754 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
24756 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24758 * tree-affine.h (aff_combination_type): New interface.
24759 (aff_combination_zero_p): Remove static.
24760 (aff_combination_const_p): New interface.
24761 (aff_combination_singleton_var_p): New interfaces.
24763 2017-05-11 Richard Biener <rguenther@suse.de>
24765 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24766 Skip unreachable blocks and destinations.
24767 (eliminate): Move stmt removal and fixup ...
24768 (fini_eliminate): ... here. Skip inserted exprs.
24769 (pass_pre::execute): Move fini_pre after fini_eliminate.
24770 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
24771 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
24772 PRE to get rid of dead code that has invalid SSA form and
24773 split critical edges again.
24775 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24777 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24779 2017-05-11 Richard Biener <rguenther@suse.de>
24781 * passes.c (execute_function_todo): Verify loops if they are
24782 said to be up-to-date.
24783 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
24784 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
24786 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
24789 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
24790 handle calling assemble_external ourself.
24793 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
24794 modes with zero size. Enhance comment.
24796 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24798 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
24799 built-ins for vec_xl and vec_xst with short and char pointer
24802 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24804 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
24805 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
24806 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
24807 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
24808 (_mm_maskz_min_round_ss): New intrinsics.
24809 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
24810 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
24811 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
24812 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
24813 (__builtin_ia32_minss_mask_round): New builtins.
24814 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24815 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24816 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
24818 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
24819 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
24821 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
24824 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24826 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
24827 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
24828 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
24829 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
24830 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
24831 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24832 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24833 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24834 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24835 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24836 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24837 (__builtin_ia32_mulss_mask_round): New builtins.
24838 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24839 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24840 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24842 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24843 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24845 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24848 2017-05-10 Julia Koval <julia.koval@intel.com>
24850 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24851 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24852 (_mm256_setr_m128i): New intrinsics.
24854 2017-05-10 Julia Koval <julia.koval@intel.com>
24856 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24857 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24858 (_mm_maskz_rcp14_ss): New intrinsics.
24859 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24860 (__builtin_ia32_rcp14ss_mask): New builtins.
24861 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24863 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24865 PR tree-optimization/51513
24866 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24867 (assert_unreachable_fallthru_edge_p): Use it.
24868 (group_case_labels_stmt): Likewise.
24869 * tree-cfg.h: Prototype it.
24870 * stmt.c: Include cfghooks.h and tree-cfg.h.
24871 (emit_case_dispatch_table) <gap_label>: New local variable.
24872 Use it to fill dispatch table gaps.
24873 Test for default_label before updating probabilities.
24874 (expand_case) <default_label>: Remove unneeded initialization.
24875 Test for unreachable default case statement and remove its edge.
24876 Set default_label accordingly.
24877 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24879 2017-05-10 Carl Love <cel@us.ibm.com>
24881 * config/rs6000/rs6000-c: Add support for built-in functions
24882 vector signed char vec_neg (vector signed char)
24883 vector signed short int vec_neg (vector short int)
24884 vector signed int vec_neg (vector signed int)
24885 vector signed long long vec_neg (vector signed long long)
24886 vector float vec_neg (vector float)
24887 vector double vec_neg (vector double)
24888 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
24890 * config/rs6000/altivec.h: Add define for vec_neg
24891 * doc/extend.texi: Update the built-in documentation for the
24892 new built-in functions.
24894 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24896 PR tree-optimization/77644
24897 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
24899 2017-05-10 Nathan Sidwell <nathan@acm.org>
24901 * dumpfile.h (TDI_lang_all): New.
24902 (TDF_KIND): New. Renumber others
24903 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24905 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24907 (get_dump_file_name): Adjust suffix generation.
24908 (dump_enable_all): Use TDF_KIND.
24909 * doc/invoke.texi (-fdump-lang-all): Document.
24911 * dumpfile.h: Tabify.
24913 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24916 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24917 Move member access before delete.
24919 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24921 * tree-inline.c (expand_call_inline): Split block at stmt
24924 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24927 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24928 are now unused after splitting mov{sf,sd}_hardfloat.
24929 (f32_lr2): Likewise.
24930 (f32_lm): Likewise.
24931 (f32_lm2): Likewise.
24932 (f32_li): Likewise.
24933 (f32_li2): Likewise.
24934 (f32_lv): Likewise.
24935 (f32_sr): Likewise.
24936 (f32_sr2): Likewise.
24937 (f32_sm): Likewise.
24938 (f32_sm2): Likewise.
24939 (f32_si): Likewise.
24940 (f32_si2): Likewise.
24941 (f32_sv): Likewise.
24942 (f32_dm): Likewise.
24943 (f32_vsx): Likewise.
24944 (f32_av): Likewise.
24945 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
24946 For movsf, order stores so the VSX stores occur before the GPR
24947 store which encourages the register allocator to use a traditional
24948 FPR instead of a GPR. For movsd, order the stores so that the GPR
24949 store comes before the VSX stores to allow the power6 to work.
24950 This is due to the power6 not having a 32-bit integer store
24951 instruction from a FPR.
24952 (movsf_hardfloat): Likewise.
24953 (movsd_hardfloat): Likewise.
24955 2017-05-09 Martin Sebor <msebor@redhat.com>
24957 PR translation/80280
24958 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
24961 PR translation/80280
24962 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
24963 data member added in r247778.
24964 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
24966 2017-05-09 Nathan Sidwell <nathan@acm.org>
24968 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
24970 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
24973 2017-05-09 Marek Polacek <polacek@redhat.com>
24975 * doc/invoke.texi: Fix typo.
24977 2017-05-09 Richard Biener <rguenther@suse.de>
24979 * tree-vrp.c (vrp_val_is_max): Adjust comment.
24980 (vrp_val_is_min): Likewise.
24981 (set_value_range_to_value): Likewise.
24982 (set_value_range_to_nonnegative): Likewise.
24983 (gimple_assign_nonzero_p): Likewise.
24984 (gimple_stmt_nonzero_p): Likewise.
24985 (vrp_int_const_binop): Likewise. Remove unreachable case.
24986 (adjust_range_with_scev): Adjust comments.
24987 (compare_range_with_value): Likewise.
24988 (extract_range_from_phi_node): Likewise.
24989 (test_for_singularity): Likewise.
24991 2017-05-09 Richard Biener <rguenther@suse.de>
24993 * tree-vrp.c (get_single_symbol): Add assert that we don't
24994 get overflowed constants as invariant part.
24995 (compare_values_warnv): Add comment before the TREE_NO_WARNING
24996 checks. Use wi::cmp instead of recursing for integer constants.
24997 (compare_values): Just ignore whether we assumed undefined
24998 overflow instead of failing the compare.
24999 (extract_range_for_var_from_comparison_expr): Add comment before the
25000 TREE_NO_WARNING sets.
25001 (test_for_singularity): Likewise.
25002 (extract_range_from_comparison): Do not disable optimization
25003 when we assumed undefined overflow.
25004 (extract_range_basic): Remove init of unused var.
25006 2017-05-09 Richard Biener <rguenther@suse.de>
25008 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25009 (extract_range_from_multiplicative_op_1): Adjust.
25010 (extract_range_from_binary_expr_1): Use int_const_binop.
25012 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25015 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25016 rs6000_store_data_bypass_p in seven define_bypass directives and
25017 in several comments.
25018 * config/rs6000/rs6000-protos.h: Add prototype for
25019 rs6000_store_data_bypass_p function.
25020 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25021 function implements slightly different (rs6000-specific) semantics
25022 than store_data_bypass_p, returning false rather than aborting
25023 with assertion error when arguments do not satisfy the
25024 requirements of store data bypass.
25025 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25026 rs6000_store_data_bypass_p.
25028 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25030 * config/xtensa/xtensa-protos.h
25031 (xtensa_initial_elimination_offset): New declaration.
25032 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25033 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25034 macro definition, add case for FRAME_POINTER_REGNUM when
25035 FRAME_GROWS_DOWNWARD.
25036 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25037 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25038 xtensa_initial_elimination_offset.
25040 2017-05-08 Nathan Sidwell <nathan@acm.org>
25042 * doc/invoke.texi: Alphabetize -fdump options.
25044 2017-05-08 Martin Sebor <msebor@redhat.com>
25046 PR translation/80280
25047 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25049 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25051 * target.def (compute_frame_layout): New optional target hook.
25052 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25053 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25054 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25056 * reload1.c (verify_initial_elim_offsets): Likewise.
25057 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25058 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25059 (arm_get_frame_offsets): Split up into this ...
25060 (arm_compute_frame_layout): ... and this function.
25062 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25064 * config/aarch64/constraints.md (Usa): New constraint.
25065 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25067 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25069 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25070 with_multilib_list after it has been checked.
25072 2017-05-08 Richard Biener <rguenther@suse.de>
25074 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25075 (bitmap_set_subtract_values): Likewise.
25077 2017-05-08 Richard Biener <rguenther@suse.de>
25079 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25080 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25082 (gimple_stmt_nonzero_warnv_p): Rename to ...
25083 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25085 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25086 (extract_range_basic): Adjust, do not disable propagation on
25087 strict overflow sensitive simplification.
25088 (vrp_visit_cond_stmt): Likewise.
25090 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25092 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25093 body size unconditionally.
25095 2017-05-07 Jeff Law <law@redhat.com>
25098 2017-05-06 Jeff Law <law@redhat.com>
25099 PR tree-optimization/78496
25100 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25103 PR tree-optimization/78496
25104 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25105 (simplify_stmt_using_ranges): Call it.
25106 (vrp_dom_walker::before_dom_children): Extract equivalences
25107 from an ASSERT_EXPR with an equality comparison against a
25110 2017-05-06 Jeff Law <law@redhat.com>
25112 PR tree-optimization/78496
25113 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25116 PR tree-optimization/78496
25117 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25118 (simplify_stmt_using_ranges): Call it.
25119 (vrp_dom_walker::before_dom_children): Extract equivalences
25120 from an ASSERT_EXPR with an equality comparison against a
25123 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25125 * lra-constraints.c (lra_copy_reg_equiv): New function.
25126 (split_reg): Use it to copy equivalence information from the
25127 original register to the spill register.
25129 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25131 PR rtl-optimization/75964
25132 * simplify-rtx.c (simplify_const_relational_operation): Remove
25133 invalid handling of comparisons of integer ABS.
25135 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25137 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25138 initialize to zero.
25139 (init_regs): Remove declaration.
25140 (function_arg_advance_32): Initialize error_p as boolean variable.
25142 2017-05-05 Nathan Sidwell <nathan@acm.org>
25144 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25145 lines. Use for (;;).
25147 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25149 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25150 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25151 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25152 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25153 VF=2 that require versioning.
25155 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25157 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25160 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25162 * diagnostic.h (diagnostic_override_option_index): Convert from
25163 macro to inline function.
25165 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25167 * diagnostic.c (last_module_changed_p): New function.
25168 (set_last_module): New function.
25169 (diagnostic_report_current_module): Convert macro usage to
25170 the above functions.
25171 * diagnostic.h (diagnostic_context::last_module): Strengthen
25172 from const line_map * to const line_map_ordinary *.
25173 (diagnostic_last_module_changed): Delete macro.
25174 (diagnostic_set_last_module): Delete macro.
25176 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25178 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25179 with diagnostic_report_diagnostic.
25180 (diagnostic_n_impl_richloc): Likewise.
25181 * diagnostic.h (report_diagnostic): Delete macro.
25182 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25183 with diagnostic_report_diagnostic.
25184 * substring-locations.c (format_warning_va): Likewise.
25186 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25188 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25189 save/restor of format_spec. Move option-printing code to...
25190 (print_option_information): ...this new function, and
25191 reimplement by simply printing to the pretty_printer,
25192 rather than appending to the format string.
25194 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25196 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25197 handling logic into...
25198 (update_effective_level_from_pragmas): ...this new function.
25200 2017-05-04 Andrew Waterman <andrew@sifive.com>
25202 * config/riscv/riscv.opt (mstrict-align): New option.
25203 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25204 (SLOW_UNALIGNED_ACCESS): Define.
25205 (riscv_slow_unaligned_access): Declare.
25206 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25208 (riscv_slow_unaligned_access): New variable.
25209 (rocket_tune_info): Set slow_unaligned_access to true.
25210 (optimize_size_tune_info): Set slow_unaligned_access to false.
25211 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25212 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25213 (riscv_option_override): Set riscv_slow_unaligned_access.
25214 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25216 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25218 * config/riscv/riscv.md: Unify indentation.
25220 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25225 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25227 (extendsi<mode>2): Add support for doing sign extension via
25228 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25229 don't have ISA 3.0 instructions.
25230 (extendsi<mode>2 splitter): Likewise.
25231 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25232 generate the normal insns since SImode can now go in vector
25233 registers. Disallow the special UNSPECs needed for previous
25234 machines to hide SImode being used. Add new insns
25235 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25236 (fix_trunc<mode>si2_stfiwx): Likewise.
25237 (fix_trunc<mode>si2_internal): Likewise.
25238 (fixuns_trunc<mode>si2): Likewise.
25239 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25240 (fctiw<u>z_<mode>_smallint): Likewise.
25241 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25242 of floating point to 32-bit integer from doing a direct move to
25243 the GPR registers to do a store.
25244 (fctiwz_<mode>): Break long line.
25246 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25248 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25249 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25250 (addr_list, addr_offset_valid_p): New.
25251 (split_address_groups): Check offset validity with above function.
25252 (gt-tree-ssa-loop-ivopts.h): Include header file.
25254 2017-05-05 Nathan Sidwell <nathan@acm.org>
25256 * config.gcc (arm*-*-*): Add missing 'fi'.
25258 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25260 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25261 included in -fopt-info does not matter.
25262 * doc/optinfo.texi (-fopt-info): Fix description of default
25263 behavour. Explicitly say order of options included in -fopt-info
25266 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25268 * config.gcc: Allow combinations of aprofile and rmprofile values for
25269 --with-multilib-list.
25270 * config/arm/t-multilib: New file.
25271 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25272 variables. Remove setting of ISA and floating-point ABI in
25273 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25274 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25275 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25276 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25278 * config/arm/t-rmprofile: Likewise except for the matches changes.
25279 * doc/install.texi (--with-multilib-list): Document the combination of
25280 aprofile and rmprofile values and warn about pitfalls in doing that.
25282 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25284 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25285 (movdi_aarch64): Likewise.
25287 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25289 PR tree-optimization/80632
25290 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25292 (build_arrays): Initialize it for virtual phis.
25293 (fix_phi_nodes): Use it for virtual phis.
25295 PR tree-optimization/80558
25296 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25297 [x, y] op z into [x op, y op z] for op & or | if conditions
25300 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25301 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25304 * config/arm/arm.md (use_literal_pool): Remove.
25305 (64-bit immediate split): No longer takes cost into consideration
25306 if arm_disable_literal_pool is enabled.
25307 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25308 used when arm_disable_literal_pool is enabled.
25309 (arm_max_const_double_inline_cost): Remove use of
25310 arm_disable_literal_pool.
25311 (push_minipool_fix): Add assert.
25312 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25313 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25314 (no_literal_pool_sf_immediate): New.
25316 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25318 PR tree-optimization/80613
25319 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25320 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25322 2017-05-05 Richard Biener <rguenther@suse.de>
25324 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25326 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25328 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25329 of this flag from insn conditions due to removal from r247495.
25331 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25333 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25335 (arm_early_store_addr_dep_ptr): Likewise.
25336 * config/arm/aarch-common-protos.h
25337 (arm_early_load_addr_dep_ptr): Add prototype.
25338 (arm_early_store_addr_dep_ptr): Likewise.
25339 * config/arm/cortex-a53.md: Add new bypasses.
25341 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25343 * tree.c (next_type_uid): Change type to unsigned.
25344 (type_hash_canon): Decrement back next_type_uid if
25345 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25346 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25349 2017-05-04 Martin Sebor <msebor@redhat.com>
25351 * builtins.c: Fix a trivial typo in a comment.
25353 PR middle-end/79234
25354 * builtins.c (check_sizes): Adjust to handle reading past the end.
25355 Avoid printing excessive upper bound of ranges. Use %E to print
25356 tree nodes instead of converting them to %wu.
25357 (expand_builtin_memchr): New function.
25358 (compute_dest_size): Rename...
25359 (compute_objsize): ...to this.
25360 (expand_builtin_memcpy): Adjust.
25361 (expand_builtin_mempcpy): Adjust.
25362 (expand_builtin_strcat): Adjust.
25363 (expand_builtin_strcpy): Adjust.
25364 (check_strncat_sizes): Adjust.
25365 (expand_builtin_strncat): Adjust.
25366 (expand_builtin_strncpy): Adjust and simplify.
25367 (expand_builtin_memset): Adjust.
25368 (expand_builtin_bzero): Adjust.
25369 (expand_builtin_memcmp): Adjust.
25370 (expand_builtin): Handle memcmp.
25371 (maybe_emit_chk_warning): Check strncat just once.
25373 2017-05-04 Martin Sebor <msebor@redhat.com>
25375 PR preprocessor/79214
25376 PR middle-end/79222
25377 PR middle-end/79223
25378 * builtins.c (check_sizes): Add inlining context and issue
25379 warnings even when -Wno-system-headers is set.
25380 (check_strncat_sizes): Same.
25381 (expand_builtin_strncat): Same.
25382 (expand_builtin_memmove): New function.
25383 (expand_builtin_stpncpy): Same.
25384 (expand_builtin): Handle memmove and stpncpy.
25386 2017-05-04 Bin Cheng <bin.cheng@arm.com>
25388 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
25389 which is not used any more.
25391 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25393 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
25395 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25397 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
25398 (cortexa53_tunings): Likewise.
25399 (cortexa57_tunings): Likewise.
25400 (cortexa72_tunings): Likewise.
25401 (cortexa73_tunings): Likewise.
25403 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25405 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
25406 Set loop alignment to 8.
25408 2017-05-04 Martin Sebor <msebor@redhat.com>
25410 PR translation/80280
25411 * builtins.c (expand_builtin_object_size): Add missing quoting to
25412 %D and like directives.
25413 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
25414 (hsa_type_for_tree_type): Same.
25415 (verify_function_arguments): Same.
25416 * symtab.c (symbol_table::change_decl_assembler_name): Same.
25417 * varasm.c (get_section): Same.
25420 2017-05-04 Martin Sebor <msebor@redhat.com>
25422 PR translation/80280
25423 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
25425 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25427 * config/aarch64/aarch64.c (generic_addrcost_table):
25428 Change HI/TI mode setting.
25430 2017-05-04 Martin Jambor <mjambor@suse.cz>
25432 PR tree-optimization/80622
25433 * tree-sra.c (comes_initialized_p): New function.
25434 (build_accesses_from_assign): Only set write lazily when
25435 comes_initialized_p is false.
25436 (analyze_access_subtree): Use comes_initialized_p.
25437 (propagate_subaccesses_across_link): Assert !comes_initialized_p
25438 instead of testing for PARM_DECL.
25440 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25442 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
25443 constraint on operand 0 to allow more general addressing modes.
25444 Adjust output template.
25445 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
25447 * config/aarch64/aarch64-protos.h
25448 (aarch64_address_valid_for_prefetch_p): Declare prototype.
25449 * config/aarch64/constraints.md (Dp): New address constraint.
25450 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
25453 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
25455 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
25456 update use of estimate_ipcp_clone_size_and_time.
25457 (estimate_local_effects): Update use of
25458 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
25459 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
25460 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
25461 Return nonspecialized time.
25463 2017-05-04 Richard Biener <rguenther@suse.de>
25465 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
25466 for the last VUSE which def dominates the PHI. Directly call
25468 (get_continuation_for_phi_1): Remove.
25470 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
25472 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
25473 to explain the use of truncating division. Cap the number of
25474 iterations to the maximum given by nb_iterations_upper_bound,
25477 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25479 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
25480 * configure: Regenerate.
25481 * config.in: Regenerate.
25482 * config/i386/driver-mingw32.c: new file.
25483 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
25484 * config.host: Link driver-mingw32.o on MinGW host.
25485 * doc/install.texi: Document new --enable-mingw-wildcard configure
25488 2017-05-04 Marek Polacek <polacek@redhat.com>
25490 PR tree-optimization/80612
25491 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
25493 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25494 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
25496 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
25497 (movt splitter): Likewise.
25498 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
25499 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
25500 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
25501 block for Thumb-1 with MOVT.
25502 (thumb2_legitimate_address_p): Move code block ...
25503 (can_avoid_literal_pool_for_label_p): ... into this new function.
25504 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
25506 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
25507 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
25508 "M-profile targets with the MOVT instruction".
25510 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25512 * config/arm/arm-builtins.c (arm_init_builtins): Rename
25513 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
25514 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
25516 2017-05-04 Martin Liska <mliska@suse.cz>
25518 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
25519 variable cond_code.
25521 2017-05-04 Richard Biener <rguenther@suse.de>
25523 * tree.c (array_at_struct_end_p): Handle arrays at struct
25524 end with flexarrays more conservatively. Refactor and treat
25525 arrays of arrays or aggregates more strict. Fix
25526 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
25527 * tree.c (array_at_struct_end_p): Adjust prototype.
25528 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
25529 * gimple-fold.c (get_range_strlen): Likewise.
25530 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
25532 2017-05-04 Richard Biener <rguenther@suse.de>
25534 PR tree-optimization/31130
25535 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
25537 (supports_overflow_infinity): Likewise.
25538 (is_negative_overflow_infinity): Likewise.
25539 (is_positive_overflow_infinity): Likewise.
25540 (is_overflow_infinity): Likewise.
25541 (stmt_overflow_infinity): Likewise.
25542 (overflow_infinity_range_p): Likewise.
25543 (usable_range_p): Remove as always returning true.
25544 (make_overflow_infinity): Remove.
25545 (negative_overflow_infinity): Likewise.
25546 (positive_overflow_infinity): Likewise.
25547 (avoid_overflow_infinity): Likewise.
25548 (set_value_range): Adjust accordingly.
25549 (set_value_range_to_nonnegative): Likewise, remove now unused
25550 overflow_infinity arg.
25551 (vrp_operand_equal_p): Adjust.
25552 (update_value_range): Likewise.
25553 (range_int_cst_singleton_p): Likewise.
25554 (operand_less_p): Likewise.
25555 (compare_values_warnv): Likewise.
25556 (extract_range_for_var_from_comparison_expr): Likewise.
25557 (vrp_int_const_binop): Likewise.
25558 (zero_nonzero_bits_from_vr): Likewise.
25559 (extract_range_from_multiplicative_op_1): Likewise.
25560 (extract_range_from_binary_expr_1): Likewise.
25561 (extract_range_from_unary_expr): Likewise.
25562 (extract_range_from_comparison): Likewise.
25563 (extract_range_basic): Likewise.
25564 (adjust_range_with_scev): Likewise.
25565 (compare_ranges): Likewise.
25566 (compare_range_with_value): Likewise.
25567 (dump_value_range): Likewise.
25568 (test_for_singularity): Likewise, remove strict_overflow_p parameter
25570 (simplify_cond_using_ranges): Adjust.
25572 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25574 * brig-builtins.def: Added a builtin for class_f64.
25575 * builtin-types.def: Added a builtin type needed by class_f64.
25577 2017-05-03 Jason Merrill <jason@redhat.com>
25579 * timevar.def: Add TV_CONSTEXPR.
25581 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25583 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
25585 2017-05-03 Martin Jambor <mjambor@suse.cz>
25587 * ipa-prop.c (ipa_update_after_lto_read): Removed.
25588 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
25589 * ipa-cp.c (ipcp_propagate_stage): Do not call
25590 ipa_update_after_lto_read.
25591 * ipa-inline.c (ipa_inline): Likewise.
25593 2017-05-03 Martin Jambor <mjambor@suse.cz>
25595 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
25596 tag. Added a default constructor and a destructor.
25597 (ipa_edge_args_sum_t): New class;
25598 (ipa_edge_args_sum): Declare.
25599 (ipa_edge_args_vector): Remove declaration.
25600 (IPA_EDGE_REF): Use ipa_edge_args_sum.
25601 (ipa_free_edge_args_substructures): Remove declaration.
25602 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
25603 (ipa_edge_args_info_available_for_edge_p): Likewise.
25604 * ipa-prop.c (ipa_edge_args_vector): Removed.
25605 (edge_removal_hook_holder): Likewise.
25606 (edge_duplication_hook_holder): Likewise.
25607 (ipa_edge_args_sum): New variable.
25608 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
25609 ipa_edge_args_vector.
25610 (ipa_free_edge_args_substructures): Likewise.
25611 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
25612 ipa_edge_args_vector.
25613 (ipa_edge_removal_hook): Turned into method
25614 ipa_edge_args_sum_t::remove.
25615 (ipa_edge_duplication_hook): Turned into method
25616 ipa_edge_args_sum_t::duplicate.
25617 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
25618 registering edge hooks.
25619 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
25620 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
25621 ipa_edge_args_sum instead of ipa_edge_args_vector.
25622 * ipa-profile.c (ipa_profile): Likewise.
25624 2017-05-03 Martin Jambor <mjambor@suse.cz>
25626 * symbol-summary.h (function_summary): New method exists.
25627 (function_summary::symtab_removal): Deallocate through release.
25628 (call_summary): New class.
25629 (gt_ggc_mx): New overload.
25630 (gt_pch_nx): Likewise.
25631 (gt_pch_nx): Likewise.
25633 2017-05-03 Jeff Law <law@redhat.com>
25635 PR tree-optimization/78496
25636 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
25637 from simplify_cond_using_ranges. Split off code to walk
25638 backwards through casts into ...
25639 (simplify_cond_using_ranges_2): New function.
25640 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
25641 (execute_vrp): After identifying jump threads, call
25642 simplify_cond_using_ranges_2.
25644 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
25647 * ipa-inline.h (inline_summary): Add ctor.
25648 (create_ggc): Do not use ggc_cleared_alloc.
25650 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
25651 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25653 * gcc.c (handle_braces): Support escaping in switch matching
25655 * doc/invoke.texi (Spec Files): Document it.
25656 Remove superfluous @code markup in items.
25658 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25660 * diagnostic-show-locus.c (struct column_range): New struct.
25661 (get_affected_columns): New function.
25662 (get_printed_columns): New function.
25663 (struct correction): New struct.
25664 (correction::ensure_capacity): New function.
25665 (correction::ensure_terminated): New function.
25666 (struct line_corrections): New struct.
25667 (line_corrections::~line_corrections): New dtor.
25668 (line_corrections::add_hint): New function.
25669 (layout::print_trailing_fixits): Reimplement in terms of the new
25671 (selftest::test_overlapped_fixit_printing): New function.
25672 (selftest::diagnostic_show_locus_c_tests): Call it.
25674 2017-05-03 Nathan Sidwell <nathan@acm.org>
25676 Canonicalize canonical type hashing
25677 * tree.h (type_hash_canon_hash): Declare.
25678 * tree.c (type_hash_list, attribute_hash_list): Move into
25679 type_hash_canon_hash.
25680 (build_type_attribute_qual_variant): Break out hash code calc into
25681 type_hash_canon_hash.
25682 (type_hash_canon_hash): New. Generic type hash computation.
25683 (build_range_type_1, build_array_type_1, build_function_type,
25684 build_method_type_directly, build_offset_type, build_complex_type,
25685 make_vector_type): Call it.
25687 2017-05-03 Richard Biener <rguenther@suse.de>
25689 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
25690 When all DRs have unknown misaligned do not always peel
25691 when there is a store but apply the same costing model as if
25692 there were only loads.
25694 2017-05-03 Richard Biener <rguenther@suse.de>
25697 PR tree-optimization/80492
25698 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25699 compare_base_decls returning dont-know properly.
25701 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
25703 * config/arm/iterators.md (CCSI): New mode iterator.
25704 (arch): New mode attribute.
25705 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
25706 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
25707 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
25708 code iterator for success result mode.
25709 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
25710 the corresponding new insn generators.
25712 2017-05-03 Bin Cheng <bin.cheng@arm.com>
25715 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25716 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25718 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
25720 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
25721 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
25722 (DDR_A): Wrap DDR argument in brackets.
25723 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
25724 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
25725 (DDR_REVERSED_P): Likewise.
25727 2017-05-03 Jakub Jelinek <jakub@redhat.com>
25729 PR tree-optimization/79472
25730 * tree-switch-conversion.c (struct switch_conv_info): Add
25731 contiguous_range and default_case_nonstandard fields.
25732 (collect_switch_conv_info): Compute contiguous_range and
25733 default_case_nonstandard fields, don't clear final_bb if
25734 contiguous_range and only the default case doesn't have the required
25736 (check_all_empty_except_final): Set default_case_nonstandard instead
25737 of failing if contiguous_range and the default case doesn't have empty
25739 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
25740 and only the default case doesn't have the required constants. Skip
25742 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
25743 if default_case_nonstandard.
25744 (build_constructors): Build constant 1 just once. Assert that default
25745 values aren't inserted in between cases if contiguous_range. Skip
25747 (build_arrays): Skip virtual phis.
25748 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
25749 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
25750 Handle virtual phis.
25751 (gen_inbound_check): Handle default_case_nonstandard case.
25752 (process_switch): Adjust check_final_bb caller. Call
25753 gather_default_values with the first non-default case instead of
25754 default case if default_case_nonstandard.
25756 2017-05-02 Nathan Sidwell <nathan@acm.org>
25758 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
25759 check. Fix formatting.
25761 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
25763 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
25764 errors when comparing specialized and unspecialized times.
25766 2017-05-02 David Malcolm <dmalcolm@redhat.com>
25768 * diagnostic-show-locus.c
25769 (layout::should_print_annotation_line_p): Make private.
25770 (layout::print_annotation_line): Make private.
25771 (layout::annotation_line_showed_range_p): Make private.
25772 (layout::show_ruler): Make private.
25773 (layout::print_source_line): Make private. Pass in line and
25774 line_width, rather than calling location_get_source_line. Drop
25776 (layout::print_leading_fixits): New method.
25777 (layout::print_any_fixits): Rename to...
25778 (layout::print_trailing_fixits): ...this, and make private.
25779 Don't print newline fixits.
25780 (diagnostic_show_locus): Move logic for printing one row into...
25781 (layout::print_line): ...this new function. Move the
25782 location_get_source_line call and error-handling from
25783 print_source_line to here. Call print_leading_fixits, and rename
25784 print_any_fixits to print_trailing_fixits.
25785 (selftest::test_fixit_insert_containing_newline): Update now that
25786 newlines are partially supported.
25787 (selftest::test_fixit_insert_containing_newline_2): New test.
25788 (selftest::test_fixit_replace_containing_newline): Update comments.
25789 (selftest::diagnostic_show_locus_c_tests): Call the new test.
25790 * edit-context.c (class added_line): New class.
25791 (class edited_line): Describe newline handling in comment.
25792 (edited_line::actually_edited_p): New method.
25793 (edited_line::print_content): Delete redundant decl.
25794 (edited_line::m_predecessors): New field.
25795 (edited_file::print_content): Call edited_line::print_content.
25796 (edited_file::print_diff): Update to support newlines.
25797 (edited_file::print_diff_hunk): Likewise.
25798 (edited_file::print_run_of_changed_lines): New function.
25799 (edited_file::print_diff_line): Convert to...
25800 (print_diff_line): ...this.
25801 (edited_file::get_effective_line_count): New function.
25802 (edited_line::edited_line): Initialize new field m_predecessors.
25803 (edited_line::~edited_line): Clean up m_predecessors.
25804 (edited_line::apply_fixit): Handle newlines.
25805 (edited_line::get_effective_line_count): New function.
25806 (edited_line::print_content): New function.
25807 (edited_line::print_diff_lines): New function.
25808 (selftest::test_applying_fixits_insert_containing_newline): New
25810 (selftest::test_applying_fixits_replace_containing_newline): New
25812 (selftest::insert_line): New function.
25813 (selftest::test_applying_fixits_multiple_lines): Add example of
25815 (selftest::edit_context_c_tests): Call the new tests.
25817 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25819 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
25820 parameter cand. Update dump information.
25821 (get_computation_cost): Update uses.
25823 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25825 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
25826 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
25827 (get_computation_at, rewrite_use_address): Update use of
25828 get_computation_aff.
25830 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25832 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25833 (get_computation): Delete.
25834 (get_computation_cost): Implement like get_computation_cost_at.
25835 Use get_computation_at.
25836 (get_computation_cost_at): Delete.
25837 (rewrite_use_nonlinear_expr): Use get_computation_at.
25838 (rewrite_use_compare, remove_unused_ivs): Ditto.
25840 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25842 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25844 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25846 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25847 (ivopts_global_cost_for_size): Rename parameter and update uses.
25848 (iv_ca_recount_cost): Update uses.
25849 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25850 candidates seperately in n_invs and n_cands.
25851 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25853 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25855 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25856 (find_inv_vars_cb): New.
25857 (find_depends): Renamed to ...
25858 (find_inv_vars): ... this.
25859 (add_candidate_1, force_var_cost): Call find_inv_vars.
25860 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25862 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25864 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25865 inv_vars. Add inv_exprs.
25866 (struct iv_cand): Rename depends_on to inv_vars.
25867 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25868 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25869 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25870 (dump_cand): Dump inv_vars.
25871 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25872 (record_invariant, find_depends, add_candidate_1): Ditto.
25873 (set_group_iv_cost, force_var_cost): Ditto.
25874 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25875 (get_computation_cost_at, get_computation_cost): Ditto.
25876 (determine_group_iv_cost_generic): Ditto.
25877 (determine_group_iv_cost_address): Ditto.
25878 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25879 (determine_group_iv_costs): Ditto.
25880 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25881 (iv_ca_set_remove_invariants): Renamed to ...
25882 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25883 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25884 (iv_ca_set_add_invariants): Renamed to ...
25885 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25886 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25887 (iv_ca_has_deps): Support inv_vars and inv_exprs.
25888 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
25889 (create_new_ivs): Remove useless dump.
25891 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25893 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
25895 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
25896 (iv_ca_set_no_cp, create_new_iv): Ditto.
25898 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25900 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25902 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25904 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25905 function tree_check2.
25907 2017-05-02 Martin Liska <mliska@suse.cz>
25909 * doc/gcov.texi: Add missing preposition.
25910 * gcov.c (function_info::function_info): Properly fill up
25911 all member variables.
25913 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25915 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25917 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25919 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25921 2017-05-02 Martin Liska <mliska@suse.cz>
25924 * lto-streamer-in.c (lto_read_tree_1): Remove
25925 LTO_STREAMER_DEBUG.
25926 * lto-streamer.c (struct tree_hash_entry): Likewise.
25927 (struct tree_entry_hasher): Likewise.
25928 (tree_entry_hasher::hash): Likewise.
25929 (tree_entry_hasher::equal): Likewise.
25930 (lto_streamer_init): Likewise.
25931 (lto_orig_address_map): Likewise.
25932 (lto_orig_address_get): Likewise.
25933 (lto_orig_address_remove): Likewise.
25934 * lto-streamer.h: Likewise.
25935 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
25936 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
25938 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
25940 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
25941 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
25942 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
25943 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
25944 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
25945 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
25946 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
25947 (mm_maskz_sub_ss): New intrinsics.
25948 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25949 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25950 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
25951 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
25952 (__builtin_ia32_subss_mask_round): New builtins.
25953 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25954 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25955 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
25957 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
25958 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25960 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25963 2017-05-02 Martin Jambor <mjambor@suse.cz>
25965 PR tree-optimization/78687
25966 * tree-sra.c (access): New field parent.
25967 (process_subtree_disqualification): New function.
25968 (disqualify_candidate): Call it.
25969 (build_accesses_from_assign): Reset write flag if creating an
25971 (build_access_subtree): Fill in parent field and also prpagate
25972 down grp_write flag.
25973 (create_artificial_child_access): New parameter set_grp_write, set
25974 grp_write to its value.
25975 (propagate_subaccesses_across_link): Also propagate grp_write flag
25977 (propagate_all_subaccesses): Push the closest parent back to work
25978 queue if add_access_to_work_queue returned true.
25980 2017-05-02 Richard Biener <rguenther@suse.de>
25982 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25983 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25984 -fstrict-overflow documentation.
25985 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25986 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
25987 flag_strict_overflow.
25988 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
25989 * lto-opts.c (lto_write_options): Do not stream it.
25990 * lto-wrapper.c (merge_and_complain): Do not handle it.
25991 * opts.c (default_options_table): Do not set -fstrict-overflow.
25992 (finish_options): Likewise do not clear it when sanitizing.
25993 * simplify-rtx.c (simplify_const_relational_operation): Do not
25994 test flag_strict_overflow.
25996 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25998 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
25999 using enabled attribute.
26000 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26001 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26002 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26003 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26004 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26005 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26006 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26007 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26008 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26009 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26011 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26013 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26015 2017-05-02 Richard Biener <rguenther@suse.de>
26017 PR tree-optimization/80591
26019 2017-04-10 Richard Biener <rguenther@suse.de>
26021 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26024 2017-05-02 Richard Biener <rguenther@suse.de>
26026 PR tree-optimization/80549
26027 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26028 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26029 headers if they do not have a preheader.
26031 2017-05-02 Martin Liska <mliska@suse.cz>
26034 * common.opt: Fix typo.
26035 * doc/invoke.texi: Likewise.
26037 2017-05-01 Jan Beulich <jbeulich@suse.com>
26039 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26040 swapping, add (x,x,m,x,n) alternative.
26042 2017-05-01 Nathan Sidwell <nathan@acm.org>
26044 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26045 unnecessary unadjusted_alignment check.
26047 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26050 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26052 * gimplify.c (gimplify_cilk_detach): New function.
26053 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26054 * tree-core.h: Document EXPR_CILK_SPAWN.
26055 * tree.h (EXPR_CILK_SPAWN): Define.
26057 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26059 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26060 to use new fixit_hint representation, using the "replace" logic.
26061 (get_line_span_for_fixit_hint): Likewise.
26062 (layout::print_any_fixits): Likewise.
26063 (selftest::test_one_liner_many_fixits): Rename to...
26064 (selftest::test_one_liner_many_fixits_1): ...this, and update
26065 comment and expected output to reflect that the multiple fix-it
26066 hints are now consolidated into one insertion.
26067 (selftest::test_one_liner_many_fixits_2): New test.
26068 (selftest::test_diagnostic_show_locus_one_liner): Update for
26070 (selftest::test_fixit_consolidation): Update for fix-it API
26072 * diagnostic.c (print_parseable_fixits): Likewise.
26073 * edit-context.c (edited_line::m_line_events): Convert from
26074 auto_vec <line_event *> to auto_vec <line_event>.
26075 (class line_event): Convert from abstract base class to a concrete
26076 class, taking over the role of replace_event.
26077 (class insert_event): Delete.
26078 (class replace_event): Rename to class line_event. Convert to
26080 (edit_context::add_fixits): Reimplement.
26081 (edit_context::apply_insert): Delete.
26082 (edit_context::apply_replace): Rename to...
26083 (edit_context::apply_fixit): ...this. Convert to half-open range.
26084 (edited_file::apply_insert): Delete.
26085 (edited_file::apply_replace): Rename to...
26086 (edited_file::apply_fixit): ...this.
26087 (edited_line::~edited_line): Drop deletion of events.
26088 (edited_line::apply_insert): Delete.
26089 (edited_line::apply_replace): Rename to...
26090 (edited_line::apply_fixit): ...this. Convert to half-open range.
26091 Update for change to type of m_line_events.
26092 * edit-context.h (edit_context::apply_insert): Delete.
26093 (edit_context::apply_replace): Rename to...
26094 (edit_context::apply_fixit): ...this.
26096 2017-05-01 Martin Sebor <msebor@redhat.com>
26098 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26101 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26104 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26105 __get_cpuid_max returns 0.
26106 (__get_cpuid_count): Ditto.
26108 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26110 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26111 replacement expression is another instance of one of its arguments.
26113 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26116 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26117 check for stack push/pop autoinc.
26118 * config/i386/i386.c (ix86_agi_dependent): Return false
26119 if the only reason why modified_in_p returned true is that
26120 addr is SP based and set_insn is a push or pop.
26122 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26124 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26127 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26130 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26131 (account_size_time): Use two predicates - exec_pred and
26133 (evaluate_conditions_for_known_args): Compute both clause and
26135 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26136 (inline_summary_t::duplicate): Update.
26137 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26139 (compute_inline_parameters): Likewise.
26140 (estimate_edge_size_and_time): Update caluclation of time.
26141 (estimate_node_size_and_time): Compute both time and nonspecialized
26143 (estimate_ipcp_clone_size_and_time): Update.
26144 (inline_merge_summary): Update.
26145 (do_estimate_edge_time): Update.
26146 (do_estimate_edge_size): Update.
26147 (do_estimate_edge_hints): Update.
26148 (inline_read_section, inline_write_summary): Stream both new predicates.
26149 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26151 (compute_inlined_call_time): Cleanup.
26152 (big_speedup_p): Update.
26153 (edge_badness): Update.
26154 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26155 (size_time_entry): Replace predicate by exec_predicate and
26156 nonconst_predicate.
26157 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26158 (estimate_edge_time): Return also nonspec_time.
26159 (reset_edge_growth_cache): Update.
26161 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26163 PR rtl-optimization/80491
26164 * ifcvt.c (noce_process_if_block): When looking for x setter
26165 with missing else_bb, don't check only the insn right before
26166 cond_earliest, but look for the last insn that x is modified in
26167 within the same bb.
26169 PR rtl-optimization/80491
26170 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26172 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26174 PR tree-optimization/80487
26175 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26177 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26179 PR tree-optimization/79697
26180 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26181 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26182 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26184 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26185 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26187 2017-04-28 Martin Sebor <msebor@redhat.com>
26189 PR tree-optimization/80523
26190 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26191 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26193 (maybe_warn, format_directive, parse_directive): Use new functions.
26194 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26196 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26198 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26200 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26202 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26203 target_header_dir): Set correctly.
26204 * configure: Regenerated.
26205 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26206 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26207 instead of SYSTEM_HEADER_DIR.
26209 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26211 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26212 (estimate_local_effects): Likewise.
26213 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26214 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26215 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26216 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26217 do_estimate_edge_time, estimate_edge_time): Likewise.
26218 * ipa-inline-analysis.c (estimate_node_size_and_time,
26219 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26220 (estimate_time_after_inlining): Remove.
26222 2017-04-28 Martin Liska <mliska@suse.cz>
26224 * doc/gcov.texi: Enhance documentation of gcov.
26226 2017-04-28 Martin Liska <mliska@suse.cz>
26228 * doc/gcov.texi: Sort options in alphabetic order.
26229 * doc/gcov-dump.texi: Likewise.
26230 * doc/gcov-tool.texi: Likewise.
26231 * gcov.c (print_usage): Likewise.
26232 * gcov-dump.c (print_usage): Likewise.
26233 * gcov-tool.c (print_merge_usage_message): Likewise.
26234 (print_rewrite_usage_message): Likewise.
26235 (print_overlap_usage_message): Likewise.
26237 2017-04-28 Martin Liska <mliska@suse.cz>
26239 PR gcov-profile/53915
26240 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26242 2017-04-28 Martin Liska <mliska@suse.cz>
26244 PR gcov-profile/79891
26245 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26246 is marked by compiler as living on a line.
26247 (get_cycles_count): Remove usage of the union.
26248 (output_intermediate_file): Likewise.
26249 (find_source): Fix GNU coding style.
26250 (accumulate_line_counts): Remove old non-all block mode.
26251 (output_lines): Remove usage of the union.
26252 * profile.c (output_location): Include all BBs, even if
26253 belonging to a same line (and file) as a previous BB.
26255 2017-04-28 Martin Liska <mliska@suse.cz>
26257 * gcov.c (process_args): Handle new argument 'w'.
26258 (read_graph_file): Assign ID to BBs.
26259 (output_branch_count): Display BB # if verbose flag is set.
26260 (output_lines): Likewise for arcs.
26261 (print_usage): Add '--verbose' option help.
26262 * doc/gcov.texi: Document --verbose (-w) option.
26264 2017-04-28 Martin Liska <mliska@suse.cz>
26266 * gcov.c (struct block_location_info): New struct.
26267 (process_file): Fill up the new structure.
26268 (read_graph_file): Replace usage of encoding by the newly added
26270 (add_line_counts): Likewise.
26271 (accumulate_line_counts): Remove usage of the union.
26272 (function_info::function_info): New function.
26273 (function_info::~function_info): Likewise.
26274 (process_file): Call delete instead of release_function.
26275 (release_function): Release the function.
26276 (release_structures): Call delete instead of release_function.
26277 (solve_flow_graph): Replace usage of num_blocks.
26278 (find_exception_blocks): Likewise.
26279 (output_lines): Fix GNU coding style.
26281 2017-04-28 Martin Liska <mliska@suse.cz>
26284 * coverage.c (coverage_remove_note_file): New function.
26285 * coverage.h: Declare the function.
26286 * toplev.c (finalize): Clean if an error has been seen.
26288 2017-04-28 Martin Liska <mliska@suse.cz>
26290 PR gcov-profile/80031
26291 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26292 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26293 * gcov.c (read_graph_file): Read just number of blocks.
26294 * profile.c (branch_prob): Do not stream 0 flags per a basic
26297 2017-04-28 Martin Liska <mliska@suse.cz>
26299 * gcov-dump.c (tag_*): Add new argument to declarations.
26300 (dump_gcov_file): Likewise.
26301 (tag_blocks): Add and use new argument depth.
26302 (tag_arcs): Likewise.
26303 (tag_lines): Likewise.
26304 (tag_counters): Likewise.
26305 (tag_summary): Likewise.
26306 (dump_working_sets): Use depth to do a proper indentation.
26308 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26311 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26312 * symtab.c (symtab_node::debug_symtab): Move definition here.
26314 2017-04-28 Richard Biener <rguenther@suse.de>
26316 * lto-streamer.h (LTO_major_version): Bump to 7.
26318 2017-04-28 Richard Biener <rguenther@suse.de>
26320 * tree-vrp.c (assert_info): New struct.
26321 (add_assert_info): New helper.
26322 (register_edge_assert_for_2): Refactor to add asserts to a vector
26324 (register_edge_assert_for_1): Likewise.
26325 (register_edge_assert_for): Likewise.
26326 (finish_register_edge_assert_for): New helper actually registering
26327 asserts where live on edge.
26328 (find_conditional_asserts): Adjust.
26329 (find_switch_asserts): Likewise.
26330 (evrp_dom_walker::try_find_new_range): Generalize.
26331 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26333 2017-04-27 Marek Polacek <polacek@redhat.com>
26336 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26337 arg10 and arg11 to itype.
26339 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26341 * doc/extend.texi (Object Size Checking): Improve grammar.
26343 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26346 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26347 that the logic for permitting reciprocal estimates matches that
26350 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26353 * tree.c (type_cache_hasher::equal): Only compare
26354 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26355 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26356 non-aggregate element types.
26357 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26358 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26360 2017-04-27 Richard Biener <rguenther@suse.de>
26362 PR middle-end/80533
26363 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26364 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26365 keeping a reference to a trailing array.
26367 2017-04-27 Richard Biener <rguenther@suse.de>
26369 PR middle-end/80539
26370 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26371 being in loop-closed SSA form conservatively.
26372 (chrec_fold_multiply_poly_poly): Likewise.
26374 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26376 PR middle-end/79665
26377 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26378 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26380 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26383 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
26384 (aarch64_function_arg_alignment): Return unsigned int again, but still
26385 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
26386 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26387 Don't emit -Wpsabi note.
26388 (aarch64_function_arg_boundary): Likewise.
26389 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26392 2017-04-26 Nathan Sidwell <nathan@acm.org>
26394 * tree.h (crc32_unsigned_n): Declare.
26395 (crc32_unsigned, crc32_unsigned): Make inline.
26396 * tree.c (crc32_unsigned_bits): Replace with ...
26397 (crc32_unsigned_n): ... this.
26398 (crc32_unsigned, crc32_byte): Remove.
26399 (crc32_string): Remove unnecessary braces.
26401 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26403 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
26404 * ipa-inline-analysis.c (MAX_TIME): Remove.
26405 (account_size_time): Use sreal for time.
26406 (dump_inline_summary): Update.
26407 (estimate_function_body_sizes): Update.
26408 (estimate_edge_size_and_time): Update.
26409 (estimate_calls_size_and_time): Update.
26410 (estimate_node_size_and_time): Update.
26411 (inline_merge_summary): Update.
26412 (inline_update_overall_summary): Update.
26413 (estimate_time_after_inlining): Update.
26414 (inline_read_section): Update.
26415 (inline_write_summary): Update.
26416 * ipa-inline.c (compute_uninlined_call_time): Update.
26417 (compute_inlined_call_time): Update.
26418 (recursive_inlining): Update.
26419 (inline_small_functions): Update.
26420 (dump_overall_stats): Update.
26421 * ipa-inline.h: Include sreal.h.
26422 (size_time_entry): Turn time to sreal.
26423 (inline_summary): Turn self_time nad time to sreal.
26425 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26427 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
26429 (sreal::stream_out, sreal::stream_in): New.
26430 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
26432 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26434 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
26437 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
26440 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
26442 (dimode_scalar_chain::compute_convert_gain): Ditto.
26443 (dimode_scalar_chain::make_vector_copies): Ditto.
26444 (dimode_scalar_chain::convert_reg): Ditto.
26445 (dimode_scalar_chain::convert_insn): Ditto.
26446 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
26447 (VI248_AVX512BW_1): New mode iterator.
26448 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
26449 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
26452 2017-04-25 Martin Sebor <msebor@redhat.com>
26454 PR tree-optimization/80497
26455 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
26456 constants are representable in HOST_WIDE_INT.
26457 (parse_directive): Ditto.
26459 2017-04-25 Martin Sebor <msebor@redhat.com>
26462 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
26463 (new_zero_array): Adjust signature.
26464 (dom_info::dom_init): Used unsigned rather that size_t.
26465 (dom_info::dom_info): Same.
26467 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26468 Jakub Jelinek <jakub@redhat.com>
26471 * config/arm/arm.c: Include gimple.h.
26472 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26473 returns negative, increment ncrn only if it returned positive.
26474 (arm_needs_doubleword_align): Return int instead of bool,
26475 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
26476 members, but if there is any such non-FIELD_DECL
26477 > PARM_BOUNDARY aligned decl, return -1 instead of false.
26478 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26479 returns negative, increment nregs only if it returned positive.
26480 (arm_setup_incoming_varargs): Likewise.
26481 (arm_function_arg_boundary): Emit -Wpsabi note if
26482 arm_needs_doubleword_align returns negative, return
26483 DOUBLEWORD_ALIGNMENT only if it returned positive.
26485 2017-04-25 Marek Polacek <polacek@redhat.com>
26488 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26489 first argument to type.
26491 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
26494 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
26495 type checks to test for compatibility instead of equality.
26497 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26498 Jakub Jelinek <jakub@redhat.com>
26501 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
26503 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
26504 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
26505 the alignment computation, but return their maximum in warn_alignment.
26506 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26507 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
26509 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
26510 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26513 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26515 * config/arc/simdext.md (dmpyh): Fix typo.
26517 2017-04-25 Richard Biener <rguenther@suse.de>
26519 PR tree-optimization/80492
26520 * alias.c (compare_base_decls): Handle registers with asm
26521 specification conservatively.
26522 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26523 compare_base_decls returning dont-know properly.
26525 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26527 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
26528 (legitimate_offset_address_p): New function.
26529 (arc_legitimate_address_p): Use above function.
26531 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26533 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
26535 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26537 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
26538 ACCH registers whenever they are available.
26540 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26542 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
26543 double regs fix when not used.
26545 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26547 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
26549 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
26550 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
26552 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26554 * config/arc/arc.c (arc_output_addsi): Check for h-register class
26555 when emitting short ADD instructions.
26557 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26559 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
26561 (cmpsi_cc_c_insn): Likewise.
26562 (cbranchsi4_scratch): Compute proper instruction length using
26563 compact_hreg_operand.
26564 * config/arc/predicates.md (compact_hreg_operand): New predicate.
26566 2017-04-25 Richard Biener <rguenther@suse.de>
26568 PR middle-end/80509
26569 * passes.c (pass_manager::pass_manager): Initialize
26570 m_name_to_pass_map.
26572 2017-04-25 Richard Biener <rguenther@suse.de>
26574 PR tree-optimization/79201
26575 * tree-ssa-sink.c (statement_sink_location): Handle calls.
26577 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26580 * config/s390/vector.md: Split MEM->GPR vector moves for
26581 non-s_operand addresses.
26583 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26586 * config/s390/predicates.md (reload_const_wide_int_operand): New
26588 * config/s390/s390.md ("movti"): Remove d/P alternative.
26589 ("movti_bigconst"): New pattern definition.
26591 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26594 * s390-protos.h (s390_expand_cs_hqi): Removed.
26595 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
26596 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
26597 modes as well as CCZ1mode and CCZmode.
26598 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
26599 signature of s390_emit_compare_and_swap.
26600 (s390_expand_cs_hqi): Likewise, make static.
26601 (s390_expand_cs_tdsi): Generate an explicit compare before trying
26602 compare-and-swap, in some cases.
26603 (s390_expand_cs): Wrapper function.
26604 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
26606 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
26607 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
26608 patterns for small and large integers. Forbid symref memory operands.
26609 Move expander to s390.c. Require cc register.
26610 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
26611 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
26612 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
26613 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
26614 symref memory operands. Remove CC mode and call s390_match_ccmode
26616 ("atomic_exchange<mode>"): Allow and implement all integer modes.
26618 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26620 * config/s390/s390.md (define_peephole2): New peephole to help
26621 combining the load-and-test pattern with volatile memory.
26623 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26625 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
26626 with CCZmode for TARGET_Z196.
26628 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26630 PR rtl-optimization/80501
26631 * combine.c (make_compound_operation_int): Set subreg_code to SET
26632 even for AND with mask of the sign bit of mode.
26634 PR rtl-optimization/80500
26635 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
26636 sum's initial value.
26638 2017-04-25 Julian Brown <julian@codesourcery.com>
26639 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26641 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
26643 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
26645 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
26647 2017-04-25 Julian Brown <julian@codesourcery.com>
26648 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26650 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
26651 (thunderx2t99_sha): New Reservation.
26653 2017-04-25 Julian Brown <julian@codesourcery.com>
26654 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26656 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
26657 type for 1-element load.
26659 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
26661 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
26663 2017-04-24 Martin Jambor <mjambor@suse.cz>
26665 PR tree-optimization/80293
26666 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
26667 char arrays not totally scalarizable if it is false.
26668 (analyze_all_variable_accesses): Pass correct value in the new
26669 parameter. Add a statistics counter.
26671 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
26673 PR middle-end/79931
26674 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
26676 2017-04-24 Richard Biener <rguenther@suse.de>
26678 PR tree-optimization/80494
26679 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
26680 out for complex types.
26682 2017-04-24 Richard Biener <rguenther@suse.de>
26684 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
26685 * tree-ssa-sccvn.c (print_scc): Print SCC size.
26686 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
26687 (DFS): Adjust and never fail.
26688 (sccvn_dom_walker::fail): Remove.
26689 (sccvn_dom_walker::before_dom_children): Adjust.
26690 (run_scc_vn): Likewise and never fail.
26691 * tree-ssa-pre.c (pass_pre::execute): Adjust.
26692 (pass_fre::execute): Likewise.
26694 2017-04-24 Richard Biener <rguenther@suse.de>
26696 PR tree-optimization/79725
26697 * tree-ssa-sink.c (statement_sink_location): Return whether
26698 failure reason was zero uses. Move that check later.
26699 (sink_code_in_bb): Deal with zero uses by removing the stmt
26702 2017-04-24 Richard Biener <rguenther@suse.de>
26705 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
26706 pointer-based references.
26708 2017-04-24 Richard Biener <rguenther@suse.de>
26711 * pass_manager.h (pass_manager::operator new): Remove.
26712 (pass_manager::operator delete): Likewise.
26713 * passes.c (pass_manager::operator new): Remove.
26714 (pass_manager::operator delete): Likewise.
26715 (pass_manager::pass_manager): Zero individual pass members.
26717 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
26720 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
26721 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
26722 Check "XEXP (src, 1)" operand here.
26723 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
26724 Check "XEXP (src, 1)" operand here.
26725 (dimode_scalar_chain::make_vector_copies): Detect count register
26726 of a shift instruction. Zero extend count register from QImode
26727 to DImode to satisfy vector shift pattern count operand predicate.
26728 Substitute vector shift count operand with a DImode copy.
26729 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
26732 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
26734 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
26735 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26736 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
26737 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
26738 (UNSPEC_NOREX_MEM): Remove definition.
26740 2017-04-21 Richard Biener <rguenther@suse.de>
26742 PR tree-optimization/79547
26743 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
26744 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
26745 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
26746 without any constraints.
26748 2017-04-21 Richard Biener <rguenther@suse.de>
26750 PR tree-optimization/78847
26751 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
26753 2017-04-21 Richard Biener <rguenther@suse.de>
26755 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
26756 (build_distinct_type_copy): Likewise.
26757 (build_variant_type_copy): Likewise.
26758 * tree.c (build_qualified_type): Pass down mem-stat info.
26759 (build_distinct_type_copy): Likewise.
26760 (build_variant_type_copy): Likewise.
26762 2017-04-21 Richard Biener <rguenther@suse.de>
26764 PR tree-optimization/80237
26765 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
26767 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
26768 for a simplified result.
26770 2016-04-21 Richard Biener <rguenther@suse.de>
26772 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
26773 sth as strict as a simple_iv but a chrec without symbols and an
26774 operand defined in the loop we are peeling (and not some subloop).
26775 (propagate_constants_for_unrolling): Propagate all constants.
26777 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26780 * config/i386/i386.c (print_reg): Remove assert for disalowed
26781 regno values, call output_operand_lossage instead.
26783 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
26786 * config/i386/constraints.md (Yc): New register constraint.
26787 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
26788 Use Yc constraint for alternative 2 of operand 0. Remove
26789 preferred_for_speed attribute.
26791 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26793 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
26794 lastprivate clauses in SIMT case.
26796 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
26798 * doc/invoke.texi (-Wextra-semi): Document new warning option.
26800 2017-04-20 Richard Biener <rguenther@suse.de>
26802 PR tree-optimization/57796
26803 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
26804 as N scalar stores.
26805 (vect_model_load_cost): Cost gathers as N scalar loads.
26807 2017-04-20 Richard Biener <rguenther@suse.de>
26809 * ggc-page.c (ggc_allocated_p): Rename to ...
26810 (safe_lookup_page_table_entry): ... this and return the lookup
26812 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
26814 2017-04-20 Richard Biener <rguenther@suse.de>
26816 PR tree-optimization/80453
26817 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
26818 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
26819 from the conditions.
26820 (vn_phi_eq): Pass them down.
26821 (vn_phi_lookup): Record them.
26822 (vn_phi_insert): Likewise.
26824 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
26826 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
26827 uninitialized variable warning to avoid buffer overrun.
26829 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26832 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26833 is suppressed for '{ 0 }' in C.
26835 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26837 * BASE-VER: Set to 8.0.0.
26839 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26841 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26842 priority .init_array and .fini_array section with SECTION_NOTYPE
26845 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26847 PR middle-end/80423
26848 * tree.h (build_array_type): Add typeless_storage default argument.
26849 * tree.c (type_cache_hasher::equal): Also compare
26850 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26851 (build_array_type): Add typeless_storage argument, set
26852 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26854 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26855 (build_array_type): Likewise. Add typeless_storage argument.
26857 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26858 Jakub Jelinek <jakub@redhat.com>
26860 PR tree-optimization/80426
26861 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26862 operation on symbolic operands, also compute the overflow for the
26863 invariant part when the operation degenerates into a negation.
26865 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26868 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26869 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26872 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26874 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26877 * config/avr/avr.c (tree.h): Include it.
26878 (cgraph.h): Include it.
26879 (avr_encode_section_info): Don't warn for uninitialized progmem
26880 variable if it's just an alias.
26882 2017-04-19 Richard Biener <rguenther@suse.de>
26885 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26886 when needed by AutoPGO.
26888 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
26891 * doc/lto.texi: Remove an extra 'that'.
26893 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
26895 PR rtl-optimization/80429
26896 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
26897 are only used in debug insns.
26899 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26900 Vladimir Makarov <vmakarov@redhat.com>
26902 * config/sparc/predicates.md (input_operand): Add comment. Return
26903 true for any memory operand when LRA is in progress.
26904 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26906 2017-04-18 Jeff Law <law@redhat.com>
26909 * mips.md ({return,simple_return}_internal): Do not overwrite
26912 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26914 PR tree-optimization/80443
26915 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26916 instead of adding 1, subtract -1 and similarly instead of subtracting
26919 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26921 PR rtl-optimization/80357
26922 * haifa-sched.c (tmp_bitmap): New variable.
26923 (model_recompute): Handle duplicate use records.
26924 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26925 (free_global_sched_pressure_data): Free it.
26927 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26930 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26931 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26932 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26933 instead of SYSTEM_HEADER_DIR.
26935 2017-04-18 Jeff Law <law@redhat.com>
26937 PR middle-end/80422
26938 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
26939 predecessors after walking up the insn chain.
26941 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26944 * dwarf2out.c (modified_type_die): Try harder not to emit internal
26945 sizetype type into debug info.
26947 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26950 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
26951 unneeded test for TARGET_UPPER_REGS_SF.
26952 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
26954 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26957 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
26958 instead of gsi_after_labels.
26960 2017-04-18 Jeff Law <law@redhat.com>
26962 * regcprop.c (maybe_mode_change): Avoid creating copies of the
26966 2017-04-13 Jeff Law <law@redhat.com>
26967 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26968 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26970 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
26973 * config/avr/avr.c (intl.h): Include it.
26974 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
26976 2017-04-18 Martin Liska <mliska@suse.cz>
26978 PR gcov-profile/78783
26979 * gcov-tool.c (gcov_output_files): Validate that destination
26980 file is either removed by the tool or by a user.
26982 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26983 Guy Benyei <guybe@mellanox.com>
26985 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26986 block, and do not negate it, the stored id is already negative.
26988 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26990 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
26992 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26995 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
26996 masks of options that should be turned off if the VSX vector
26997 options are turned off.
26998 (OTHER_P8_VECTOR_MASKS): Likewise.
26999 (OTHER_VSX_VECTOR_MASKS): Likewise.
27000 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27001 rs6000_disable_incompatible_switches to validate no type switches
27003 (rs6000_incompatible_switch): New function to disallow turning on
27004 other vector options if -mno-vsx, -mno-power8-vector, or
27005 -mno-power9-vector are specified.
27007 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27009 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27011 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27013 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27014 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27015 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27016 (ARG_POINTER_CFA_OFFSET): Likewise.
27018 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27020 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27021 conditions to take advantage of various optimizations.
27023 2017-04-13 Jeff Law <law@redhat.com>
27025 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27026 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27027 (zero_extendsidi2_dext): Likewise.
27029 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27032 * fold-const.c (fold_ternary_loc): Revert
27033 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27036 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27038 PR rtl-optimization/80343
27039 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27040 new scratch pseudo.
27042 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27045 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27046 to ubsan_encode_value.
27048 2017-04-13 Jeff Law <law@redhat.com>
27050 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27051 appearing in DEBUG_INSNs.
27053 2017-04-13 Martin Liska <mliska@suse.cz>
27055 PR gcov-profile/80413
27056 * gcov-io.c (gcov_write_string): Copy to buffer just when
27057 allocated size is greater than zero.
27059 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27062 * dwarf2out.c (decls_for_scope): Ignore declarations of
27063 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27065 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27068 * ipa-visibility.c (non_local_p): Fix typos.
27069 (localize_node): When localizing symbol in same comdat group,
27070 dissolve the group only when we know external symbols are going
27072 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27074 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27076 PR tree-optimization/79390
27077 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27078 order does not result in usable sequence, retry with reversed operand
27084 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27085 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27086 op0 instead of fold_convert_loc (loc, type, arg0).
27088 2017-04-12 Jeff Law <law@redhat.com>
27090 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27091 has a delay slot in the generated code.
27093 * config/cris/cris.md (cris_preferred_reload_class): Return
27094 GENNONACR_REGS rather than GENERAL_REGS.
27096 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27099 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27100 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27101 signedness of the result type.
27103 2017-04-12 Richard Biener <rguenther@suse.de>
27104 Jeff Law <law@redhat.com>
27106 PR tree-optimization/80359
27107 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27108 trim stores to TARGET_MEM_REFs.
27110 2017-04-12 Richard Biener <rguenther@suse.de>
27112 PR tree-optimization/79390
27113 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27114 threading case even more.
27116 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27119 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27120 for quad_address_p for TImode, instead of just not indexed_address.
27122 2017-04-12 Richard Biener <rguenther@suse.de>
27123 Bernd Edlinger <bernd.edlinger@hotmail.de>
27125 PR middle-end/79671
27126 * alias.c (component_uses_parent_alias_set_from): Handle
27127 TYPE_TYPELESS_STORAGE.
27128 (get_alias_set): Likewise.
27129 * tree-core.h (tree_type_common): Add typeless_storage flag.
27130 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27131 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27132 for types containing members with TYPE_TYPELESS_STORAGE.
27133 (place_field): Likewise.
27134 (layout_type): Likewise for ARRAY_TYPE.
27135 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27136 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27137 TYPE_TYPELESS_STORAGE.
27138 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27140 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27143 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27144 first argument to type.
27146 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27150 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27151 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27152 (rs6000_expand_binop_builtin): Likewise.
27153 (rs6000_expand_ternop_builtin): Likewise; also add missing
27154 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27155 vshasigma built-ins.
27156 * doc/extend.texi: Document that vec_xxpermdi's third argument
27157 must be a constant.
27159 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27161 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27162 Use shift_const cost parameter when calculating gain of STV shifts.
27164 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27166 PR rtl-optimization/70478
27167 * lra-constraints.c (process_alt_operands): Check memory for
27168 disfavoring memory insn operand.
27170 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27172 PR middle-end/80100
27173 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27174 left shift in unsigned HOST_WIDE_INT type.
27176 PR rtl-optimization/80385
27177 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27178 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27181 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27182 if they have any depend clauses.
27184 2017-04-11 Martin Liska <mliska@suse.cz>
27187 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27188 * ipa-split.c (split_function): Create a local comdat symbol
27189 if caller is in a comdat group.
27191 2017-04-11 Martin Liska <mliska@suse.cz>
27194 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27197 2017-04-11 Martin Sebor <msebor@redhat.com>
27199 PR middle-end/80364
27200 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27201 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27203 (directive::set_width, directive::set_precision, format_character):
27205 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27208 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27211 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27212 conflict, set target->arch_name instead of target->cpu_name.
27214 2017-04-11 Richard Biener <rguenther@suse.de>
27216 PR tree-optimization/80374
27217 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27218 build_zero_cst, remove fold_convertible_p check again.
27220 2017-04-11 Martin Liska <mliska@suse.cz>
27223 * ubsan.c (instrument_object_size): Do not instrument register
27226 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27229 * config/i386/i386-builtin-types.def
27230 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27231 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27232 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27233 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27234 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27235 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27236 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27237 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27238 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27239 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27240 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27241 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27242 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27243 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27244 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27245 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27246 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27247 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27248 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27249 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27250 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27251 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27252 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27253 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27254 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27255 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27256 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27257 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27258 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27259 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27260 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27261 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27262 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27263 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27264 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27265 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27266 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27267 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27268 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27269 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27270 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27271 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27272 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27273 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27274 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27275 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27277 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27278 flag to second_arg_count, handle 4 argument function type _COUNT
27279 aliases, handle second_arg_count on second argument rather than last.
27281 2017-04-10 Jeff Law <law@redhat.com>
27283 PR tree-optimization/80374
27284 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27285 record anything if we can not convert integer_zero_node to the
27288 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27291 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27292 Enhance special handling given to the TARGET_P9_MINMAX option in
27293 relation to certain other options.
27295 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27297 PR tree-optimization/80153
27298 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27299 remove POINTER_PLUS_EXPR's base part directly, rather than through
27302 2017-04-10 Richard Biener <rguenther@suse.de>
27303 Bin Cheng <bin.cheng@arm.com>
27305 PR tree-optimization/80153
27306 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27307 the first element of pointer type aff_tree. Build result expr in
27309 (add_elt_to_tree): Convert to type unconditionally. Remove other
27310 fold_convert calls.
27311 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27312 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27314 2017-04-10 Richard Biener <rguenther@suse.de>
27316 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27319 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27321 PR rtl-optimization/70478
27322 * lra-constraints.c (curr_small_class_check): New.
27323 (update_and_check_small_class_inputs): New.
27324 (process_alt_operands): Update curr_small_class_check. Disfavor
27325 alternative insn memory operands. Check available regs for small
27328 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27331 * config/mips/mips.opt (-mvirt): Update description.
27332 * doc/invoke.texi (-mvirt): Likewise.
27334 2017-04-10 Richard Biener <rguenther@suse.de>
27336 PR middle-end/80362
27337 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27338 looking for NEGATE_EXPR in -A / -B to A / B folding.
27340 2017-04-10 Martin Liska <mliska@suse.cz>
27342 PR gcov-profile/80224
27343 * gcov.c (print_usage): Fix usage string.
27344 (get_gcov_intermediate_filename): Remove.
27345 (output_gcov_file): Use both for normal and intermediate format.
27346 (generate_results): Do not initialize special file for
27347 intermediate format.
27349 2017-04-10 Richard Biener <rguenther@suse.de>
27351 PR tree-optimization/80304
27352 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27355 2017-04-10 Nathan Sidwell <nathan@acm.org>
27358 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27359 (rs6000_init_builtins): Use it.
27361 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27363 * config/arm/arm.md (<mrc>): Add mode to SET source.
27364 (<mrrc>): Likewise.
27366 2017-04-10 Richard Biener <rguenther@suse.de>
27368 PR middle-end/80344
27369 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27371 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27374 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27375 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27376 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27377 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27378 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27379 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27380 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27381 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27382 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27383 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
27384 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
27385 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
27386 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
27387 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
27388 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
27389 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
27390 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
27391 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
27392 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
27393 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
27394 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
27395 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
27396 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
27398 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
27400 PR rtl-optimization/70478
27401 * lra-constraints.c: Reverse the last patch.
27403 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
27405 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
27406 Add comment for WCHAR_T.
27408 2017-04-08 Martin Liska <mliska@suse.cz>
27411 2017-04-07 Martin Liska <mliska@suse.cz>
27414 * ipa-split.c (split_function): Add function part to a same comdat
27417 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27420 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
27422 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
27424 * rs6000/rs6000.c (vec_load_pendulum): Rename...
27425 (vec_pairing): ...to this.
27426 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
27427 (rs6000_sched_init): Adjust for name change.
27428 (struct rs6000_sched_context): Likewise.
27429 (rs6000_init_sched_context): Likewise.
27430 (rs6000_set_sched_context): Likewise.
27432 2017-04-07 Jakub Jelinek <jakub@redhat.com>
27438 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
27440 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
27441 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
27442 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
27444 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
27446 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
27448 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27450 PR rtl-optimization/70703
27451 * ira-color.c (update_conflict_hard_regno_costs): Use
27452 int64_t instead of HOST_WIDE_INT.
27454 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27456 PR rtl-optimization/70478
27457 * lra-constraints.c (process_alt_operands): Disfavor alternative
27458 insn memory operands.
27460 2017-04-07 Jeff Law <law@redhat.com>
27462 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
27463 CALL and NOTE_INSN_CALL_ARG_LOCATION.
27465 2017-04-07 Martin Liska <mliska@suse.cz>
27468 * config/aarch64/aarch64.c (aarch64_process_target_attr):
27469 Show error message instead of an ICE.
27471 2017-04-07 Martin Liska <mliska@suse.cz>
27474 * ipa-split.c (split_function): Add function part to a same comdat
27477 2017-04-07 Richard Biener <rguenther@suse.de>
27479 PR middle-end/80341
27480 * tree.c (get_unwidened): Also handle ! for_type case for
27482 * convert.c (do_narrow): Split out from ...
27483 (convert_to_integer_1): ... here. Do not pass final truncation
27484 type to get_unwidened for TRUNC_DIV_EXPR.
27486 2017-04-07 Richard Biener <rguenther@suse.de>
27488 * tree-affine.c (wide_int_ext_for_comb): Take type rather
27490 (aff_combination_const): Adjust.
27491 (aff_combination_scale): Likewise.
27492 (aff_combination_add_elt): Likewise.
27493 (aff_combination_add_cst): Likewise.
27494 (aff_combination_convert): Likewise.
27495 (add_elt_to_tree): Likewise. Remove unused argument.
27496 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
27498 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
27500 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
27502 * config/arm/arm.c (arm_default_short_enums): Use
27503 ARM_DEFAULT_SHORT_ENUMS.
27504 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
27506 2017-04-06 Jakub Jelinek <jakub@redhat.com>
27509 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
27510 members with redundant out-of-class redeclaration.
27512 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27515 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
27516 * config/i386/i386.md (*zero_extendsidi2):
27517 Add (?*x,*x) and (?*v,*v) alternatives.
27519 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27522 * config/i386/i386.c (ix86_expand_builtin)
27523 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
27524 mode from insn data. Convert operands to insn operand mode.
27525 Copy operands that don't satisfy insn predicate to a register.
27527 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
27529 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
27532 2017-04-06 Richard Biener <rguenther@suse.de>
27534 PR tree-optimization/80334
27535 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
27536 preserve alignment of accesses.
27538 2017-04-06 Richard Biener <rguenther@suse.de>
27540 PR tree-optimization/80262
27541 * tree-sra.c (build_ref_for_offset): Preserve address-space
27543 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
27544 Drop useless address-space information on MEM_REF offsets.
27546 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
27548 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
27550 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27552 PR rtl-optimization/70703
27553 * ira-color.c (update_conflict_hard_regno_costs): Use
27554 HOST_WIDE_INT instead of long.
27556 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
27559 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
27560 not defined for x86_64 target. Add -mmmx target option when __SSE2__
27562 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
27563 for x86_64 target. Handle -m3dnowa option.
27565 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27567 PR rtl-optimization/70703
27568 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
27569 (update_conflict_hard_regno_costs): Use long instead of unsigned
27570 arithmetic for cost calculation.
27572 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27573 Bernd Edlinger <bernd.edlinger@hotmail.de>
27576 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
27579 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
27582 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
27583 ptr_mode with Pmode throughout.
27584 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
27585 into probe_stack_range and use DImode.
27587 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
27590 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
27591 call_eh_return is true.
27593 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27595 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
27596 Initialize last_match_fntype_index.
27598 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27601 * tree-nvr.c: Include internal-fn.h.
27602 (pass_return_slot::execute): Ignore internal calls without
27605 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27606 Richard Biener <rguenther@suse.de>
27609 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
27610 captures used multiple times, except for the last use.
27611 * generic-match-head.c: Include gimplify.h.
27613 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27615 PR tree-optimization/79390
27616 * target.h (struct noce_if_info): Declare.
27617 * targhooks.h (default_noce_conversion_profitable_p): Declare.
27618 * target.def (noce_conversion_profitable_p): New target hook.
27619 * ifcvt.h (struct noce_if_info): New type, moved from ...
27620 * ifcvt.c (struct noce_if_info): ... here.
27621 (noce_conversion_profitable_p): Renamed to ...
27622 (default_noce_conversion_profitable_p): ... this. No longer
27624 (noce_try_store_flag_constants, noce_try_addcc,
27625 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
27626 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
27627 instead of noce_conversion_profitable_p.
27628 * config/i386/i386.c: Include ifcvt.h.
27629 (ix86_option_override_internal): Don't override
27630 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
27631 (ix86_noce_conversion_profitable_p): New function.
27632 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
27633 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
27634 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
27635 * doc/tm.texi: Regenerated.
27637 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27639 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
27642 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27645 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
27646 instructions for small multiply cores.
27648 2017-04-04 Jeff Law <law@redhat.com>
27650 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
27652 (mips_expand_vec_perm_const): Initialize elements in orig_perm
27653 that are not set by the loop over the elements.
27655 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27658 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
27659 int mode, convert_modes it to mode as unsigned, otherwise use
27660 lowpart_subreg to mode rather than SImode.
27661 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
27662 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
27663 Use DImode instead of SImode for the shift count operand.
27664 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
27667 2017-04-04 Richard Biener <rguenther@suse.de>
27669 PR middle-end/80281
27670 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
27671 arithmetic done for the negate or the plus. Simplify.
27672 (A - (-B) -> A + B): Likewise.
27673 * fold-const.c (split_tree): Make sure to not negate pointers.
27675 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
27677 PR rtl-optimization/60818
27678 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
27679 a compare of comparisons with the thing compared if this results
27680 in a different machine mode.
27682 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
27684 * alias.c (base_alias_check): Fix typo in comment.
27685 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
27686 * cgraphunit.c (symbol_table::compile): Likewise.
27687 * collect2.c (maybe_run_lto_and_relink): Likewise.
27688 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
27689 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
27690 * config/avr/avr.c (avr_map_op_t): Likewise.
27691 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
27692 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
27693 * config/epiphany/epiphany.md (movcc): Likewise.
27694 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
27695 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
27697 * config/mips/mips.c (mips_save_restore_reg): Likewise.
27698 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
27699 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
27700 * config/sh/sh.c (sh_rtx_costs): Likewise.
27701 * fold-const.c (fold_truth_andor): Likewise.
27702 * genautomata.c (collapse_flag): Likewise.
27703 * gengtype.h (struct type::u::s): Likewise.
27704 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
27705 * input.c (FORMAT_AMOUNT): Likewise.
27706 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
27707 (known_aggs_to_agg_replacement_list): Likewise.
27708 * ipa-inline-analysis.c: Likewise.
27709 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
27710 * ipa-polymorphic-call.c
27711 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
27712 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
27713 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
27715 * modulo-sched.c (apply_reg_moves): Likewise.
27716 * omp-expand.c (build_omp_regions_1): Likewise.
27717 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
27718 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
27719 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
27720 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
27721 * value-prof.c: Likewise.
27722 * var-tracking.c (val_reset): Likewise.
27724 2017-04-03 Richard Biener <rguenther@suse.de>
27726 PR tree-optimization/80275
27727 * fold-const.c (split_address_to_core_and_offset): Handle
27730 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
27732 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
27733 descriptors is at least equal to that of functions.
27735 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27737 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
27739 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
27742 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
27743 (mov<IMOD4:mode>): New expander.
27744 (*mov<IMOD4:mode>_internal): New insn and split pattern.
27746 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
27748 PR rtl-optimization/79405
27749 * fwprop.c (propagations_left): New variable.
27750 (forward_propagate_into): Decrement it.
27751 (fwprop_init): Initialize it.
27752 (fw_prop): If the variable has reached zero, stop propagating.
27753 (fwprop_addr): Ditto.
27755 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27758 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
27759 a FUNCTION_DECL, pass it as decl instead of origin to
27762 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
27764 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
27767 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
27770 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
27771 TARGET_VSX_SMALL_INTEGER.
27773 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27775 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
27776 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
27778 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27780 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
27781 extraction from odd-numbered MSA register.
27783 2017-03-31 Jakub Jelinek <jakub@redhat.com>
27785 PR middle-end/80173
27786 * expmed.c (store_bit_field_1): Don't attempt to create
27787 a word subreg out of hard registers wider than word if they
27788 have HARD_REGNO_NREGS of 1 for their mode.
27790 PR middle-end/80163
27791 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
27792 conversions to integer types wider than word and pointer.
27795 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
27796 (rtx_equal_for_cselib_p): Pass 0 to it.
27797 * cselib.c (cselib_hasher::equal): Likewise.
27798 (rtx_equal_for_cselib_1): Add depth argument. If depth
27799 is 128, don't look up VALUE locs and punt. Increment
27800 depth in recursive calls when walking VALUE locs.
27802 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
27804 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
27805 (make_gcov_file_name): Use the canonical path name for generating
27807 (read_line): Fix handling of files with ascii null bytes.
27809 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
27811 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
27812 to initialise a vector register instead
27813 of using a const_int.
27815 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27817 PR translation/80189
27818 * gimplify.c (omp_default_clause): Use %qs instead of %s in
27819 diagnostic messages.
27821 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
27824 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
27825 (dfp_diex_<mode>): Update mode of operand 1.
27826 * doc/extend.texi (dxex, dxexq): Document change to return type.
27827 (diex, diexq): Document change to argument type.
27829 2017-03-30 Martin Jambor <mjambor@suse.cz>
27832 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27833 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27834 it reflects the signature changes performed at the callee side.
27835 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27836 to cgraph_build_function_type_skip_args.
27837 (build_function_decl_skip_args): Adjust call to the above function.
27839 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27842 * config/i386/sse.md
27843 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27844 register as dest whenever it is a MEM not rtx_equal_p to the
27845 corresponding dup operand, and when forcing into reg move the
27846 reg into the memory afterwards.
27847 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27848 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27849 for the force_reg mode.
27850 (avx512vl_vextractf128<mode>): Use register as dest either
27851 always when a MEM, or when it is a MEM not rtx_equal_p to the
27852 corresponding dup operand, or even not when it is a CONST_VECTOR
27853 depending on the mode and lo vs. hi.
27854 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27856 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27857 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27858 Likewise. Require that operands[2] is even.
27859 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27860 Remove extraneous parens. Require that operands[2] is a multiple
27862 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27863 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27864 disallow memory then.
27866 2017-03-30 Richard Biener <rguenther@suse.de>
27868 PR tree-optimization/77498
27869 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27870 to non-constants over backedges.
27872 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27874 PR rtl-optimization/80233
27875 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27876 as last_combined_insn. Do not test for BARRIER_P separately.
27878 2017-03-29 Andreas Schwab <schwab@suse.de>
27881 * calls.c (prepare_call_address): Convert funexp to Pmode before
27882 copying to temp reg.
27884 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27886 PR tree-optimization/80158
27887 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
27888 Handle possible future case of more than one alternate
27890 (replace_rhs_if_not_dup): Likewise.
27891 (replace_one_candidate): Likewise.
27893 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
27895 PR rtl-optimization/80193
27896 * ira.c (ira): Do not check allocation for LRA.
27898 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
27900 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27901 (nvptx_output_simt_exit): Declare.
27902 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27903 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27904 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27905 (nvptx_declare_function_name): ...here. Emit declaration of local
27906 memory space buffer for omp_simt_enter insn.
27907 (nvptx_output_unisimt_switch): New.
27908 (nvptx_output_softstack_switch): New.
27909 (nvptx_output_simt_enter): New.
27910 (nvptx_output_simt_exit): New.
27911 * config/nvptx/nvptx.h (struct machine_function): New fields
27912 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27913 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27914 (UNSPECV_SIMT_EXIT): Ditto.
27915 (omp_simt_enter_insn): New insn.
27916 (omp_simt_enter): New expansion.
27917 (omp_simt_exit): New insn.
27918 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27920 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27921 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27922 (expand_GOMP_SIMT_EXIT): New.
27923 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27924 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27925 (GOMP_SIMT_EXIT): Ditto.
27926 * target-insns.def (omp_simt_enter): New insn.
27927 (omp_simt_exit): Ditto.
27928 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27930 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27931 (lower_rec_input_clauses): Likewise.
27932 (lower_lastprivate_clauses): Handle SIMT privatization.
27934 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
27935 (ompdevlow_adjust_simt_enter): New.
27936 (find_simtpriv_var_op): New.
27937 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
27938 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
27940 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
27941 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
27942 (copy_decl_for_dup_finish): Ditto.
27944 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
27946 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
27949 * config/i386/i386.c (ix86_option_override_internal): Always
27950 allow -mpreferred-stack-boundary=3 for 64-bit targets.
27952 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27954 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
27956 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27958 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
27959 mark new edge's irreducible flag accordign to it.
27960 (vect_do_peeling): Check loop preheader edge's irreducible flag
27961 and pass it to function slpeel_add_loop_guard.
27963 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
27965 PR tree-optimization/80218
27966 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27967 Update block frequencies and counts.
27969 2017-03-28 Richard Biener <rguenther@suse.de>
27971 PR tree-optimization/78644
27972 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
27973 of a simplification result we may not use it at all.
27975 2017-03-28 Richard Biener <rguenther@suse.de>
27978 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
27979 without arguments, generate default definition of a SSA name.
27981 2017-03-28 Richard Biener <rguenther@suse.de>
27983 PR middle-end/80222
27984 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27985 TYPE_REF_CAN_ALIAS_ALL references.
27986 * fold-const.c (fold_indirect_ref_1): Likewise.
27988 2017-03-28 Martin Liska <mliska@suse.cz>
27991 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
27992 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
27994 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27995 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
27997 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
27998 (EXTRA_SPECS): Define.
27999 (SUBTARGET_EXTRA_SPECS): Likewise.
28000 (SUBTARGET_CPP_SPEC): Likewise.
28001 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28002 SUBTARGET_EXTRA_SPECS.
28003 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28005 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28007 * config/arc/simdext.md (vst64_insn): Update pattern.
28008 (vld32wh_insn): Likewise.
28009 (vld32wl_insn): Likewise.
28010 (vld64_insn): Likewise.
28011 (vld32_insn): Likewise.
28013 2017-03-28 Marek Polacek <polacek@redhat.com>
28016 * fold-const.c (fold_comparison): Use protected_set_expr_location
28017 instead of SET_EXPR_LOCATION.
28019 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28021 * tree.c (add_expr): Avoid name lookup warning.
28023 2017-03-27 Jeff Law <law@redhat.com>
28025 PR tree-optimization/80216
28026 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28027 function name. Limit recursion depth.
28028 (record_temporary_equivalences): Corresponding changes.
28030 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28032 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28035 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28038 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28040 * cfgcleanup.c (reg_note_cfa_p): New array.
28041 (insns_have_identical_cfa_notes): New function.
28042 (old_insns_match_p): Don't cross-jump in between /f
28043 and non-/f instructions. If both i1 and i2 are frame related,
28044 verify all CFA notes, their order and content.
28046 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28049 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28050 HImode and SImode with zero extend to DImode to one insn.
28051 (bswap<mode>2_extenddi): Likewise.
28052 (bswapsi2_extenddi): Likewise.
28053 (bswaphi2_extendsi): Likewise.
28054 (bswaphi2): Combine bswap HImode and SImode into one insn.
28055 Separate memory insns from swapping register.
28056 (bswapsi2): Likewise.
28057 (bswap<mode>2): Likewise.
28058 (bswaphi2_internal): Delete, no longer used.
28059 (bswapsi2_internal): Likewise.
28060 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28061 store, and gpr<-gpr swap insns.
28062 (bswap<mode>2_store): Likewise.
28063 (bswaphi2_reg): Register only splitter, combine with the splitter.
28064 (bswaphi2 splitter): Likewise.
28065 (bswapsi2_reg): Likewise.
28066 (bswapsi2 splitter): Likewise.
28067 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28068 the insns into load, store, and register/register insns.
28069 (bswapdi2_ldbrx): Likewise.
28070 (bswapdi2_load): Likewise.
28071 (bswapdi2_store): Likewise.
28072 (bswapdi2_reg): Likewise.
28074 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28076 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28077 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28079 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28082 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28084 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28085 special handling for target option conflicts between dform
28086 options (-mpower9-dform, -mpower9-dform-vector,
28087 -mpower9-dform-scalar) and -mno-direct-move.
28089 2017-03-27 Richard Biener <rguenther@suse.de>
28091 PR tree-optimization/80181
28092 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28094 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28096 * config/arc/predicates.md (move_double_src_operand): Replace the
28097 call to move_double_src_operand with a call to address_operand.
28099 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28101 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28102 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28103 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28105 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28107 * config/arc/predicates.md (long_immediate_loadstore_operand):
28108 Consider scaled addresses cases.
28110 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28112 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28113 restored when in interrupt.
28114 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28115 doesn't have delay slot.
28117 2017-03-27 Richard Biener <rguenther@suse.de>
28120 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28121 inlined thunk clones.
28123 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28126 * asan.c (instrument_derefs): Copy over last operand from
28127 original COMPONENT_REF to the new COMPONENT_REF with
28128 DECL_BIT_FIELD_REPRESENTATIVE.
28129 * ubsan.c (instrument_object_size): Likewise.
28131 2017-03-27 Richard Biener <rguenther@suse.de>
28133 PR tree-optimization/80170
28134 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28135 sure DR/SCEV didnt fold in constants we do not see when looking
28136 at the reference base alignment.
28138 2017-03-27 Richard Biener <rguenther@suse.de>
28140 PR middle-end/80171
28141 * gimple-fold.c (fold_ctor_reference): Properly guard against
28142 NULL return value from canonicalize_constructor_val.
28144 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28147 * config/i386/i386.c (ix86_expand_builtin)
28148 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28149 flags reg setting and flags reg using instructions.
28150 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28151 clobbering instructions to zero extend op2.
28153 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28155 * doc/install.texi (Configuration) <--with-aix-soname>:
28156 Update link to AIX ld.
28158 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28160 PR rtl-optimization/80160
28161 PR rtl-optimization/80159
28162 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28163 reg_alternate_class into account.
28165 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28168 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28169 to consider in curr_insn_transform.
28171 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28173 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28174 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28175 and emit_mode_inner.
28177 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28179 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28180 argument to the overloaded builtin variants. Use the new flag to
28181 deprecate certain builtin variants.
28182 * config/s390/s390-builtin-types.def: Add new builtin types.
28183 * config/s390/s390-builtins.h: Support new flags field for
28184 overloaded builtins.
28185 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28186 (s390_macro_to_expand): Enable vector float data type.
28187 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28188 builtins by incrementing the __VEC__ version number.
28189 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28191 (s390_resolve_overloaded_builtin): Emit error messages depending
28192 on the builtin flags.
28193 * config/s390/s390.c (s390_expand_builtin): Support additional
28194 flags argument. Change error message to match the messages
28195 emitted in s390-c.c.
28196 * config/s390/s390.md: New UNSPEC_* constants.
28197 (op_type): Add new instruction types.
28198 * config/s390/vecintrin.h: Add new builtins and test data class
28200 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28201 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28202 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28203 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28204 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28205 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28207 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28208 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28209 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28210 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28212 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28213 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28214 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28215 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28217 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28218 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28219 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28220 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28221 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28222 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28223 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28225 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28226 ("vec_scatter_element<V_HW_4:mode>_DI")
28227 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28228 ("vec_fpint<mode>", "vflls")
28229 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28230 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28231 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28232 ("*vec_cmphe<mode>_cc"): ... these.
28234 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28235 mode constant instead of magic value.
28237 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28239 * config/s390/s390.c (s390_expand_vec_compare): Support other
28240 vector floating point modes than just V2DF.
28241 (s390_expand_vcond): Likewise.
28242 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28243 (s390_cannot_change_mode_class): Prevent mode changes between TF
28244 and V1TF in vector registers.
28245 * config/s390/s390.md (DF, SF): New mode attributes.
28246 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28247 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28248 SFmode support for VRs.
28249 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28251 (VFT, VF_HW): New mode iterators.
28252 (vw, sdx): New mode attributes.
28253 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28254 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28255 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28256 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28257 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28258 also the new vector floating point modes. Renaming to ...
28260 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28261 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28262 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28263 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28264 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28265 ("vec_unordered<mode>"): ... these.
28267 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28268 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28269 ("*vec_extendv2df"): New insn definitions.
28271 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28273 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28274 ("mulditi3_2", "*muldi3_sign"): New patterns.
28275 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28276 rename the pattern definition.
28278 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28280 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28282 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28284 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28286 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28287 instruction if possible.
28288 * config/s390/vector.md (vec_halfnumelts): New mode
28290 ("*vec_vllezlf<mode>"): New pattern.
28292 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28294 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28295 ("popcountv4si2", "popcountv2di2"): Rename to ...
28296 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28297 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28299 ("popcount<mode>2_vxe"): New pattern.
28301 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28303 * common/config/s390/s390-common.c (processor_flags_table): Add
28305 * config.gcc: Add arch12.
28306 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28307 Default to arch12 for unknown CPU model numbers.
28308 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28309 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28310 PROCESSOR_max sanity check.
28311 * config/s390/s390-opts.h (enum processor_type): Add
28313 * config/s390/s390.c (processor_table): Add arch12.
28314 (s390_expand_builtin): Add check for B_VXE flag.
28315 (s390_issue_rate): Add PROCESSOR_ARCH12.
28316 (s390_get_sched_attrmask): Likewise.
28317 (s390_get_unit_mask): Likewise.
28318 (s390_sched_score): Enable z13 scheduling for arch12.
28319 (s390_sched_reorder): Likewise.
28320 (s390_sched_variable_issue): Likewise.
28321 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28323 (s390_tune_attr): Use z13 scheduling also for arch12.
28324 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28325 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28326 (TARGET_VXE_P): New macros.
28327 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28328 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28329 * config/s390/s390.opt: Add arch12 as processor_type.
28331 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28333 * config/s390/s390.md
28334 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28335 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28336 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28338 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28339 Rename expanders to ...
28341 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28342 ("fixuns_truncdddi2_emu"): ... these.
28344 ("fixuns_trunc<mode>si2_emu"): New expander.
28346 ("*fixuns_truncdfdi2_z13"): Rename to ...
28347 ("*fixuns_truncdfdi2_vx"): ... this.
28349 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28351 * config/s390/2964.md: Remove the single element vector compare
28352 instructions which are no longer used.
28353 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28355 (s390_canonicalize_comparison): Remove handling of DFmode
28357 (s390_expand_vec_compare_scalar): Remove function.
28358 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28359 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28361 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28363 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28365 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28367 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28368 will anyway by matched by mov<mode>_64dfp.
28370 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28372 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28373 vlef/vstef. Add missing operand to vleif.
28375 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28377 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28378 pair for all vector types with 64 bit elements.
28379 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28380 * config/s390/vector.md (V_HW_64): ... here.
28381 (V_128_NOSINGLE): New mode iterator.
28382 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28383 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
28384 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
28385 ("*vec_load_pairv2di"): Change to ...
28386 ("*vec_load_pair<mode>"): ... this one.
28388 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28390 * config/s390/constraints.md: Add comments.
28391 (jKK): Reject element sizes > 8 bytes.
28392 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
28394 * config/s390/s390.md: Add the s_operand checks formerly in
28395 s390_split_ok_p to various splitters where they are still
28397 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
28398 for 128 bit vectors. Plus two splitters.
28400 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28402 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
28405 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28408 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
28409 error if the boundary argument is not constant.
28411 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28413 PR rtl-optimization/80112
28414 * loop-doloop.c (doloop_condition_get): Don't check condition
28415 if cmp isn't SET with IF_THEN_ELSE src.
28417 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28419 PR tree-optimization/80158
28420 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
28421 replacing a candidate statement, also replace it for the
28422 candidate's alternate interpretation.
28423 (replace_rhs_if_not_dup): Likewise.
28424 (replace_one_candidate): Likewise.
28426 2017-03-24 Richard Biener <rguenther@suse.de>
28428 PR tree-optimization/80167
28429 * graphite-isl-ast-to-gimple.c
28430 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
28432 (translate_isl_ast_to_gimple::get_rename): Likewise.
28434 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28436 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
28437 handling of certain combinations of target options, including the
28438 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
28439 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
28441 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28444 * config/arm/arm.md (*load_multiple): Add reload_completed to
28445 matching condition.
28447 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28448 Richard Biener <rguenth@suse.de>
28450 PR tree-optimization/79908
28451 PR tree-optimization/80136
28452 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28453 been cast away, gimplify_and_add suffices.
28455 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
28457 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
28459 2017-03-23 Richard Biener <rguenther@suse.de>
28461 PR tree-optimization/80032
28462 * gimplify.c (gimple_push_cleanup): Forced unconditional
28463 cleanups still have to go to the conditional_cleanups
28466 2017-03-22 Jakub Jelinek <jakub@redhat.com>
28468 PR tree-optimization/80072
28469 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
28471 (next_operand_entry_id): Change type to unsigned int.
28472 (sort_by_operand_rank): Make sure to return the right return value
28473 even if unsigned fields are bigger than INT_MAX.
28474 (struct oecount): Change cnt and id type to unsigned int.
28475 (oecount_hasher::equal): Formatting fix.
28476 (oecount_cmp): Make sure to return the right return value
28477 even if unsigned fields are bigger than INT_MAX.
28478 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
28481 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
28482 TREE_READONLY on result if writing it more than once.
28485 * doc/invoke.texi (-fsanitize=thread): Document that with
28486 -fnon-call-exceptions atomics are not able to throw
28490 * tsan.c: Include tree-eh.h.
28491 (instrument_builtin_call): Call maybe_clean_eh_stmt or
28492 maybe_clean_or_replace_eh_stmt where needed.
28493 (instrument_memory_accesses): Add cfg_changed argument.
28494 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
28495 if it returned true.
28496 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
28498 PR rtl-optimization/63191
28499 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
28500 wrapper function, moved the whole old content into ...
28501 (ix86_delegitimize_address_1): ... this. New inline function.
28502 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
28503 true as last argument instead of ix86_delegitimize_address.
28505 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28507 * config/aarch64/aarch64.c (generic_branch_cost): Copy
28508 cortexa57_branch_cost.
28510 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28512 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
28514 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28517 * doc/md.texi (Constraints): Document wA constraint.
28518 * config/rs6000/constraints.md (wA): New.
28519 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
28520 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
28521 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
28522 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
28524 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
28527 * gimplify.c (is_oacc_declared): New function.
28528 (oacc_default_clause): Use it to set default flags for acc declared
28529 variables inside parallel regions.
28530 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
28531 declared variables.
28532 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
28533 declare attribute to any decl as necessary.
28535 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
28538 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
28539 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
28540 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
28541 (arm_arch_lpae): This.
28542 * config/arm/arm.c (arm_arch7ve): Rename into ...
28543 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
28544 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
28547 2017-03-22 Martin Liska <mliska@suse.cz>
28550 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
28551 error message instead of an ICE.
28553 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28555 * doc/extend.texi (6.11 Additional Floating Types): Revise.
28557 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
28559 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28561 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28564 2017-03-21 Martin Sebor <msebor@redhat.com>
28566 * doc/extend.texi: Use "cannot" instead of "can't."
28567 * doc/hostconfig.texi: Same.
28568 * doc/install.texi: Same.
28569 * doc/invoke.texi: Same.
28570 * doc/loop.texi: Same.
28571 * doc/md.texi: Same.
28572 * doc/objc.texi: Same.
28573 * doc/rtl.texi: Same.
28574 * doc/tm.texi: Same.
28575 * doc/tm.texi.in: Same.
28576 * doc/trouble.texi: Same.
28578 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
28581 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
28582 (collect_checksum_attributes): Set it.
28583 (die_checksum_ordered): Use it.
28585 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28587 PR tree-optimization/79908
28588 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
28589 change: For a VA_ARG whose LHS has been cast away, use
28590 force_gimple_operand to construct the side effects.
28592 2017-03-21 David Malcolm <dmalcolm@redhat.com>
28594 PR translation/80001
28595 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
28596 more amenable to translation.
28597 (oacc_loop_auto_partitions): Likewise.
28599 2017-03-21 Marek Polacek <polacek@redhat.com>
28600 Martin Sebor <msebor@redhat.com>
28602 PR tree-optimization/80109
28603 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
28604 on INTEGRAL_TYPE_P.
28606 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28607 Segher Boessenkool <segher@kernel.crashing.org>
28610 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
28611 check reg_used_between_p between insn and one of succ or succ2
28612 depending on if succ is artificial insn not inserted into insn
28615 2017-03-21 Martin Liska <mliska@suse.cz>
28617 PR gcov-profile/80081
28618 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
28619 * doc/gcc.texi: Include gcov-dump stuff.
28620 * doc/gcov-dump.texi: New file.
28622 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
28624 PR rtl-optimization/79150
28625 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
28626 conditional jump, if the jump is the last insn of the loop.
28628 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28629 Richard Biener <rguenth@suse.de>
28631 PR tree-optimization/79908
28632 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28633 been cast away, use force_gimple_operand to construct the side
28636 2017-03-21 Martin Liska <mliska@suse.cz>
28638 PR libfortran/79956
28639 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
28642 2017-03-21 Brad Spengler <spender@grsecurity.net>
28645 * plugin.c (htab_hash_plugin): New function.
28646 (add_new_plugin): Use it and adjust.
28647 (parse_plugin_arg_opt): Adjust.
28648 (init_one_plugin): Likewise.
28650 2017-03-21 Richard Biener <rguenther@suse.de>
28652 PR tree-optimization/80032
28653 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
28654 if set force the cleanup to happen unconditionally.
28655 (gimplify_target_expr): Push inserted clobbers with force_uncond
28656 to avoid them being removed by control-dependent DCE.
28658 2017-03-21 Richard Biener <rguenther@suse.de>
28660 PR tree-optimization/80122
28661 * tree-inline.c (copy_bb): Do not expans va-arg packs or
28662 va_arg_pack_len when the inlined call stmt requires pack
28664 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
28666 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28669 * tsan.c (instrument_builtin_call): If the memory model argument
28670 is not a constant, assume it is valid.
28673 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
28676 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28678 PR rtl-optimization/79910
28679 * combine.c (can_combine_p): Do not allow combining an I0 or I1
28680 if its dest is used by an insn before I2 (other than the combined
28681 insns themselves, which are properly handled already).
28683 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
28686 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28688 * combine.c (record_used_regs): New static function.
28689 (try_combine): Handle situations where there is an additional
28690 instruction between I2 and I3 which needs to have a LOG_LINK
28694 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28696 * combine.c (try_combine): Delete redundant i1 test. Call
28697 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28699 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28702 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
28703 alternatives 13/14.
28705 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28707 PR tree-optimization/80054
28708 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
28709 the optimization if a PHI or any of its arguments is not dominated
28710 by the candidate's basis. Use gphi* rather than gimple* as
28712 (replace_profitable_candidates): Clean up a gimple* variable that
28713 should be a gphi* variable.
28715 2017-03-20 Martin Sebor <msebor@redhat.com>
28718 * doc/extend.texi (attribute constructor): Document present limitation.
28720 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
28723 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
28724 __POWER9_VECTOR__ #ifdef control, change template definition to
28725 use Power9-specific built-in function.
28726 (vec_any_eq): Likewise.
28727 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
28728 to control outcomes from this test.
28729 (vector_ae_<mode>p): For VEC_F modes, likewise.
28731 2017-03-20 Ian Lance Taylor <iant@google.com>
28733 * config/i386/i386.c (ix86_function_regparm): Save an extra
28734 register for -fsplit-stack with DECL_STATIC_CHAIN.
28736 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28739 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
28740 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
28742 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
28744 * config/riscv/riscv.c (riscv_print_operand): Use "fence
28746 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
28749 2017-03-20 Marek Polacek <polacek@redhat.com>
28752 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
28754 2017-03-20 Richard Biener <rguenther@suse.de>
28756 PR tree-optimization/80113
28757 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
28758 allocate extra SSA name for PHI def.
28759 (add_close_phis_to_outer_loops): Likewise.
28760 (add_close_phis_to_merge_points): Likewise.
28761 (copy_loop_close_phi_args): Likewise.
28762 (copy_cond_phi_nodes): Likewise.
28764 2017-03-20 Martin Liska <mliska@suse.cz>
28766 PR middle-end/79753
28767 * tree-chkp.c (chkp_build_returned_bound): Do not build
28768 returned bounds for a LHS that's not a BOUNDED_P type.
28770 2017-03-20 Martin Liska <mliska@suse.cz>
28774 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
28775 COMPLEX_CST and VECTOR_CST.
28777 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28780 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
28781 target operand. A new splitter adds the clobber statement in case
28782 the target operand is dead anyway.
28784 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
28786 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
28787 to age-old versions of binutils and glibc.
28789 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
28791 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
28793 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28795 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
28797 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
28799 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
28800 requirement for binutils 2.13.
28802 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
28804 * combine.c (try_combine): Delete redundant i1 test. Call
28805 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28807 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
28809 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
28810 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
28812 <riscv64-*-elf>: Re-arrange section
28813 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
28814 <riscv32-*-linux>: Likewise.
28815 <riscv64-*-elf>: Likewise
28816 <riscv64-*-linux>: Likewise.
28818 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
28821 * aarch64.opt(verbose-cost-dump): Fix typo.
28823 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
28826 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
28827 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
28829 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28831 * reload.c (find_reloads): When reloading a nonoffsettable address,
28832 use RELOAD_OTHER for it and its address reloads.
28834 PR rtl-optimization/79910
28835 * combine.c (record_used_regs): New static function.
28836 (try_combine): Handle situations where there is an additional
28837 instruction between I2 and I3 which needs to have a LOG_LINK
28840 2017-03-17 Jeff Law <law@redhat.com>
28842 PR tree-optimization/71437
28843 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28844 conditional in the hash table first.
28845 (vrp_dom_walker::before_dom_children): Extract condition from
28846 ASSERT_EXPR. Record condition, its inverion and any implied
28847 conditions as well.
28849 2017-03-17 Marek Polacek <polacek@redhat.com>
28850 Markus Trippelsdorf <markus@trippelsdorf.de>
28852 PR tree-optimization/80079
28853 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28856 2017-03-17 Richard Biener <rguenther@suse.de>
28858 PR middle-end/80075
28859 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28860 Properly verify the LHS before the RHS possibly claims to be
28862 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28865 2017-03-17 Martin Jambor <mjambor@suse.cz>
28867 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28868 (List of -O2 options): Likewise.
28869 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28872 2017-03-17 Tom de Vries <tom@codesourcery.com>
28874 * gcov-dump.c (print_usage): Print bug_report_url.
28876 2017-03-17 Richard Biener <rguenther@suse.de>
28878 PR middle-end/80050
28879 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28880 (parser::peek): Likewise.
28882 2017-03-17 Richard Biener <rguenther@suse.de>
28884 PR tree-optimization/80048
28885 * sese.c (free_sese_info): Properly release rename_map and
28886 copied_bb_map elements.
28888 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
28890 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
28891 Add linked-list forward and backlinks. Insert on
28892 construction, remove on destruction.
28893 (class pass_store_merging): Add m_stores_head field.
28894 (pass_store_merging::terminate_and_process_all_chains):
28895 Iterate over m_stores_head list.
28896 (pass_store_merging::terminate_all_aliasing_chains):
28898 (pass_store_merging::execute): Check for debug stmts first.
28899 Push new chains onto the m_stores_head stack.
28901 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28904 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28905 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28906 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28908 2017-03-16 Jeff Law <law@redhat.com>
28910 PR tree-optimization/71437
28911 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28912 member function. Implementation moved into after_dom_children
28913 member function and into the threader's thread_outgoing_edges
28915 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28916 some code into new thread_outgoing_edges.
28917 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28918 definition. Simplify marker handling (do it here). Assume we always
28919 have the available expression and the const/copies tables.
28920 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28922 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28923 * tree-vrp.c (equiv_stack): No longer file scoped.
28924 (vrp_dom_walker): New class.
28925 (vrp_dom_walker::before_dom_children): New member function.
28926 (vrp_dom_walker::after_dom_children): Likewise.
28927 (identify_jump_threads): Setup domwalker. Use it rather than
28928 walking edges in a random order by hand. Simplify setup/finalization.
28929 (finalize_jump_threads): Remove.
28930 (vrp_finalize): Do not call identify_jump_threads here.
28931 (execute_vrp): Do it here instead and call thread_through_all_blocks
28934 PR tree-optimization/71437
28935 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
28937 (simplify_stmt_for_jump_threading): Add basic_block argument. All
28939 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
28940 (dom_opt_dom_walker::thread_across_edge): Remove
28941 handle_dominating_asserts argument. All callers changed.
28942 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
28943 changes. Remove calls to lhs_of_dominating_assert. Other
28944 uses of handle_dominating_asserts turn into unconditional code
28945 (simplify_control_stmt_condition_1): Likewise.
28946 (simplify_control_stmt_condition): Likewise.
28947 (thread_through_normal_block, thread_across_edge): Likewise.
28948 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
28949 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
28950 object if it is not an SSA_NAME.
28951 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
28952 before calling into the VRP specific simplifiers.
28953 (identify_jump_threads): Remove handle_dominating_asserts
28956 2017-03-16 Jakub Jelinek <jakub@redhat.com>
28959 * tree-diagnostic.c (default_tree_printer): No longer static.
28960 * tree-diagnostic.h (default_tree_printer): New prototype.
28962 2017-03-16 Tamar Christina <tamar.christina@arm.com>
28964 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
28965 Change ins into fmov.
28967 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28969 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
28970 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
28971 Use h_con constraint for operand 1.
28972 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
28973 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
28975 2017-03-15 Jeff Law <law@redhat.com>
28977 PR tree-optimization/71437
28978 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
28979 (record_temporary_equivalences): Use it.
28981 PR tree-optimization/71437
28982 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28983 tree-ssa-scopedtables.
28984 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28985 (record_conditions, record_cond, vuse_eq): Likewise.
28986 (record_edge_info): Adjust to API tweak of record_conditions.
28987 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
28988 (record_temporary_equivalences, optimize_stmt): Likewise.
28989 (eliminate_redundant_computations): Likewise.
28990 (record_equivalences_from_stmt): Likewise.
28991 * tree-ssa-scopedtables.c: Include options.h and params.h.
28992 (vuse_eq): New function, moved from tree-ssa-dom.c
28993 (build_and_record_new_cond): Likewise.
28994 (record_conditions): Likewise. Accept vector of conditions rather
28995 than edge_equivalence structure for first argument.
28996 for the first argument.
28997 (avail_exprs_stack::lookup_avail_expr): New member function, moved
28998 from tree-ssa-dom.c.
28999 (avail_exprs_stack::record_cond): Likewise.
29000 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29001 from tree-ssa-dom.c.
29002 (avail_exprs_stack): Add new member functions lookup_avail_expr
29004 (record_conditions): Declare.
29006 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29009 * lra-constraints.c (process_alt_operands): Increase reject for
29010 reloading an input/output operand.
29012 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29015 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29016 insns to convert from signed/unsigned char/short to IEEE 128-bit
29018 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29020 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29023 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29024 subreg of inner mode for values already in registers.
29026 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29028 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29029 iteration reg is used after the loop.
29031 2017-03-14 Martin Sebor <msebor@redhat.com>
29033 PR tree-optimization/79800
29034 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29035 precision in negative-positive range.
29036 (format_floating): Call non-const overload with adjusted precision.
29038 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29041 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29044 2017-03-14 Martin Sebor <msebor@redhat.com>
29046 PR middle-end/80020
29047 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29048 * builtins.def (aligned_alloc): Use it.
29051 * Makefile.in (GTFILES): Add calls.c.
29052 * calls.c: Include "gt-calls.h".
29054 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29056 PR rtl-optimization/79728
29057 * regs.h (struct target_regs): New field
29058 x_contains_allocatable_regs_of_mode.
29059 (contains_allocatable_regs_of_mode): New macro.
29060 * reginfo.c (init_reg_sets_1): Initialize it, and change
29061 contains_reg_of_mode so it includes global regs as well.
29062 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29063 rather than contains_regs_of_mode.
29065 2017-03-14 Martin Liska <mliska@suse.cz>
29067 * doc/invoke.texi: Document options that can't be combined with
29068 -fcheck-pointer-bounds.
29070 2017-03-14 Martin Liska <mliska@suse.cz>
29072 PR middle-end/79831
29073 * doc/invoke.texi (-Wchkp): Document the option.
29075 2017-03-14 Martin Liska <mliska@suse.cz>
29077 * Makefile.in: Install gcov-dump.
29079 2017-03-14 Martin Liska <mliska@suse.cz>
29081 * multiple_target.c (expand_target_clones): Bail out for
29082 an invalid attribute.
29084 2017-03-14 Richard Biener <rguenther@suse.de>
29086 * alias.c (struct alias_set_entry): Pack properly.
29087 * cfgloop.h (struct loop): Likewise.
29088 * cse.c (struct set): Likewise.
29089 * ipa-utils.c (struct searchc_env): Likewise.
29090 * loop-invariant.c (struct invariant): Likewise.
29091 * lra-remat.c (struct cand): Likewise.
29092 * recog.c (struct change_t): Likewise.
29093 * rtl.h (struct address_info): Likewise.
29094 * symbol-summary.h (function_summary): Likewise.
29095 * tree-loop-distribution.c (struct partition): Likewise.
29096 * tree-object-size.c (struct object_size_info): Likewise.
29097 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29098 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29099 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29100 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29101 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29102 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29103 (struct _stmt_vec_info): Likewise.
29105 2017-03-14 Martin Liska <mliska@suse.cz>
29108 * multiple_target.c (create_dispatcher_calls): Check that
29109 a target can create a function dispatcher.
29111 2017-03-14 Martin Liska <mliska@suse.cz>
29114 * multiple_target.c (expand_target_clones): Drop local.local
29115 flag for default implementation.
29117 2017-03-14 Richard Biener <rguenther@suse.de>
29119 PR tree-optimization/80030
29120 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29122 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29124 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29125 gcc_fallthrough() instead of __attribute__((fallthrough));
29127 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29129 * doc/gcc.texi: Remove "up" link to (DIR).
29130 * doc/gccint.texi: Ditto.
29132 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29134 * doc/install.texi (Specific) <avr>: Remove reference to
29137 2017-03-13 Jeff Law <law@redhat.com>
29139 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29140 attribute rather than comments.
29142 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29143 match_scratch operand is highest.
29145 2017-03-13 Martin Liska <mliska@suse.cz>
29147 PR middle-end/78339
29148 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29149 is a CHKP clone, use original declaration.
29151 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29153 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29154 (arc_conditional_register_usage): Use a different allocation order
29155 when optimizing for size.
29156 * common/config/arc/arc-common.c (arc_option_optimization_table):
29157 Section anchors default on when optimizing for size.
29159 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29161 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29163 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29165 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29166 * config/arc/arc.md (cpu_facility): Add cd variant.
29167 (*movqi_insn): Add code density variant.
29168 (*movhi_insn): Likewise.
29169 (*movqi_insn): Likewise.
29170 (*addsi3_mixed): Likewise.
29171 (subsi3_insn): Likewise.
29173 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29175 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29177 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29179 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29180 expressions with MINUS and UNARY ops.
29182 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29185 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29187 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29188 between vec_select and vector argument.
29189 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29190 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29191 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29192 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29193 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29194 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29196 2017-03-13 Richard Biener <rguenther@suse.de>
29199 * params.def (vect-max-peeling-for-alignment): Fix typo.
29201 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29203 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29204 issue that only occurred with binutils below 2.18.
29206 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29208 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29209 refer to binutils 2.11/2.12 minimum.
29211 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29213 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29214 ftp.kernel.org and simplify binutils requirement.
29216 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29218 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29220 (Optimize Options): Ditto. Also remove redundancy.
29222 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29224 PR translation/79848
29225 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29227 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29228 to G_ to avoid double translation.
29230 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29232 PR translation/79923
29233 * auto-profile.c (get_combined_location): Convert leading
29234 character of diagnostics to lower case and remove trailing period.
29235 (read_profile): Likewise for various diagnostics.
29236 * config/arm/arm.c (arm_option_override): Remove trailing period
29237 from various diagnostics.
29238 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29239 (msp430_expand_delay_cycles): Likewise.
29241 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29244 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29245 full command-line argument, rather than just "str".
29246 (aarch64_validate_march): Likewise.
29247 (aarch64_validate_mtune): Likewise.
29249 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29251 PR rtl-optimization/78911
29252 * lra-assigns.c (must_not_spill_p): New function.
29253 (spill_for): Use it.
29255 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29257 PR tree-optimization/79981
29258 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29259 ATOMIC_COMPARE_EXCHANGE ifn result.
29260 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29261 IFN_ATOMIC_COMPARE_EXCHANGE.
29263 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29266 * opts.c (parse_sanitizer_options): Add missing question mark to
29267 "did you mean" message.
29269 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29271 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29273 (VMULEUH_UNS): Likewise.
29274 (VMULOUB_UNS): Likewise.
29275 (VMULOUH_UNS): Likewise.
29276 * config/rs6000/rs6000.c (builtin_function_type): Remove
29277 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29279 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29282 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29283 x with result of extra_parsing_for_operand_code_0.
29284 (function_reader::extra_parsing_for_operand_code_0): Convert
29285 return type from void to rtx, returning x. When reading
29286 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29287 larger size containing struct block_symbol.
29289 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29291 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29292 -mfloat128-hardware without -m64.
29294 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29297 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29298 entries to the case statement that marks unsigned arguments to
29299 overloaded functions.
29301 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29303 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29304 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29306 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29309 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29310 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29312 2017-03-10 Martin Liska <mliska@suse.cz>
29316 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29317 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29320 2017-03-10 Olivier Hainque <hainque@adacore.com>
29322 * tree-switch-conversion (array_value_type): Start by resetting
29323 candidate type to it's main variant.
29325 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29327 PR rtl-optimization/79909
29328 * combine.c (try_combine): Use simplify_replace_rtx on individual
29329 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29330 of the whole CALL_INSN_FUNCTION_USAGE.
29332 PR tree-optimization/79972
29333 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29334 get_range_info on SSA_NAMEs. Formatting fixes.
29336 2017-03-10 Richard Biener <rguenther@suse.de>
29337 Jakub Jelinek <jakub@redhat.com>
29339 PR tree-optimization/77975
29340 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29341 edge to be constant.
29342 (get_val_for): For constant x return it. Formatting fix.
29343 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29344 would use the same bases as the current one.
29346 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29348 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29349 instead of vec_select for V1TImode.
29350 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29352 (VSX_LE_128): Add V1TI to this mode iterator.
29353 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29354 (*vsx_le_perm_store_<mode>): Likewise.
29355 (pre-reload splitter for VSX stores): Likewise.
29356 (post-reload splitter for VSX stores): Likewise.
29357 (*vsx_xxpermdi2_le_<mode>): Likewise.
29358 (*vsx_lxvd2x2_le_<mode>): Likewise.
29359 (*vsx_stxvd2x2_le_<mode>): Likewise.
29361 2017-03-09 Michael Eager <eager@eagercon.com>
29363 Correct failures with --enable-checking=yes,rtl.
29365 * config/microblaze/microblaze.c (microblaze_expand_shift):
29366 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29367 test for const0_rtx.
29368 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29369 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29371 2017-03-09 Richard Biener <rguenther@suse.de>
29373 PR tree-optimization/79977
29374 * graphite-scop-detection.c (scop_detection::merge_sese):
29375 Handle the case of extra exits to blocks dominating the entry.
29377 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29379 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29382 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
29384 PR rtl-optimization/79949
29385 * lra-constraints.c (process_alt_operands): Check memory when
29386 trying to predict a cycle. Print about the overall increase.
29388 2017-03-09 Richard Biener <rguenther@suse.de>
29390 PR middle-end/79971
29391 * gimple-expr.c (useless_type_conversion_p): Preserve
29392 TYPE_SATURATING for fixed-point types.
29394 2017-03-09 Richard Biener <rguenther@suse.de>
29397 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
29398 alignment of BLKmode params.
29400 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29403 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
29404 (VALL_NO_V2Q): Likewise.
29406 * config/aarch64/aarch64-simd.md
29407 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
29409 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
29410 VALL_NO_V2Q mode iterator.
29411 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
29413 2017-03-09 Martin Liska <mliska@suse.cz>
29415 PR tree-optimization/79631
29416 * tree-chkp-opt.c (chkp_is_constant_addr): Call
29417 tree_int_cst_sign_bit just for INTEGER constants.
29419 2017-03-09 Martin Liska <mliska@suse.cz>
29423 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
29426 2017-03-09 Marek Polacek <polacek@redhat.com>
29429 * tree.c (inchash::add_expr): Handle TREE_VEC.
29431 2017-03-09 Martin Liska <mliska@suse.cz>
29434 (chkp_narrow_size_and_offset): New function.
29435 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
29436 (void chkp_parse_bit_field_ref): New function.
29437 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
29438 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
29440 2017-03-09 Martin Liska <mliska@suse.cz>
29443 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
29444 (chkp_find_bounds_1): Remove gcc_unreachable.
29446 2017-03-09 Jakub Jelinek <jakub@redhat.com>
29449 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
29450 BUILT_IN_SYNC*, determine the access type from the size suffix and
29451 always build a MEM_REF with that type. Handle forgotten
29452 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
29455 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
29456 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
29457 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
29458 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
29459 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
29460 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
29461 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
29462 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
29463 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
29464 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
29465 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
29466 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
29467 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
29468 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
29469 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
29470 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
29471 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
29472 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
29473 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
29474 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
29475 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
29476 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
29477 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
29478 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
29479 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
29480 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
29481 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
29482 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
29483 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
29484 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
29485 definitions outside of __OPTIMIZE__ guarded section.
29488 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
29489 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
29490 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
29491 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
29494 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29496 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
29497 ("vfenez<mode>"): Add missing constraints.
29499 2017-03-08 Martin Sebor <msebor@redhat.com>
29502 * config/nds32/nds32.c (nds32_option_override):
29503 Fix misspelled diagnostic.
29505 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29508 * gimplify.c (gimplify_omp_for): Replace index var in outer
29509 taskloop statement with an artificial variable and add
29510 OMP_CLAUSE_PRIVATE clause for it.
29512 2017-03-08 Richard Biener <rguenther@suse.de>
29514 PR tree-optimization/79955
29515 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
29516 for accesses that are completely outside of the variable.
29518 2017-03-08 Andrew Haley <aph@redhat.com>
29520 PR tree-optimization/79943
29521 * tree-ssa-loop-split.c (compute_new_first_bound): When
29522 calculating the new upper bound, (END-BEG) should be added, not
29525 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29527 * config/avr/avr.md (setmemhi): Make sure match_dup
29528 operand number comes before match_scratch.
29530 2017-03-08 Richard Biener <rguenther@suse.de>
29532 PR tree-optimization/79920
29533 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
29534 with ncopies == 1 to ...
29535 (vect_transform_slp_perm_load): ... here. Properly compute
29536 all element loads by iterating VF times over the group. Do
29537 not handle ncopies (computed in a broken way) in
29538 vect_create_mask_and_perm.
29540 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29543 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
29544 is a uniform vector, use uniform_vector_p return value instead of
29545 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
29547 2017-03-07 Marek Polacek <polacek@redhat.com>
29549 PR middle-end/79809
29550 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
29551 (alloca_call_type): Likewise.
29553 2017-03-07 Martin Liska <mliska@suse.cz>
29555 * gcov.c (process_args): Put comment to correct location.
29557 2017-03-07 Martin Liska <mliska@suse.cz>
29559 PR middle-end/68270
29560 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
29561 Use array_at_struct_end_p instead of DECL_CHAIN (field).
29562 (chkp_narrow_bounds_for_field): Likewise.
29563 (chkp_parse_array_and_component_ref): Pass one more argument to
29566 2017-03-07 Richard Biener <rguenther@suse.de>
29568 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
29571 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
29573 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
29574 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
29576 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29579 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
29580 to end of description.
29581 (PARAM_MAX_STORES_TO_MERGE): Likewise.
29583 2017-03-07 Jakub Jelinek <jakub@redhat.com>
29585 PR rtl-optimization/79901
29586 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
29588 (*avx512f_<code><mode>3<mask_name>): ... this.
29589 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
29590 iterator instead of VI8_AVX2_AVX512BW.
29592 PR rtl-optimization/79901
29593 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
29594 min/max expander, expand it using expand_vec_cond_expr.
29597 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
29600 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29603 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
29604 to void * for PCH reasons.
29605 * dwarf2out.c (output_loc_operands, output_die): Cast
29606 v.val_vec.array to unsigned char *.
29608 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
29611 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
29614 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
29616 PR rtl-optimization/79571
29617 * lra-constraints.c (process_alt_operands): Calculate static
29618 reject and subtract it from overall when only addresses will be
29621 2017-03-06 Julia Koval <julia.koval@intel.com>
29624 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
29625 incoming stack boundary to 128 for 64-bit targets.
29627 2017-03-06 Richard Biener <rguenther@suse.de>
29629 PR tree-optimization/79894
29630 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
29631 to NULL after folding it.
29633 2017-03-06 Richard Biener <rguenther@suse.de>
29635 PR tree-optimization/79824
29636 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
29637 check disabling peeling for gaps.
29639 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
29641 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
29642 attributes): Document gettimeofday.
29644 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29646 * config/s390/s390.c (s390_option_override_internal): Set
29647 PARAM_MIN_VECT_LOOP_BOUND
29649 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29651 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
29652 * config/s390/s390.md: Likewise.
29654 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29657 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
29658 (<avx2_avx512>_perm<mode>): Rename to ...
29659 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
29661 (<avx512>_perm<mode>_mask): Rename to ...
29662 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
29664 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
29665 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
29666 instead of VI8F_256_512.
29667 (avx512f_perm<mode>): New define_expand.
29668 (avx512f_perm<mode>_mask): Likewise.
29669 (avx512f_perm<mode>_1<mask_name>): New define_insn.
29670 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
29672 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29674 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
29675 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
29677 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
29679 2017-03-06 Martin Liska <mliska@suse.cz>
29682 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
29683 when having a SSA NAME w/o VAR_DECL assigned to it.
29685 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29687 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
29688 msa_dpsub_<su>_d): Fix MODE for vec_select.
29690 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
29692 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
29694 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
29696 2017-03-06 Richard Biener <rguenther@suse.de>
29698 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
29699 * plugin.c (register_plugin_info): Likewise.
29700 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
29702 2017-03-05 Jakub Jelinek <jakub@redhat.com>
29704 * config/i386/sse.md (sse_storehps, sse_storelps,
29705 avx_<castmode><avxsizesuffix>_<castmode>,
29706 avx512f_<castmode><avxsizesuffix>_<castmode>,
29707 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
29708 in condition that at least one operand is not a MEM.
29710 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29712 PR middle-end/79805
29713 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
29714 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
29716 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
29717 gimple_call_nothrow_p flag based on whether original builtin can throw.
29718 If it can, emit following stmts on the fallthrough edge.
29719 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
29720 don't create new bb if inserting just debug stmts on the edge, try to
29721 insert them on the fallthru bb or just reset debug stmts.
29723 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
29726 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
29727 restore recog_data (including the operand rtxes inside it) around
29728 the call to get_insn_template.
29730 2017-03-03 Martin Sebor <msebor@redhat.com>
29732 PR tree-optimization/79699
29733 * context.c (context::~context): Free MPFR caches to avoid
29734 a memory leak on program exit.
29736 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29738 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
29739 Use wide_int::ulow () instead of .elt (0).
29741 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29743 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
29744 (*pushxf): Limit oF constraint to 32bit targets and add oC
29745 constraint for 64bit targets.
29746 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
29747 (*pushdf): Change rmF constraint to rmC.
29749 2017-03-03 Martin Liska <mliska@suse.cz>
29751 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
29752 Remove unused variable.
29754 2017-03-03 Jakub Jelinek <jakub@redhat.com>
29757 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
29758 is a memory operand, increase num_memory.
29759 (ix86_expand_args_builtin): Likewise.
29761 2017-03-03 Jan Hubicka <jh@suse.cz>
29764 * ipa-devirt.c (maybe_record_node): Properly handle
29765 __cxa_pure_virtual visibility.
29767 2017-03-03 Martin Liska <mliska@suse.cz>
29769 PR tree-optimization/79803
29770 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
29772 (pass_loop_prefetch::execute): Disabled optimization if an
29773 assumption about L1 cache size is not met.
29775 2017-03-03 Martin Liska <mliska@suse.cz>
29777 PR rtl-optimization/79574
29778 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
29779 (hash_scan_set): Likewise.
29780 (dump_hash_table): Likewise.
29781 (hoist_code): Likewise.
29783 2017-03-03 Richard Biener <rguenther@suse.de>
29785 * fixed-value.c (fixed_from_string): Restore use of elt (1)
29786 in place of uhigh ().
29787 (fixed_convert_from_real): Likewise.
29789 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
29792 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
29794 2017-03-03 Richard Biener <rguenther@suse.de>
29796 PR middle-end/79818
29797 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
29798 TYPE_OVERFLOW_UNDEFINED check.
29800 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29802 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
29804 (vector_ae_<mode>_p): Likewise.
29805 (vector_nez_<mode>_p): Likewise.
29806 (vector_ne_v2di_p): Likewise.
29807 (vector_ae_v2di_p): Likewise.
29808 (vector_ne_<mode>_p): Likewise.
29809 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
29811 (vsx_tsqrt<mode>2_fe): Likewise.
29813 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
29816 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
29818 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29820 PR rtl-optimization/79780
29821 * cprop.c (one_cprop_pass): When second and further conditional trap
29822 in a single basic block is turned into an unconditional trap, turn it
29823 into a deleted note to avoid RTL verification failures.
29825 2017-03-02 Richard Biener <rguenther@suse.de>
29827 * fold-const.c (const_binop): Use ulow () instead of elt (0).
29829 2017-03-02 Richard Biener <rguenther@suse.de>
29831 PR tree-optimization/79345
29833 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29834 param and abort the walk, returning -1 if it is hit.
29835 (walk_aliased_vdefs): Take a limit param and pass it on.
29836 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29837 defaulting to 0 and return a signed int.
29838 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29839 (check_defs): New helper.
29840 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29841 about uninitialized memory.
29842 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29843 bogus uninitialized warning.
29844 (fixed_convert_from_real): Likewise.
29846 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29848 PR tree-optimization/66768
29849 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29850 iv_use if base object can't be determined.
29852 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29854 PR tree-optimization/79345
29855 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29856 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29857 (get_pattern_stats): Initialize it.
29858 * genemit.c (gen_expand): Verify match_scratch numbers come after
29859 match_operand/match_dup numbers.
29860 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29861 match_scratch numbers.
29862 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29864 * config/s390/s390.md (trunctdsd2): Likewise.
29866 2017-03-02 Richard Biener <rguenther@suse.de>
29868 * wide-int.h (wide_int_storage::operator=): Implement in terms
29871 2017-03-02 Richard Biener <rguenther@suse.de>
29873 PR tree-optimization/79777
29874 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29875 the to insert expression to sth existing.
29877 2017-03-01 Martin Sebor <msebor@redhat.com>
29879 PR middle-end/79692
29880 * gimple-ssa-sprintf.c
29881 (directive::known_width_and_precision): New function.
29882 (format_integer): Use it.
29883 (get_mpfr_format_length): Consider the full range of precision
29884 when computing %g output with the # flag. Set the likely byte
29885 count to 3 rather than 1 when precision is indeterminate.
29886 (format_floating): Correct the lower bound of precision.
29888 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29890 * doc/invoke.texi: Document default code model for 64-bit Linux.
29892 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29895 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
29896 udiv rather than div since input pattern is unsigned.
29898 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
29900 * config/i386/i386.c (print_reg): Warn for values of
29901 unsupported size in integer register.
29903 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29906 * config/rs6000/predicates.md (current_file_function_operand): Do
29907 not allow self calls to be local if the function is replaceable.
29909 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29912 * config/rs6000/altivec.h (vec_ctz and others): Change the
29913 preprocessor macro that controls conditional compilation from
29914 _ARCH_PWR9 to __POWER9_VECTOR__.
29915 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29916 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29917 control (instead of _ARCH_PWR9 control) so that template
29918 definition uses power9-specific function.
29919 (vec_any_eq): Likewise.
29920 (vec_all_ne): Change macro definition to use a power9-specific
29921 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29922 _ARCH_PWR9 control).
29923 (vec_any_eq) Likewise.
29924 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29925 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29926 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29927 support for xvcmpnedp instruction.
29928 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29929 macro expansion so that Power9 implementation of vec_all_ne does
29930 not use the AltiVec predicate framework.
29931 (VCMPNEH_P): Likewise.
29932 (VCMPNEW_P): Likewise.
29933 (VCMPNED_P): Likewise.
29934 (VCMPNEFP_P): Likewise.
29935 (VCMPNEDP_P): Likewise.
29936 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
29937 implementation of vec_any_eq to not use AltiVec predicate
29939 (VCMPAEH_P): Likewise.
29940 (VCMPAEW_P): Likewise.
29941 (VCMPAED_P): Likewise.
29942 (VCMPAEFP_P): Likewise.
29943 (VCMPAEDP_P): Likewise.
29944 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
29945 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
29946 not use the AltiVec predicate framework.
29947 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
29948 of vec_any_eq to not use AltiVec predicate framework.
29949 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29950 support for predefined __POWER9_VECTOR__ macro to indicate that
29951 Power9 instruction selection is enabled.
29952 (altivec_overloaded_builtins): Remove extraneous
29953 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
29954 function argument types RS6000_BTI_bool_V16QI and
29955 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
29956 entry for overloaded function argument types RS6000_BTI_bool_V4SI
29957 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
29958 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
29959 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
29960 Power9 for implementations of vec_cmpne. Change the signature for
29961 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
29962 (representing vec_all_ne) to remove the previously described first
29963 argument of type RS6000_BTI_INTSI, as this was an artifact of
29964 reliance on the AltiVec predicate framework, which is no longer
29965 used in the implementation of these functions. Add
29966 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
29967 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
29968 since, unlike the AltiVec predicate framework implementation, we
29969 do not share function descriptors between vec_alle and vec_anyeq.
29970 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
29971 set of modes that receive special treatment even when
29972 TARGET_P9_VECTOR is true. The special treatment emits code that
29973 does not depend on Power9 instructions.
29974 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
29975 define_expand to not rely on AltiVec predicate framework.
29976 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29978 (vector_ne_v2di_p): Change this define_expand to not rely on
29979 AltiVec predicate framework.
29980 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29982 (vector_ne_<mode>_p): Change this define_expand to not rely on
29983 AltiVec predicate framework.
29984 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29986 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
29987 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
29988 define_insn pattern.
29989 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
29990 define_insn pattern because the xvcmpne<VSs>. instruction is not
29992 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
29993 instruction is not supported.
29995 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29997 * config/nvptx/nvptx.c: Include intl.h.
29999 2017-03-01 Martin Jambor <mjambor@suse.cz>
30002 * ipa-prop.h (ipa_bits): Removed field known.
30003 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30004 to pointers. Adjusted their comments to warn about their sharing.
30005 (ipcp_transformation_summary): Change bits to a vector of pointers.
30006 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30007 (ipa_get_ipa_bits_for_value): Declare.
30008 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30009 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30010 (ipa_bits_hash_table): Likewise.
30011 (ipa_vr_ggc_hash_traits): Likewise.
30012 (ipa_vr_hash_table): Likewise.
30013 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30014 being pointers and vr_known being removed.
30015 (ipa_set_jf_unknown): Likewise.
30016 (ipa_get_ipa_bits_for_value): New function.
30017 (ipa_set_jfunc_bits): Likewise.
30018 (ipa_get_value_range): New overloaded functions.
30019 (ipa_set_jfunc_vr): Likewise.
30020 (ipa_compute_jump_functions_for_edge): Use the above functions to
30021 construct bits and vr parts of jump functions.
30022 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30023 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30025 (ipcp_grow_transformations_if_necessary): Also allocate
30026 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30028 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30029 them. Fix too long lines.
30030 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30031 vr_known being removed.
30032 (ipa_read_jump_function): Use new setter functions to construct bits
30033 and vr parts of jump functions or set them to NULL.
30034 (write_ipcp_transformation_info): Adjust for bits being pointers.
30035 (read_ipcp_transformation_info): Likewise.
30036 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30038 Include gt-ipa-prop.h.
30039 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30041 (ipcp_store_bits_results): Likewise.
30042 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30043 Do not write to existing jump functions but use a temporary instead.
30045 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30048 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30049 attempt to use its first operand as BIT_FIELD_REF base.
30051 2017-03-01 Richard Biener <rguenther@suse.de>
30053 PR middle-end/79721
30054 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30055 interpolating formula in wrapping arithmetic.
30056 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30058 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30060 PR tree-optimization/79734
30061 * tree-vect-generic.c (expand_vector_condition): Optimize
30062 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30063 Handle VEC_COND_EXPR where comparison has different inner width from
30064 type's inner width.
30066 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30068 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30069 markup, and similar issues. Remove @opindex entries for things
30070 that aren't options. Add missing -mmpy-option entries.
30072 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30074 PR tree-optimization/79737
30075 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30076 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30077 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30078 instead of byte_size. Formatting fix.
30079 (shift_bytes_in_array_right): Formatting fix.
30081 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30084 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30085 condition on optimize for the leaf function test.
30087 2017-02-28 Martin Liska <mliska@suse.cz>
30090 * read-rtl-function.c (function_reader::handle_unknown_directive):
30091 Bail out when one uses -flto.
30093 2017-02-28 Martin Liska <mliska@suse.cz>
30095 * common.opt: Replace space with tabular for options of <number>
30097 * config/i386/i386.opt: Show <number> value for
30098 -mlarge-data-threshold.
30099 * opts.c (print_filtered_help): Do not display number in hexadecimal
30102 2017-02-28 Martin Liska <mliska@suse.cz>
30104 * common.opt: Fix --help=option -Q for options which are of
30107 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30109 * config/i386/i386.c (print_reg): Error out for values
30110 of 8-bit size in invalid integer register.
30112 2017-02-28 Martin Sebor <msebor@redhat.com>
30114 PR tree-optimization/79691
30115 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30117 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30120 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30121 gcc_unreachable with output_operand_lossage.
30123 2017-02-28 Richard Biener <rguenther@suse.de>
30125 PR tree-optimization/79740
30126 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30128 (visit_nary_op): Insert the nary into the hashtable if we
30129 pattern-matched sth.
30130 * tree-ssa-pre.c (eliminate_insert): Robustify.
30132 2017-02-28 Richard Biener <rguenther@suse.de>
30134 PR middle-end/79731
30135 * fold-const.c (decode_field_reference): Reject out-of-bound
30138 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30140 * config/i386/i386.c: Include intl.h.
30141 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30142 instead of just cond ? "..." : "...".
30143 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30144 * coverage.c (read_counts_file): Likewise.
30145 * omp-offload.c: Include intl.h.
30146 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30147 of just cond ? "..." : "...".
30148 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30149 of just cond ? "..." : "...".
30151 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30154 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30156 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30157 'tune for' CPU name.
30158 * config/arm/arm-cpu-data.h: Regenerated.
30160 2017-02-28 Richard Biener <rguenther@suse.de>
30162 PR tree-optimization/79732
30163 * tree-inline.c (expand_call_inline): Do not shadow var.
30165 2017-02-28 Richard Biener <rguenther@suse.de>
30167 PR tree-optimization/79723
30168 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30169 address-space properly.
30171 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30173 * doc/optinfo.texi (Optimization groups): Fix option used for
30175 * doc/invoke.texi (-fopt-info): Document "omp".
30176 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30177 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30178 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30179 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30180 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30182 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30184 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30185 OPTGROUP_OMP with "omp".
30187 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30190 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30191 for arithmetic shift of unsigned V2DI.
30193 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30195 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30196 arc/linux.h headers.
30197 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30198 (LINK_SPEC): Likewise.
30199 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30200 (EXTRA_SPECS): Likewise.
30201 (STARTFILE_SPEC): Likewise.
30202 (ENDFILE_SPEC): Likewise.
30203 (LIB_SPEC): Likewise.
30204 (TARGET_SDATA_DEFAULT): Likewise.
30205 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30206 (MULTILIB_DEFAULTS): Likewise.
30207 (DWARF2_UNWIND_INFO): Likewise.
30208 * config/arc/big.h: New file.
30209 * config/arc/elf.h: Likewise.
30210 * config/arc/linux.h: Likewise.
30211 * config/arc/t-uClibc: Remove.
30213 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30215 PR tree-optimization/77536
30216 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30217 (tree_transform_and_unroll_loop): Use above function to compute the
30218 estimated niter of unrolled loop and use it when scaling profile.
30219 Also use count info rather than frequency if it's non-zero.
30220 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30221 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30222 (vect_transform_loop): Call above function.
30224 2017-02-27 Richard Biener <rguenther@suse.de>
30226 PR tree-optimization/45397
30227 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30228 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30229 (visit_nary_op): Add pattern matching for CSEing sign-changed
30230 or truncated operations with wider ones.
30232 2017-02-27 Richard Biener <rguenther@suse.de>
30234 PR tree-optimization/79690
30235 * tree-vect-stmts.c (vectorizable_store): Use vector type
30236 built from the DR with address-space.
30238 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30240 * doc/invoke.texi (Optimize Options): Refine the description
30241 of asan-use-after-return.
30243 2017-02-25 Alan Modra <amodra@gmail.com>
30245 PR rtl-optimization/79584
30246 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30247 base, not ad->base_term, the reg within base. Remove assertion
30248 that ad->base == ad->base_term. Replace gen_int_mode using
30249 bogus mode with const0_rtx.
30251 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30253 PR middle-end/79396
30254 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30255 FMA_EXPR like tcc_binary or tcc_unary.
30257 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30260 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30262 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30263 (output_loc_operands): Handle DW_OP_call_ref and
30264 DW_OP_GNU_variable_value.
30265 (struct variable_value_struct): New type.
30266 (struct variable_value_hasher): Likewise.
30267 (variable_value_hash): New variable.
30268 (string_types): Remove.
30269 (copy_loc_descr): New function.
30270 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30271 (prepend_loc_descr_to_each): New function.
30272 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30273 instead of add_loc_descr_to_each if the first argument is single
30274 location list and the second has multiple.
30275 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30276 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30277 when looking for variable value which doesn't have other location info.
30278 (loc_list_from_tree): Formatting fix.
30279 (gen_array_type_die): Simplify DW_AT_string_length handling.
30280 (adjust_string_types): Remove.
30281 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30282 string_types. Call resolve_variable_values.
30283 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30284 (resolve_addr_in_expr): Likewise. Add A argument.
30285 (copy_deref_exprloc): Remove deref argument. Adjust for the
30286 original expression being DW_OP_GNU_variable_value with optionally
30287 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30288 optionally after it.
30289 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30290 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30291 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30292 (variable_value_hasher::hash, variable_value_hasher::equal): New
30294 (resolve_variable_value_in_expr, resolve_variable_value,
30295 resolve_variable_values, note_variable_value_in_expr,
30296 note_variable_value): New functions.
30297 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30300 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30303 * opts.h (handle_generated_option): Add GENERATED_P argument.
30304 * opts-common.c (handle_option): Adjust function comment.
30305 (handle_generated_option): Add GENERATED_P argument, pass it to
30307 (control_warning_option): Pass false to handle_generated_option
30309 * opts.c (maybe_default_option): Pass true to handle_generated_option
30311 * optc-gen.awk: Likewise.
30313 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30315 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30316 a REG, look at the REG it is a SUBREG of.
30317 (splitter for cmpeqsi_t): Ditto.
30319 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30321 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30322 the special USEs with the pattern of the insn, not the insn itself.
30324 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30327 * doc/invoke.texi: Document -mload-store-pairs.
30329 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30330 Sandra Loosemore <sandra@codesourcery.com>
30332 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30333 argument isn't a CONST_INT.
30334 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30335 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30336 (nios2_validate_compare): Bypass alternate compare logic if *op2
30337 is not a CONST_INT.
30338 (ldstwm_operation_p): Return false if first_base is not a REG or
30339 if first_offset is not a CONST_INT.
30341 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30343 * config/cris/cris.md: Use correct operand in a define_peephole2.
30345 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30347 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30349 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30351 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30352 this_insn if it is an INSN or JUMP_INSN.
30353 (force_offsettable): Look at base, not at addr.
30354 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30355 on things that aren't necessarily CONST_INTs.
30357 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30359 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30360 -mfpmath=sse is the default also for x86-32 targets with SSE2
30361 instruction set when @option{-ffast-math} is enabled
30363 2017-02-24 Jeff Law <law@redhat.com>
30365 PR rtl-optimizatoin/79286
30366 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30369 2017-02-24 Richard Biener <rguenther@suse.de>
30371 PR tree-optimization/79389
30372 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30375 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30377 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30378 function comment to reflect reality.
30379 (loop_exits_before_overflow): Fix typo in function description.
30381 2017-02-24 Richard Biener <rguenther@suse.de>
30383 PR tree-optimization/79389
30384 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
30385 properly that a threading opportunity exists. Detect conditional
30386 copy/constant propagation opportunities.
30388 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
30390 * config/visium/visium.md (type): Add trap.
30391 (b): New mode attribute.
30392 (*btst): Rename into...
30393 (*btst<mode>): ...this and adjust.
30394 (*cbranchsi4_btst_insn): Rename into...
30395 (*cbranch<mode>4_btst_insn): ...this and adjust.
30396 (trap): New define_insn.
30398 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30400 PR tree-optimization/79389
30401 * ifcvt.c (struct noce_if_info): Add rev_cond field.
30402 (noce_reversed_cond_code): New function.
30403 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
30404 reversed_comparison_code. Formatting fix.
30405 (noce_try_store_flag): Test rev_cond != NULL in addition to
30406 reversed_comparison_code.
30407 (noce_try_store_flag_constants): Likewise.
30408 (noce_try_store_flag_mask): Likewise.
30409 (noce_try_addcc): Use rev_cond if non-NULL instead of
30410 reversed_comparison_code.
30411 (noce_try_cmove_arith): Likewise. Formatting fixes.
30412 (noce_try_minmax, noce_try_abs): Clear rev_cond.
30413 (noce_find_if_block): Initialize rev_cond.
30414 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
30415 instead of false as last argument never attempt to reverse it
30418 2017-02-23 Bin Cheng <bin.cheng@arm.com>
30420 PR tree-optimization/79663
30421 * tree-predcom.c (combine_chains): Process refs in reverse order
30422 only for ZERO length chains, and add explaining comment.
30424 2017-02-23 Jeff Law <law@redhat.com>
30426 PR tree-optimization/79578
30427 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
30428 in call to operand_equal_p.
30430 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
30433 * config/i386/cpuid.h: Fix another undefined behavior.
30435 2017-02-23 Richard Biener <rguenther@suse.de>
30437 PR tree-optimization/79683
30438 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
30439 vector types for data-refs.
30441 2017-02-23 Martin Liska <mliska@suse.cz>
30443 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
30445 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30447 PR middle-end/79665
30448 * internal-fn.c (get_range_pos_neg): Moved to ...
30449 * tree.c (get_range_pos_neg): ... here. No longer static.
30450 * tree.h (get_range_pos_neg): New prototype.
30451 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
30452 are known to be in between 0 and signed maximum inclusive, try to
30453 expand both unsigned and signed divmod and use the cheaper one from
30456 2017-02-22 Jeff Law <law@redhat.com>
30458 PR tree-optimization/79578
30459 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
30460 to compare base operands.
30462 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
30465 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
30466 gpc_reg_operand instead of fpr_reg_operand.
30468 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
30470 * config/mips/mips.c (mips_return_in_memory): Force FP
30471 vector types to be returned in memory for o32 ABI.
30473 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30475 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
30476 instead of DW_TAG_member for static data member declarations and don't
30477 set no_linkage_name for static inline data members.
30478 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
30481 2017-02-22 Martin Liska <mliska@suse.cz>
30483 * doc/invoke.texi: Replace inequality signs with square brackets
30486 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
30489 * lra-constraints.c (simplify_operand_subreg): Handle
30490 WORD_REGISTER_OPERATIONS targets.
30492 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30495 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
30496 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
30497 elimination by swapping fld*.
30499 2017-02-22 Richard Biener <rguenther@suse.de>
30501 PR tree-optimization/79673
30502 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
30503 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
30504 irrelevant address-space qualifiers and avoiding a
30505 ADDR_SPACE_CONVERT_EXPR from fold_convert.
30507 2017-02-22 Richard Biener <rguenther@suse.de>
30509 PR tree-optimization/79666
30510 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
30511 to not symbolically negate if that may introduce undefined
30514 2017-02-22 Martin Liska <mliska@suse.cz>
30517 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
30518 * data-streamer-out.c (streamer_write_gcov_count_stream):
30520 * value-prof.c (stream_out_histogram_value): Make assert more
30521 precise based on type of counter.
30523 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
30526 * config/i386/i386.md (standard_x87sse_constant_load splitter):
30527 Use nonimmediate_operand instead of memory_operand for operand 1.
30528 (float-extend standard_x87sse_constant_load splitter): Ditto.
30530 2017-02-21 Jeff Law <law@redhat.com>
30532 PR tree-optimization/79621
30533 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
30534 blocks with edges to themselves.
30536 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30539 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
30540 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
30541 Use gimple_call_builtin_p.
30544 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
30545 on temporarily removed DEBUG_INSNs.
30547 PR tree-optimization/79649
30548 * tree-loop-distribution.c (classify_partition): Give up on
30549 non-generic address space loads/stores.
30551 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
30553 * doc/loop.texi (Loop manipulation): Remove nonexistent
30554 tree_ssa_loop_version from the documentation.
30555 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
30557 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30560 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
30561 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
30562 * config/rs6000/rs6000.c: Include except.h.
30563 (rs6000_expand_split_stack_prologue): Call
30564 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
30566 2017-02-21 Martin Jambor <mjambor@suse.cz>
30569 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
30570 have been analyzed.
30572 2017-02-21 Martin Jambor <mjambor@suse.cz>
30574 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
30575 for backward compatibility only.
30576 * doc/invoke.texi (Option Summary): Remove all references to
30577 -fipa-cp-alignment.
30579 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
30583 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30585 * lra-constraints.c (curr_insn_transform): Handle
30586 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30588 2017-02-21 Martin Liska <mliska@suse.cz>
30590 * config/i386/i386.opt: Replace -masm-dialect with -masm.
30592 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
30594 PR translation/79638
30595 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
30597 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
30600 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
30601 (arm_function_ok_for_sibcall): Return false for an indirect call by
30602 descriptor if all the argument registers are used.
30603 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
30604 alignment of the function.
30606 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30608 PR tree-optimization/61441
30609 * simplify-rtx.c (simplify_const_unary_operation): For
30610 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
30611 the sNaN unmodified.
30613 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30615 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30616 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30617 instead of SYSTEM_HEADER_DIR.
30619 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
30620 Martin Liška <mliska@suse.cz>
30622 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
30623 Fix typos and grammar, use active voice, and clarify.
30625 2017-02-20 Marek Polacek <polacek@redhat.com>
30627 PR middle-end/79537
30628 * gimplify.c (gimplify_expr): Handle unused *&&L;.
30631 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
30633 2017-02-20 Jakub Jelinek <jakub@redhat.com>
30636 * config/i386/i386.c (ix86_expand_builtin): Handle
30637 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
30638 ix86_builtins_isa[fcode].isa as a requirement of those
30639 flags and any other flag in the bitmask.
30640 (ix86_init_mmx_sse_builtins): Use 0 instead of
30641 ~OPTION_MASK_ISA_64BIT as mask.
30642 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
30643 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
30644 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
30645 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
30647 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30650 * lra-constraints.c (split_reg): Check requested split mode
30651 is supported by the register.
30653 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30655 * lra-constraints.c (simplify_operand_subreg): Remove early
30658 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30661 * lra-constraints.c (curr_insn_transform): Tighten condition
30662 for converting SUBREG reloads from OP_OUT to OP_INOUT.
30664 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30667 * lra-constraints.c (curr_insn_transform): Handle
30668 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30670 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
30673 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
30675 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
30677 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
30680 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
30683 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30685 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
30688 2017-02-18 Jakub Jelinek <jakub@redhat.com>
30690 * final.c (last_columnnum, override_columnnum): New variables.
30691 (final_start_function): Set last_columnnum, pass it to begin_prologue
30692 hook and pass 0 to dwarf2out_begin_prologue.
30693 (final_scan_insn): Update override_columnnum. Pass last_columnnum
30694 to source_line debug hook.
30695 (notice_source_line): Compute last_columnnum and for debug_column_info
30696 return true on column changes.
30697 * debug.h (struct gcc_debug_hooks): Add column argument to
30698 source_line and begin_prologue hooks.
30699 (debug_nothing_int_charstar_int_bool): Remove prototype.
30700 (debug_nothing_int_int_charstar,
30701 debug_nothing_int_int_charstar_int_bool): New prototypes.
30702 (dwarf2out_begin_prologue): Add column argument.
30703 * debug.c (do_nothing_debug_hooks): Adjust source_line and
30704 begin_prologue hooks.
30705 (debug_nothing_int_charstar_int_bool): Remove.
30706 (debug_nothing_int_int_charstar,
30707 debug_nothing_int_int_charstar_int_bool): New functions.
30708 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
30709 through to dwarf2out_source_line.
30710 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
30711 (dwarf2out_source_line): Add column argument, emit it if requested.
30712 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
30714 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30715 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
30716 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
30717 through to dwarf2out_begin_prologue.
30718 (vmsdbgout_source_line): Add column argument, pass it through to
30719 dwarf2out_source_line.
30720 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
30721 dbxout_source_line caller.
30722 (dbxout_source_line): Add column argument.
30724 * common.opt (gno-column-info, gcolumn-info): New options.
30725 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
30726 (check_die): Also test for multiple DW_AT_decl_column attributes.
30727 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
30728 DW_AT_decl_column if requested.
30729 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
30731 (gen_variable_die): Likewise.
30732 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
30733 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
30736 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
30737 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
30738 (ix86_handle_option): Handle OPT_m3dnowa.
30739 * doc/invoke.texi (-m3dnowa): Document.
30740 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
30741 -m3dnowa instead of -m3dnow -march=athlon.
30744 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
30745 instead of gcc_assert for K, r and R code checks. Formatting fixes.
30747 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30750 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
30751 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
30752 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
30753 generator for vsx_xxpermdi_<mode>_be.
30754 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
30755 force big-endian semantics.
30756 (vsx_xxpermdi_<mode>_be): New define_expand with same
30757 implementation as previous version of vsx_xxpermdi_<mode>.
30759 2017-02-17 Jakub Jelinek <jakub@redhat.com>
30761 PR tree-optimization/79327
30762 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
30763 variable, its initialization and use.
30765 2017-02-17 Julia Koval <julia.koval@intel.com>
30767 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
30768 (OPTION_MASK_ISA_PKU_UNSET): New.
30769 (ix86_handle_option): Handle -mrdpid.
30770 * config/i386/cpuid.h (bit_RDPID): New.
30771 * config/i386/driver-i386.c (host_detect_local_cpu):
30772 Detect RDPID feature.
30773 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
30774 * config/i386/i386-c.c (ix86_target_macros_internal):
30776 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
30777 (ix86_valid_target_attribute_inner_p): Add "rdpid".
30778 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
30779 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
30780 * config/i386/i386.md (define_insn "rdpid"): New.
30781 * config/i386/i386.opt Add -mrdpid.
30782 * config/i386/immintrin.h (_rdpid_u32): New.
30784 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
30786 PR rtl-optimization/79541
30787 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
30788 instead of transforming it into USE.
30790 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
30792 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
30793 If HONOR_SNANS (SFmode) force the input to a register.
30794 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
30795 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
30796 an frsp or similar insn.
30798 2017-02-17 Martin Liska <mliska@suse.cz>
30800 PR rtl-optimization/79577
30801 * params.def (selsched-max-sched-times): Increase minimum to 1.
30803 2017-02-17 Martin Liska <mliska@suse.cz>
30805 PR rtl-optimization/79574
30806 * gcse.c (want_to_gcse_p): Prevent integer overflow.
30808 2017-02-17 Martin Liska <mliska@suse.cz>
30810 PR tree-optimization/79529
30811 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
30812 ssa_defined_default_def_p to handle cases which are implicitly
30814 * tree-ssa.c (ssa_defined_default_def_p): New function.
30815 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
30816 which are implicitly defined.
30817 * tree-ssa.h (ssa_defined_default_def_p): Declare.
30819 2017-02-17 Richard Biener <rguenther@suse.de>
30821 PR middle-end/79576
30822 * params.def (max-ssa-name-query-depth): Limit to 10.
30824 2017-02-17 Richard Biener <rguenther@suse.de>
30826 PR tree-optimization/79552
30827 * tree-ssa-structalias.c (visit_loadstore): Properly verify
30830 2017-02-17 Richard Biener <rguenther@suse.de>
30833 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30835 2017-02-17 Marek Polacek <polacek@redhat.com>
30837 PR middle-end/79536
30838 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30839 (fold_negate_expr): New wrapper.
30841 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30843 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30844 Correct terminology and de-emphasize pre-standard behavior.
30846 2017-02-16 Alan Modra <amodra@gmail.com>
30848 PR rtl-optimization/79286
30849 * ira.c (def_dominates_uses): New function.
30850 (update_equiv_regs): Don't create an equivalence for insns that
30851 may trap where the register def does not dominate the use.
30853 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30855 PR rtl-optimization/78127
30856 * lra.c (lra): Call lra_eliminate before finish the loop after
30859 2017-02-16 Richard Biener <rguenther@suse.de>
30861 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30863 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30864 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30865 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30866 (isl_val_int_from_wi): New function.
30867 (extract_affine_gmp): Rename to ...
30868 (extract_affine_wi): ... this, take a widest_int.
30869 (extract_affine_int): Just wrap extract_affine_wi.
30870 (add_param_constraints): Use isl_val_int_from_wi.
30871 (add_loop_constraints): Likewise, and extract_affine_wi.
30873 2017-02-15 Jeff Law <law@redhat.com>
30875 PR middle-end/79521
30876 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30877 ira_init_register_move_cost_if_necessary.
30879 2017-02-15 Martin Sebor <msebor@redhat.com>
30881 PR middle-end/32003
30882 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30883 removed in a prior commit.
30885 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30887 PR tree-optimization/79347
30888 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
30889 counters during peeling.
30891 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
30893 * Makefile.in (site.exp): Remove "set ISLVER".
30895 2017-02-15 Jakub Jelinek <jakub@redhat.com>
30898 * real.c (real_from_integer): Call real_convert even for decimal.
30900 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30903 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30905 2017-02-14 Andrew Pinski <apinski@cavium.com>
30907 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30908 cores and change the partno/implementer to be correct.
30909 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30910 the 'B" as the implementer.
30911 * config/aarch64/aarch64-tune.md: Regenerate.
30913 2017-02-14 Carl Love <cel@us.ibm.com>
30915 * config/rs6000/rs6000.c: Add case statement entry to make the
30916 xvcvuxdsp built-in argument unsigned.
30917 * config/rs6000/vsx.md: Fix the source and return operand types so they
30918 match the instruction definitions from the ISA document. Fix typo
30919 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30922 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30925 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30926 member early_clobber_alts.
30927 * lra-lives.c (reg_early_clobber_p): New.
30928 (process_bb_lives): Use it.
30929 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30930 (debug_operand_data): Initialize early_clobber_alts.
30931 (setup_operand_alternative): Set up early_clobber_alts.
30932 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30933 alternatives to new_insn_reg.
30934 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
30936 (lra_update_insn_regno_info): Pass the new arg.
30938 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30940 PR middle-end/79505
30941 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
30942 (new_oacc_loop_raw): Don't clear already cleared fields.
30945 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
30946 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
30947 _mm512_prefetch_i64gather_ps): New inline functions and macros.
30949 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
30952 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
30954 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
30957 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
30958 the extra instruction to the right place to store 128-bit constant
30961 2017-02-14 Martin Sebor <msebor@redhat.com>
30963 PR middle-end/79448
30964 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
30965 warning for strings of unknown length.
30967 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
30969 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
30971 2017-02-14 Jeff Law <law@redhat.com>
30974 * ira-costs.c (scan_one_insn): Initialize register move costs
30975 for pseudos seen in USE/CLOBBER insns.
30977 PR tree-optimization/79095
30978 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
30979 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
30980 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
30981 if the operands are known to be not equal, then the resulting range
30983 (intersect_ranges): If the new range is ~[0,0] and the old range is
30984 wide, then prefer ~[0,0].
30985 * tree-vrp.c (overflow_comparison_p_1): New function.
30986 (overflow_comparison_p): New function.
30987 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
30988 if NAME is used in an overflow test.
30989 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
30990 overflow check that can be expressed as an equality test, then adjust
30991 ops to be that equality test.
30993 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30995 * config/s390/s390-builtin-types.def: Remove flags argument.
30996 * config/s390/s390.c (s390_init_builtins): Likewise.
30998 2017-02-14 Martin Liska <mliska@suse.cz>
31000 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31001 vector. Fix trailing white spaces.
31003 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31005 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31008 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31010 PR rtl-optimization/68664
31011 * config/arm/arm.c (arm_sched_can_speculate_insn):
31012 New function. Declare prototype.
31013 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31015 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31017 PR rtl-optimization/68664
31018 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31020 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31022 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31024 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31025 max skip bytes for function, loop and jump.
31027 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31029 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31030 ABS_EXPR for gimple dump.
31032 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31035 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31037 PR tree-optimization/79408
31038 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31039 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31040 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31041 also if rhs1 is INTEGER_CST.
31043 2017-02-14 Richard Biener <rguenther@suse.de>
31045 PR middle-end/79432
31046 * tree-into-ssa.c (insert_phi_nodes): When the function can
31047 have abnormal edges rewrite SSA names with broken use-def
31048 dominance out of SSA and register them for PHI insertion.
31050 2017-02-13 Martin Sebor <msebor@redhat.com>
31052 PR middle-end/79496
31053 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31054 clearing info.nowrite flag when snprintf size argument is a range.
31056 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31058 * cprop.c (cprop_jump): Add missing space in string literal.
31059 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31060 (get_constraint_for_component_ref): Likewise.
31061 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31062 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31063 * lra-constraints.c (process_alt_operands): Likewise.
31064 * ipa-inline.c (inline_small_functions): Likewise.
31065 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31066 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31067 * trans-mem.c (diagnose_tm_1_op): Likewise.
31068 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31069 (grid_parallel_clauses_gridifiable): Likewise.
31071 * config/nvptx/mkoffload.c (process): Add space in between
31074 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31075 "MOD4_SSE_REGS" and "ALL_REGS".
31077 * spellcheck.c (test_data): Add , in between "foo" and "food".
31079 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31082 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31083 boundary crossing check and subsequent code generation agree.
31085 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31087 * config/aarch64/aarch64.c (has_memory_op): Delete.
31088 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31091 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31093 PR rtl-optimization/79388
31094 PR rtl-optimization/79450
31095 * combine.c (distribute_notes): When removing TEM_INSN for which
31096 corresponding dest has last value recorded, invalidate that last
31099 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31101 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31102 of explicit '@'. Add missing assembly comment marker on branch costs
31105 2017-02-13 Nathan Sidwell <nathan@acm.org>
31107 * gengtype-lex.l (<in_struct>): Add '/'.
31109 2017-02-13 Martin Liska <mliska@suse.cz>
31112 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31114 2017-02-13 Richard Biener <rguenther@suse.de>
31116 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31118 * configure: Re-generate.
31119 * config.in: Likewise.
31120 * graphite-dependences.c: Simplify as if
31121 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31122 * graphite-isl-ast-to-gimple.c: Likewise.
31123 * graphite-optimize-isl.c: Likewise.
31124 * graphite-poly.c: Likewise.
31125 * graphite-sese-to-poly.c: Likewise.
31126 * graphite.h: Likewise.
31127 * toplev.c: Include isl/version.h and use isl_version () for
31128 printing the ISL version.
31129 * doc/install.texi: Update ISL requirement.
31131 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31133 * doc/standards.texi (Standards): Update reference to
31136 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31138 * doc/extend.texi (Named Address Spaces): sourceware.org now
31140 * doc/install.texi (Binaries): Ditto.
31143 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31145 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31146 standard terminology "stringize"/"stringizing" throughout.
31147 * doc/cppinternals.texi: Likewise.
31149 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31151 * doc/extend.texi: Fix some spelling mistakes and typos.
31152 * doc/invoke.texi: Likewise.
31154 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31157 * params.def (inline-min-speedup) Change from 10 to 8.
31159 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31161 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31164 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31167 * ipa-inline-analysis.c (get_minimal_bb): New function.
31168 (record_modified): Use it.
31169 (remap_edge_change_prob): Handle also ancestor functions.
31171 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31173 * doc/contrib.texi (Contributors): Remove broken link into
31174 the Mauve CVS repository.
31176 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31178 PR middle-end/79454
31179 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31180 result computation whenever lhs doesn't have vector mode, not
31181 just when it has BLKmode.
31183 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31185 * doc/makefile.texi (profiledbootstrap): Refer to the
31186 installation instructions only in textual form.
31188 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31191 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31193 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31195 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31196 (Specific): Update mingw-w64 reference.
31198 (Specific): Remove broken link to Renesas RX processor.
31200 2017-02-10 Richard Biener <rguenther@suse.de>
31202 * toplev.c (process_options): Do not mention obsolete graphite
31203 options when printing sorry message about missing graphite support.
31204 Mention -floop-nest-optimize.
31206 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31208 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31209 (vtst_p16): Likewise.
31210 (vtstq_p8): Likewise.
31211 (vtstq_p16): Likewise.
31213 (vtstq_p64): Likewise.
31214 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31215 (vset_lane_p64): New.
31216 (vsetq_lane_p64): New.
31218 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31220 PR tree-optimization/79411
31221 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31222 stmt operands are SSA_NAMEs used in abnormal phis.
31223 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31226 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31229 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31232 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31234 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31236 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31238 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31242 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31243 not arbitrary TREE_CONSTANT.
31246 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31247 "omp declare target link" attribute unless is_global_var.
31248 * omp-offload.c (find_link_var_op): Likewise.
31250 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31251 Chung-Lin Tang <cltang@codesourcery.com>
31253 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31255 (gimplify_adjust_omp_clauses): Don't delete TILE.
31256 (gimplify_omp_for): Deal with TILE.
31257 * internal-fn.c (expand_GOACC_TILE): New function.
31258 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31260 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31261 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31263 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31264 avoid DIV for outermost collapse var.
31265 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31266 Remove out of date comments, fix whitespace.
31267 * omp-general.c (omp_extract_for_data): Deal with tiling.
31268 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31269 adjust OLF_DIM_BASE value.
31270 (struct omp_for_data): Add tiling field.
31271 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31272 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31273 for auto loops. Remove default auto determining, moved to
31274 oacc_loop_fixed_partitions.
31275 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31276 stmts, add e_mask field.
31277 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31278 (oacc_thread_numbers): Use oacc_dim_call.
31279 (oacc_xform_tile): New.
31280 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31281 (finish_oacc_loop): Adjust for ifns vector.
31282 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31283 add case for GOACC_TILE fns.
31284 (oacc_loop_xform_loop): Delete.
31285 (oacc_loop_process): Iterate over call list directly, and add
31286 handling for GOACC_TILE fns.
31287 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31289 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31290 vector partitioning to outer loops. Assign 2 partitions to loops
31291 when available. Add TILE handling.
31292 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31293 (execite_oacc_device_lower): Process GOACC_TILE fns,
31294 ignore unknown specs.
31295 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31296 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31297 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31299 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31301 * configure.ac (ACX_BUGURL): Update.
31302 * configure: Regenerate.
31304 2017-02-09 Richard Biener <rguenther@suse.de>
31306 PR tree-optimization/69823
31307 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31308 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31310 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31312 * config/arc/arc-c.def: Add __NPS400__ definition.
31313 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31314 (TARGET_NPS400): Define.
31316 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31318 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31320 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31321 pointer, arch_info.
31322 (arc_cpu_types): Fill the arch_info field with a pointer into the
31323 arc_arch_types table.
31324 (arc_selected_cpu): Declare.
31325 * config/arc/arc.c (arc_selected_cpu): Make global.
31326 (arc_selected_arch): Delete.
31327 (arc_base_cpu): Delete.
31328 (arc_override_options): Remove references to deleted variables,
31329 update access to arch information.
31330 (ARC_OPT): Update access to arch information.
31331 (ARC_OPTX): Likewise.
31332 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31333 (TARGET_ARC600): Update access to arch information.
31334 (TARGET_ARC601): Likewise.
31335 (TARGET_ARC700): Likewise.
31336 (TARGET_EM): Likewise.
31337 (TARGET_HS): Likewise.
31338 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31341 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31344 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31345 condition/operands for integer GE/LE/GEU/LEU operations.
31347 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31349 PR translation/79397
31350 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31353 2017-02-08 Martin Jambor <mjambor@suse.cz>
31356 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31357 whether allocation happened.
31358 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31359 nothing was allocated.
31361 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31363 PR tree-optimization/79408
31364 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31365 constant, but SSA_NAME with a known integer range, use the minimum
31366 of that range instead of op1 to determine if modulo can be replaced
31367 with its first operand.
31369 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31371 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31373 2017-02-08 Richard Biener <rguenther@suse.de>
31375 PR tree-optimization/71824
31376 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31377 Check all loops contained in the merged region.
31379 2017-02-07 Andrew Pinski <apinski@cavium.com>
31381 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31383 2017-02-07 Andrew Pinski <apinski@cavium.com>
31385 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
31386 (thunderxt88): Likewise.
31387 (thunderxt81): Disable LSE and change v8.1 to v8.
31388 (thunderxt83): Likewise.
31390 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31391 Richard Biener <rguenther@suse.de>
31393 PR middle-end/79399
31394 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
31395 type from int to size_t.
31396 * ira-costs.c (struct_costs_size): Change type from int to size_t.
31398 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31400 PR rtl-optimization/79386
31401 * cprop.c (bypass_conditional_jumps): Initialize
31402 bypass_last_basic_block already before splitting bbs after
31403 unconditional traps...
31404 (bypass_conditional_jumps): ... rather than here.
31407 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
31408 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
31409 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
31410 fix -masm=intel patterns.
31412 2017-02-07 Richard Biener <rguenther@suse.de>
31414 PR tree-optimization/79256
31415 PR middle-end/79278
31416 * builtins.c (get_object_alignment_2): Use min_align_of_type
31417 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
31418 and ADJUST_FIELD_ALIGN.
31420 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
31422 * doc/tm.texi: Regenerate.
31423 * stor-layout.c (layout_decl): Adjust.
31424 (update_alignment_for_field): Likewise.
31425 (place_field): Likewise.
31426 (min_align_of_type): Likewise.
31427 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
31428 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
31429 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
31430 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
31431 * config/frv/frv.c (frv_adjust_field_align): Likewise.
31432 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
31433 * config/i386/i386.c (x86_field_alignment): Likewise.
31434 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
31435 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
31436 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
31437 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
31438 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
31439 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
31443 2017-01-30 Richard Biener <rguenther@suse.de>
31445 PR tree-optimization/79256
31446 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31447 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31450 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
31452 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
31453 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
31454 builtins to SImode and emit a zero-extend, if necessary.
31456 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31458 * docs/invoke.texi (RISC-V Options): Alphabetize.
31460 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31462 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
31465 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31467 * config/riscv/riscv.c: New file.
31468 * common/config/riscv/riscv-common.c: Likewise.
31469 * config.gcc: Likewise.
31470 * config/riscv/constraints.md: Likewise.
31471 * config/riscv/elf.h: Likewise.
31472 * config/riscv/generic.md: Likewise.
31473 * config/riscv/linux.h: Likewise.
31474 * config/riscv/multilib-generator: Likewise.
31475 * config/riscv/peephole.md: Likewise.
31476 * config/riscv/pic.md: Likewise.
31477 * config/riscv/predicates.md: Likewise.
31478 * config/riscv/riscv-builtins.c: Likewise.
31479 * config/riscv/riscv-c.c: Likewise.
31480 * config/riscv/riscv-ftypes.def: Likewise.
31481 * config/riscv/riscv-modes.def: Likewise.
31482 * config/riscv/riscv-opts.h: Likewise.
31483 * config/riscv/riscv-protos.h: Likewise.
31484 * config/riscv/riscv.h: Likewise.
31485 * config/riscv/riscv.md: Likewise.
31486 * config/riscv/riscv.opt: Likewise.
31487 * config/riscv/sync.md: Likewise.
31488 * config/riscv/t-elf-multilib: Likewise.
31489 * config/riscv/t-linux: Likewise.
31490 * config/riscv/t-linux-multilib: Likewise.
31491 * config/riscv/t-riscv: Likewise.
31492 * configure.ac: Likewise.
31493 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
31494 Waterman as RISC-V maintainers.
31495 * doc/install.texi: Add RISC-V entries.
31496 * doc/invoke.texi: Add RISC-V options section.
31497 * doc/md.texi: Add RISC-V constraints section.
31498 * configure: Regenerated.
31500 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
31503 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
31504 false values to be constant vectors with all 0 or all 1 bits set.
31505 (vcondu<mode><mode>): Likewise.
31506 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
31508 (fpmask_comparison_operator): Update comment.
31509 (vecint_comparison_operator): New predicate.
31510 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
31511 vector conditionals when the true and false values are constant
31512 vectors with all 0 bits or all 1 bits set.
31514 2017-02-06 Martin Sebor <msebor@redhat.com>
31516 PR tree-optimization/79376
31517 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
31519 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
31521 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
31522 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
31523 to simplify split condition.
31525 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31527 * omp-expand.c (oxpand_omp_atomic_fetch_op,
31528 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
31531 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
31533 PR rtl-optimization/68664
31534 * target.def (can_speculate_insn): New hook.
31535 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
31536 * doc/tm.texi: Regenerate.
31537 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
31538 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
31539 (rs6000_sched_can_speculate_insn): New function.
31541 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31543 PR tree-optimization/79284
31544 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
31545 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
31546 vectorizable_mask_load_store, vectorizable_operation,
31547 vect_is_simple_cond, get_same_sized_vectype): Use it instead
31548 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
31549 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
31550 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
31551 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
31552 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31553 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
31554 is_gimple_assign (stmt). Replace another such test with
31555 is_gimple_assign (stmt).
31557 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31560 * config/avr/avr.c (rtl-iter.h): Include it.
31561 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
31562 (avr_legitimate_combined_insn): ...and implementation.
31564 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
31566 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
31567 * config/s390/s390.c (s390_const_operand_ok)
31568 (s390_canonicalize_comparison, s390_extract_part)
31569 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
31570 (s390_contiguous_bitmask_p, s390_rtx_costs)
31571 (legitimize_pic_address): Likewise.
31572 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
31573 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
31574 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
31575 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
31576 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
31578 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31580 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
31581 REGNO($0) == REGNO($1).
31583 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31585 * config/s390/linux.h(SIZE_TYPE): Add comment.
31587 2017-02-06 Julian Brown <julian@codesourcery.com>
31588 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31589 Virendra Pathak <virendra.pathak@broadcom.com>
31591 * config/aarch64/aarch64-cores.def: Change the scheduler
31593 * config/aarch64/aarch64.md: Include thunderx2t99.md.
31594 * config/aarch64/thunderx2t99.md: New file.
31596 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31598 * doc/standards.texi (Go Language): Update link to language
31601 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31603 * tree-eh.c (lower_resx): Sanitize profile.
31604 (cleanup_empty_eh_move_lp): Likewise.
31606 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31609 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
31611 * cfgloopmanip.h (loop_version): Update prototype.
31612 * modulo-sched.c (sms_schedule): Update call of loop_version.
31613 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
31614 * tree-parloops.c (gen_parallel_loop): Likewise.
31615 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
31616 * tree-ssa-loop-split.c (split_loop): Likewise.
31617 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
31618 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
31620 2017-02-05 Martin Liska <mliska@suse.cz>
31623 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
31625 (print_operand_address): Initialize a struct to zero.
31627 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31629 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
31630 garbage collector only in textual form.
31632 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31634 * doc/extend.texi (x86 specific memory model extensions for
31635 transactional memory): Simplify a phrase.
31637 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
31640 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
31641 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
31642 (atomic_storedi_1): Likewise.
31644 2017-02-04 Jakub Jelinek <jakub@redhat.com>
31646 PR tree-optimization/79338
31647 * tree-parloops.c (gather_scalar_reductions): Don't call
31648 vect_analyze_loop_form for loop->inner before destroying loop's
31651 2017-02-03 Martin Sebor <msebor@redhat.com>
31653 PR tree-optimization/79327
31654 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
31655 when precision has resulted in leading zeros.
31656 (format_integer): Adjust the likely counter to assume an unknown
31657 argument that may be zero is non-zero.
31659 2017-02-03 Jason Merrill <jason@redhat.com>
31662 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
31663 avoid copying non-taken branch.
31665 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31667 PR tree-optimization/79340
31668 * tree-vect-loop.c (vectorizable_reduction): Release
31669 vec_defs elements after safe_splicing them into other vectors.
31672 PR tree-optimization/79327
31673 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
31674 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
31676 (format_integer): Use wide_int_to_tree instead of build_int_cst
31677 + to_?hwi. If argmin is NULL, just set argmin and argmax to
31678 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
31679 of shortest and longest sequence.
31681 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
31683 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
31684 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
31686 2017-02-03 Walter Lee <walt@tilera.com>
31689 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
31690 after initial stackframe link reg save.
31691 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
31693 2017-02-03 Jakub Jelinek <jakub@redhat.com>
31696 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
31697 wu for stxssp alternative.
31699 2017-02-03 Martin Sebor <msebor@redhat.com>
31701 PR tree-optimization/79352
31702 * gimple-fold.c (get_range_strlen): Add argument.
31703 (get_range_strlen): Change return type to bool.
31704 (get_maxval_strlen): Pass in a dummy argument.
31705 * gimple-fold.h (get_range_strlen): Change return type to bool.
31706 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
31707 * tree.h (array_at_struct_end_p): Add argument.
31708 * tree.c (array_at_struct_end_p): Handle it.
31710 2017-02-03 Martin Liska <mliska@suse.cz>
31713 * multiple_target.c (create_dispatcher_calls): Redirect edge
31714 from a caller of a dispatcher.
31715 (expand_target_clones): Make the clones local.
31716 (ipa_target_clone): Do both target clones and resolvers.
31717 (ipa_dispatcher_calls): Remove the pass.
31718 (pass_dispatcher_calls::gate): Likewise.
31719 (make_pass_dispatcher_calls): Likewise.
31720 * passes.def (pass_target_clone): Put as very first IPA early
31723 2017-02-03 Martin Liska <mliska@suse.cz>
31725 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
31726 in case of a function with ifunc attribute.
31728 2017-02-03 Martin Liska <mliska@suse.cz>
31730 * cgraph.c (cgraph_node::dump): Dump function version info.
31731 * symtab.c (symtab_node::dump_base): Add missing new line.
31733 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31735 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
31736 (ifcombine_ifandif): Use it.
31738 2017-02-03 Martin Liska <mliska@suse.cz>
31740 * doc/invoke.texi: Document default value for
31741 use-after-scope-direct-emission-threshold.
31743 2017-02-03 Martin Liska <mliska@suse.cz>
31745 PR tree-optimization/79339
31746 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
31747 (format_floating): Likewise.
31749 2017-02-03 Martin Liska <mliska@suse.cz>
31752 * ipa-prop.c (ipa_node_params_t::insert): Remove current
31754 (ipa_node_params_t::remove): Likewise.
31755 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
31756 initialization from removed ipa_node_params_t::insert.
31757 (ipa_node_params::~ipa_node_params): Move from removed
31758 ipa_node_params_t::release.
31759 * symbol-summary.h (symbol_summary::m_released): New member.
31760 Do not release a summary twice. Do not allow to call finalizer
31761 for types of a summary that live in GGC memory.
31763 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31765 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
31768 2017-02-02 Martin Sebor <msebor@redhat.com>
31770 PR middle-end/79275
31771 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
31772 (format_string): Tighten up the range of output for non-constant
31773 strings and correct the expected range for wide non-constant strings.
31775 2017-02-02 Martin Sebor <msebor@redhat.com>
31777 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
31779 PR middle-end/32003
31780 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
31782 (-fdump-tree-@var): Add to index and document how to come up
31783 with pass-specific option and dump file names.
31784 (-fdump-passes): Clarify where to look for output.
31786 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
31788 PR middle-end/77445
31789 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
31790 statistics of the analyzed path; allow threading for speed when
31791 any of BBs along the path are optimized for speed.
31793 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
31795 PR middle-end/78468
31796 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
31797 settings of the virtual registers.
31800 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
31802 * explow.c (get_dynamic_stack_size): Take known alignment of stack
31803 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
31806 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31808 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
31809 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
31811 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31813 * config/s390/s390.md: Add missing comments with the expanded
31815 * config/s390/vector.md: Likewise.
31816 * config/s390/vx-builtins.md: Likewise.
31818 2017-02-02 Jakub Jelinek <jakub@redhat.com>
31821 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
31822 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
31823 conditions on a single line.
31825 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31827 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31828 __S390_VX__ to __VX__.
31830 2017-02-01 Andrew Pinski <apinski@cavium.com>
31832 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31833 stmt_info to record_stmt_cost.
31834 (vect_get_known_peeling_cost): Pass stmt_info if known to
31836 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31837 cpu_vector_cost field into
31838 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31839 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31840 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31841 splitting of scalar_stmt_cost and vec_stmt_cost.
31842 (thunderx_vector_cost): Likewise.
31843 (cortexa57_vector_cost): LIkewise.
31844 (exynosm1_vector_cost): Likewise.
31845 (xgene1_vector_cost): Likewise.
31846 (thunderx2t99_vector_cost): Improve after the splitting of the two
31848 (aarch64_builtin_vectorization_cost): Update for the splitting of
31849 scalar_stmt_cost and vec_stmt_cost.
31851 2017-02-01 Torvald Riegel <triegel@redhat.com>
31852 Richard Henderson <rth@redhat.com>
31854 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31855 conditional on existance of a fast atomic load.
31856 * optabs-query.c (can_atomic_load_p): New function.
31857 * optabs-query.h (can_atomic_load_p): Declare it.
31858 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31859 no fast atomic load is available for the particular size of access.
31860 (expand_atomic_compare_and_swap): Likewise.
31861 (expand_atomic_load): Likewise.
31862 (expand_atomic_store): Likewise.
31863 (expand_atomic_fetch_op): Likewise.
31864 * testsuite/lib/target-supports.exp
31865 (check_effective_target_sync_int_128): Remove x86 because it provides
31866 no fast atomic load.
31867 (check_effective_target_sync_int_128_runtime): Likewise.
31869 2017-02-01 Richard Biener <rguenther@suse.de>
31871 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31872 (graphite_transform_loops): Provide opt-info for optimized nests.
31873 * tree-parloop.c (parallelize_loops): Provide opt-info for
31874 parallelized loops.
31876 2017-02-01 Richard Biener <rguenther@suse.de>
31878 PR middle-end/79315
31879 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31880 was not set before.
31882 2017-02-01 Richard Biener <rguenther@suse.de>
31884 PR tree-optimization/71824
31885 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31886 Verify the loops are valid in the merged SESE region.
31887 (scop_detection::can_represent_loop_1): Check analyzing the
31888 evolution of the number of iterations in the region succeeds.
31890 2017-01-31 Ian Lance Taylor <iant@golang.org>
31892 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
31893 REG_ARGS_SIZE note to 32-bit push insns and call insn.
31895 2017-01-31 David Malcolm <dmalcolm@redhat.com>
31897 PR preprocessor/79210
31898 * input.c (get_substring_ranges_for_loc): Replace line_width
31899 assertion with error-handling.
31901 2017-01-31 Richard Biener <rguenther@suse.de>
31903 PR tree-optimization/77318
31904 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31905 (create_pw_aff_from_tree): Take loop parameter.
31906 (add_condition_to_pbb): Pass loop of the condition to
31907 create_pw_aff_from_tree.
31909 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31911 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31912 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31914 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31918 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31920 (convert_int_to_float128): Likewise.
31921 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31922 (convert_int_to_float128): Likewise.
31923 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31924 (UNSPEC_IEEE128_CONVERT): Likewise.
31925 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31926 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31927 Use local variables for IBM extended format.
31928 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31929 (fix_trunc<mode>si2_fprs): Likewise.
31930 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31931 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31932 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31933 to know that we can now have integers of all sizes in vector
31935 (fix<uns>_<mode>di2_hw): Likewise.
31936 (float<uns>_<mode>si2_hw): Likewise.
31937 (fix_<mode>si2_hw): Likewise.
31938 (fixuns_<mode>si2_hw): Likewise.
31939 (float<uns>_<mode>di2_hw): Likewise.
31940 (float_<mode>di2_hw): Likewise.
31941 (float_<mode>si2_hw): Likewise.
31942 (floatuns_<mode>di2_hw): Likewise.
31943 (floatuns_<mode>si2_hw): Likewise.
31944 (xscvqp<su>wz_<mode>): Delete, no longer used.
31945 (xscvqp<su>dz_<mode>): Likewise.
31946 (xscv<su>dqp_<mode>): Likewise.
31947 (ieee128_mfvsrd_64bit): Likewise.
31948 (ieee128_mfvsrd_32bit): Likewise.
31949 (ieee128_mfvsrwz): Likewise.
31950 (ieee128_mtvsrw): Likewise.
31951 (ieee128_mtvsrd_64bit): Likewise.
31952 (ieee128_mtvsrd_32bit): Likewise.
31954 2017-01-31 Martin Liska <mliska@suse.cz>
31957 * ipa-prop.c (ipa_free_all_node_params): Call release method
31958 instead of ~sumbol_summary to not to trigger double times
31961 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
31963 PR tree-optimization/71691
31964 * bitmap.h (class auto_bitmap): New.
31965 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
31966 is_maybe_undefined instead of ssa_undefined_value_p.
31968 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31970 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31971 __S390_ARCH_LEVEL__ to __ARCH__.
31973 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31975 PR tree-optimization/79267
31976 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
31977 if should_remove_lhs_p is true.
31979 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31982 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31983 (add_alignment_attribute): New.
31984 (base_type_die): Add alignment attribute.
31985 (subrange_type_die): Likewise.
31986 (modified_type_die): Likewise.
31987 (gen_array_type_die): Likewise.
31988 (gen_descr_array_type_die: Likewise.
31989 (gen_enumeration_type_die): Likewise.
31990 (gen_subprogram_die): Likewise.
31991 (gen_variable_die): Likewise.
31992 (gen_field_die): Likewise.
31993 (gen_ptr_to_mbr_type_die): Likewise.
31994 (gen_struct_or_union_type_die): Likewise.
31995 (gen_subroutine_type_die): Likewise.
31996 (gen_typedef_die): Likewise.
31997 (base_type_cmp): Compare alignment attribute.
31999 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32002 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32003 (setb_unsigned) New pattern for setb with CCUNS.
32004 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32005 subfc./subfe sequence to avoid overflow problems. Generate a
32006 shorter sequence with cmpld/setb for power9.
32007 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32008 for generating subfc. instruction.
32009 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32010 now uses this instruction.
32012 2017-01-30 Ian Lance Taylor <iant@google.com>
32015 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32016 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32018 2017-01-30 Martin Sebor <msebor@redhat.com>
32020 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32021 Move constant to the right of a relational operator.
32022 (get_mpfr_format_length, format_character, format_string): Ditto.
32023 (should_warn_p, maybe_warn): Same.
32025 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32027 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32030 * asan.c (get_translation_unit_decl): Remove function.
32031 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32033 2017-01-30 Martin Liska <mliska@suse.cz>
32035 PR gcov-profile/79259
32036 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32037 -fprofile-generate.
32039 2017-01-30 Martin Liska <mliska@suse.cz>
32042 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32043 Initialize variables with NULL value.
32045 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32048 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32050 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32052 2017-01-30 Richard Biener <rguenther@suse.de>
32054 PR tree-optimization/79276
32055 * tree-vrp.c (process_assert_insertions): Properly adjust common
32056 when removing a duplicate.
32058 2017-01-30 Richard Biener <rguenther@suse.de>
32060 PR tree-optimization/79256
32061 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32062 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32064 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32066 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32069 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32070 ("*r<noxa>sbg_<mode>_sll_bitmask")
32071 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32072 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32073 Use contiguous_bitmask_nowrap_operand.
32075 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32078 * config/rs6000/altivec.h (vec_xl): Revise #define.
32079 (vec_xst): Likewise.
32081 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32083 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32085 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32087 PR rtl-optimization/79194
32088 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32089 traps before call to bypass_conditional_jumps.
32091 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32093 PR tree-optimization/71374
32094 * lra-constraints.c (check_conflict_input_operands): New.
32095 (match_reload): Use it.
32097 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32100 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32101 account to calculate conflict_set.
32103 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32105 PR rtl-optimization/78559
32106 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32107 other_insn in combine.
32109 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32111 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32112 uint16_type_node for BT_UINT16.
32114 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32116 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32117 "RTL Tests" to menu.
32118 (GIMPLE Tests): New node.
32119 (RTL Tests): New node.
32121 2017-01-27 Richard Biener <rguenther@suse.de>
32123 PR tree-optimization/79245
32124 * tree-loop-distribution.c (distribute_loop): Apply cost
32125 modeling also to detected patterns.
32127 2017-01-27 Richard Biener <rguenther@suse.de>
32129 PR tree-optimization/71433
32130 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32131 (compare_assert_loc): New function.
32132 (process_assert_insertions): Sort and optimize assert locations
32133 to remove duplicates and push down identical assertions on
32134 edges to their destination block.
32136 2017-01-27 Richard Biener <rguenther@suse.de>
32138 PR tree-optimization/79244
32139 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32140 out SSA names even if abnormal.
32142 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32144 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32145 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32146 instead of MPFR_RNDN.
32148 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32151 * arm.c (arm_option_override): Don't call build_target_option_node
32152 until after doing all option overrides.
32153 (arm_valid_target_attribute_tree): Likewise.
32155 2017-01-27 Martin Liska <mliska@suse.cz>
32157 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32158 for {cd}tors and C++ {cd}tors.
32160 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32162 * config/s390/s390.md ("*setmem_long_and")
32163 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32165 2017-01-26 Martin Sebor <msebor@redhat.com>
32167 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32170 2017-01-26 Martin Sebor <msebor@redhat.com>
32172 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32173 HAVE_DFmode before using XFmode or DFmode.
32174 (parse_directive): Avoid using the z length modifier to avoid
32175 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32177 PR middle-end/78703
32178 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32179 to accept adjustment as an array.
32180 (get_int_range): New function.
32181 (struct directive): Make width and prec arrays.
32182 (directive::set_width, directive::set_precision): Call get_int_range.
32183 (format_integer, format_floating): Handle width and precision ranges.
32184 (format_string, parse_directive): Same.
32186 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32189 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32190 comdat_type_p set, just clone them, but keep the children in the
32194 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32195 which have direct callers with -fvar-tracking-assignments enabled
32197 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32198 inside of type units.
32200 2017-01-26 Martin Sebor <msebor@redhat.com>
32202 PR middle-end/78703
32203 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32205 (struct format_result): Replace number_chars, number_chars_min,
32206 and number_chars_max with a single member of struct result_range.
32208 (format_result::operator+=): Adjust.
32209 (struct fmtresult): Remove bounded. Handle likely and unlikely
32211 (fmtresult::adjust_for_width_or_precision): New function.
32212 (fmtresult:type_max_digits): New function.
32213 (bytes_remaining): Handle likely and unlikely counters.
32214 (min_bytes_remaining): Remove.
32215 (format_percent): Simplify.
32216 (format_integer, format_floating): Set likely and unlikely counters.
32217 (get_string_length, format_character, format_string): Same.
32218 (format_plain, should_warn_p): New function.
32219 (maybe_warn): Call should_warn_p. Update diagnostic messages
32220 and handle those for all directives, including plain strings.
32221 (format_directive): Handle likely and unlikely counters.
32222 Remove unnecessary quoting from diagnostics. Add an informational
32224 (add_bytes): Remove.
32225 (pass_sprintf_length::compute_format_length): Simplify.
32226 (try_substitute_return_value): Handle likely and unlikely counters.
32228 2017-01-26 Carl Love <cel@us.ibm.com>
32230 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32231 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32233 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32236 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32237 endianess for subregs into account.
32238 * lra-constraints.c (lra_constraints): Do risky transformations
32239 always on the first iteration.
32240 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32241 last_call_used_reg_set.
32242 (process_bb_lives): Define and use last_call_used_reg_set.
32243 * lra.c (lra): Always continue after lra_constraints on the first
32246 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32248 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32250 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32252 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32254 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32255 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32256 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32257 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32258 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32259 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32260 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32261 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32262 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32264 2017-01-26 Marek Polacek <polacek@redhat.com>
32267 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32268 for the third operand.
32270 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32272 PR middle-end/79236
32273 * omp-low.c (struct omp_context): Add simt_stmt field.
32274 (scan_omp_for): Return omp_context *.
32275 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32276 context to the _simt_ SIMD stmt.
32277 (lower_omp_for): For combined SIMD with sibling _simt_
32278 SIMD, make sure to use the same decls in _looptemp_
32279 clauses as in the sibling.
32281 2017-01-26 David Sherwood <david.sherwood@arm.com>
32283 PR middle-end/79212
32284 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32287 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32290 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32291 emit fld b; fld a; if possible.
32293 * brig-builtins.def: Update copyright years.
32294 * config/arm/arm_acle_builtins.def: Update copyright years.
32296 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32299 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32300 constraint instead of o for the stxsd instruction.
32302 2017-01-25 Carl Love <cel@us.ibm.com>
32304 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32305 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32307 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32309 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32311 2017-01-25 Richard Biener <rguenther@suse.de>
32313 PR tree-optimization/69264
32314 * target.def (vector_alignment_reachable): Improve documentation.
32315 * doc/tm.texi: Regenerate.
32316 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32318 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32319 earlier changes with respect to TYPE_USER_ALIGN.
32320 (vector_alignment_reachable_p): Likewise. Improve dumping.
32322 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32325 * config/arm/arm.md (xordi3): Force constant operand into a register
32328 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32330 * doc/invoke.texi (-fstore-merging): Correct default optimization
32331 levels at which it is enabled.
32332 (-O): Move -fstore-merging from list to...
32335 2017-01-25 Richard Biener <rguenther@suse.de>
32338 * omp-expand.c: Include debug.h.
32339 (expand_omp_taskreg): Make sure to generate early debug before
32340 outlining anything from a function.
32341 (expand_omp_target): Likewise.
32342 (grid_expand_target_grid_body): Likewise.
32344 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32347 * asan.c (get_translation_unit_decl): New function.
32348 (asan_add_global): Extract modules file name from globals
32349 TRANSLATION_UNIT_DECL name.
32351 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32354 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32355 for long calls with APCS frame and VFP.
32357 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32359 * cfg.c (original_copy_tables_initialized_p): New function.
32360 * cfg.h (original_copy_tables_initialized_p): New decl.
32361 * cfgrtl.c (relink_block_chain): Guard the call to
32362 free_original_copy_tables with a call to
32363 original_copy_tables_initialized_p.
32364 * cgraph.h (symtab_node::native_rtl_p): New decl.
32365 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32366 (symtab_node::needed_p): Don't assert for early assembly output
32367 for __RTL functions.
32368 (cgraph_node::finalize_function): Set "force_output" for __RTL
32370 (cgraph_node::analyze): Bail out early for __RTL functions.
32371 (analyze_functions): Update assertion to support __RTL functions.
32372 (cgraph_node::expand): Bail out early for __RTL functions.
32373 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32375 * function.h (struct function): Update comment for field
32377 * gimple-expr.c: Include "tree-pass.h".
32378 (gimple_has_body_p): Return false for __RTL functions.
32379 * Makefile.in (OBJS): Add run-rtl-passes.o.
32380 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32382 (gcc::pass_manager::get_clean_slate): New accessor.
32383 * passes.c: Include "insn-addr.h".
32384 (should_skip_pass_p): Add logging. Update logic for running
32385 "expand" to be compatible with both __GIMPLE and __RTL. Guard
32386 property-provider override so it is only done for gimple passes.
32388 (skip_pass): New function.
32389 (execute_one_pass): Call skip_pass when skipping passes.
32390 * read-md.c (md_reader::read_char): Support filtering
32391 the input to a subset of line numbers.
32392 (md_reader::md_reader): Initialize fields
32393 m_first_line and m_last_line.
32394 (md_reader::read_file_fragment): New function.
32395 * read-md.h (md_reader::read_file_fragment): New decl.
32396 (md_reader::m_first_line): New field.
32397 (md_reader::m_last_line): New field.
32398 * read-rtl-function.c (function_reader::create_function): Only
32399 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
32400 curr_properties. Set DECL_INITIAL to a dummy block.
32401 (read_rtl_function_body_from_file_range): New function.
32402 * read-rtl-function.h (read_rtl_function_body_from_file_range):
32404 * run-rtl-passes.c: New file.
32405 * run-rtl-passes.h: New file.
32407 2017-01-24 Jeff Law <law@redhat.com>
32409 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
32412 2017-01-24 Bin Cheng <bin.cheng@arm.com>
32414 PR tree-optimization/79159
32415 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
32416 (record_nonwrapping_iv): Improve boundary using above function if no
32417 value range information.
32419 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
32420 Martin Jambor <mjambor@suse.cz>
32422 * brig-builtins.def: New file.
32423 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
32424 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
32425 (DEF_HSAIL_SAT_BUILTIN): Likewise.
32426 (DEF_HSAIL_INTR_BUILTIN): Likewise.
32427 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
32428 * builtin-types.def (BT_INT8): New.
32429 (BT_INT16): Likewise.
32430 (BT_UINT8): Likewise.
32431 (BT_UINT16): Likewise.
32432 (BT_FN_ULONG): Likewise.
32433 (BT_FN_UINT_INT): Likewise.
32434 (BT_FN_UINT_ULONG): Likewise.
32435 (BT_FN_UINT_LONG): Likewise.
32436 (BT_FN_UINT_PTR): Likewise.
32437 (BT_FN_ULONG_PTR): Likewise.
32438 (BT_FN_INT8_FLOAT): Likewise.
32439 (BT_FN_INT16_FLOAT): Likewise.
32440 (BT_FN_UINT32_FLOAT): Likewise.
32441 (BT_FN_UINT16_FLOAT): Likewise.
32442 (BT_FN_UINT8_FLOAT): Likewise.
32443 (BT_FN_UINT64_FLOAT): Likewise.
32444 (BT_FN_UINT16_UINT32): Likewise.
32445 (BT_FN_UINT32_UINT16): Likewise.
32446 (BT_FN_UINT16_UINT16_UINT16): Likewise.
32447 (BT_FN_INT_PTR_INT): Likewise.
32448 (BT_FN_UINT_PTR_UINT): Likewise.
32449 (BT_FN_LONG_PTR_LONG): Likewise.
32450 (BT_FN_ULONG_PTR_ULONG): Likewise.
32451 (BT_FN_VOID_UINT64_UINT64): Likewise.
32452 (BT_FN_UINT8_UINT8_UINT8): Likewise.
32453 (BT_FN_INT8_INT8_INT8): Likewise.
32454 (BT_FN_INT16_INT16_INT16): Likewise.
32455 (BT_FN_INT_INT_INT): Likewise.
32456 (BT_FN_UINT_FLOAT_UINT): Likewise.
32457 (BT_FN_FLOAT_UINT_UINT): Likewise.
32458 (BT_FN_ULONG_UINT_UINT): Likewise.
32459 (BT_FN_ULONG_UINT_PTR): Likewise.
32460 (BT_FN_ULONG_ULONG_ULONG): Likewise.
32461 (BT_FN_UINT_UINT_UINT): Likewise.
32462 (BT_FN_VOID_UINT_PTR): Likewise.
32463 (BT_FN_UINT_UINT_PTR: Likewise.
32464 (BT_FN_UINT32_UINT64_PTR): Likewise.
32465 (BT_FN_INT_INT_UINT_UINT): Likewise.
32466 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
32467 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
32468 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
32469 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
32470 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
32471 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
32472 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
32473 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
32474 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
32475 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
32476 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
32477 * doc/frontends.texi: List BRIG FE.
32478 * doc/install.texi (Testing): Add BRIG tesring requirements.
32479 * doc/invoke.texi (Overall Options): Mention BRIG.
32480 * doc/standards.texi (Standards): Doucment BRIG HSA version.
32482 2017-01-24 Richard Biener <rguenther@suse.de>
32484 PR translation/79208
32485 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
32487 2017-01-24 Martin Jambor <mjambor@suse.cz>
32490 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
32491 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
32492 and known_contexts.
32494 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
32496 PR middle-end/79123
32497 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
32498 casts from signed to unsigned really don't have a range.
32500 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
32502 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
32503 GMP_RNDx for compatiblity.
32505 2017-01-24 Martin Liska <mliska@suse.cz>
32508 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
32509 that would prevent us to call alloca with -1 as argument.
32511 2017-01-24 Jakub Jelinek <jakub@redhat.com>
32513 * dwarf2out.c (output_compilation_unit_header, output_file_names):
32514 Avoid -Wformat-security warning.
32516 2017-01-23 Andrew Pinski <apinski@cavium.com>
32518 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
32521 2017-01-23 Martin Sebor <msebor@redhat.com>
32523 PR middle-end/78703
32524 * gimple-ssa-sprintf.c (warn_level): New global.
32525 (format_integer): Use it here and throughout the rest of the file.
32526 Use the same switch to compute sign as base.
32527 (maybe_warn): New function.
32528 (format_directive): Factor out warnings into maybe_warn.
32529 Add debugging output. Use warn_level.
32530 (add_bytes): Use warn_level.
32531 (pass_sprintf_length::compute_format_length): Add debugging output.
32532 (try_substitute_return_value): Same.
32533 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
32535 PR middle-end/78703
32536 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
32537 (struct fmtresult, format_integer, format_floating): Adjust.
32538 (fmtresult::fmtresult): Set max correctly in two argument ctor.
32539 (get_string_length, format_string,format_directive): Same.
32540 (pass_sprintf_length::compute_format_length): Same.
32541 (try_substitute_return_value): Simplify slightly.
32543 PR middle-end/78703
32544 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
32545 (fmtresult::operator+=): Outlined.
32546 (struct fmtresult): Add ctors.
32547 (struct conversion_spec): Rename...
32548 (struct directive): ...to this. Add and remove data members.
32549 (directive::set_width, directive::set_precision): New functions.
32550 (format_percent): Use fmtresult ctor.
32551 (get_width_and_precision): Remove.
32552 (format_integer): Make naming changes. Avoid computing width and
32554 (format_floating): Same. Adjust indentation.
32555 (format_character, format_none): New functions.
32556 (format_string): Moved character handling to format_character.
32557 (format_directive): Remove arguments, change return type.
32558 (parse_directive): New function.
32559 (pass_sprintf_length::compute_format_length): Move directive
32560 parsing to parse_directive.
32562 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32564 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
32565 (assign_assembler_name_if_needed): ... this.
32566 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
32567 (assign_assembler_name_if_needed): ... this.
32568 (free_lang_data_in_cgraph): Adjust callers.
32569 * cgraphunit.c (cgraph_node::analyze): Likewise.
32570 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
32573 2017-01-23 Richard Biener <rguenther@suse.de>
32575 PR tree-optimization/79088
32576 PR tree-optimization/79188
32577 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
32578 resetting loop bounds after last path deletion. Reset loop
32579 bounds of the target loop, make code match the comments.
32580 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
32581 Make sure loops need no fixups.
32583 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32585 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
32586 exponent support with double type for first argument.
32587 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
32588 type returned by __builtin_vec_extract_sig,
32589 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
32590 functions from "vector int" to "vector unsigned int" or from
32591 "vector long long int" to "vector unsigned long long int".
32592 Changed type returned by __builtin_vec_extract_exp,
32593 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
32594 functions from "vector int" to "vector unsigned int" or from
32595 "vector long long int" to "vector unsigned long long int".
32596 Changed return type of __builtin_vec_test_data_class,
32597 __builtin_vec_test_data_class_sp, and
32598 __builtin_vec_test_data_class_dp from "vector int" to
32599 "vector bool int" or from "vector long long int" to "vector bool
32600 long long int" and changed second argument type from "unsigned
32601 int" to "int". Added new overloaded function forms "vector float
32602 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
32603 "vector float __builtin_vec_insert_exp_sp (vector float, vector
32604 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
32605 double, vector unsigned long long int)" and "vector double
32606 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
32607 long int)". Changed return type of
32608 __builtin_scalar_test_data_class and
32609 __builtin_scalar_test_data_class_sp and
32610 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
32611 int" and changed second argument from "unsigned int" to "int".
32612 Changed type returned by __builtin_scalar_test_neg,
32613 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
32614 from "int" to "bool int". Added new overloaded function form
32615 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
32616 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
32617 exponent double-precision with floating point first argument.
32618 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
32619 documentation of scalar_test_data_class, scalar_test_neg,
32620 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
32621 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
32622 vec_test_data_class built-in functions to reflect refinements in
32623 their type signatures.
32625 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
32627 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
32629 (aarch64_elf_asm_destructor): Likewise.
32631 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
32633 PR rtl-optimization/78634
32634 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
32635 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
32636 * ifcvt.c (noce_try_cmove): Add missing cost check.
32638 PR rtl-optimization/71724
32639 * combine.c (if_then_else_cond): Look for situations where it is
32640 beneficial to undo the work of one of the recursive calls.
32642 2017-01-23 Bin Cheng <bin.cheng@arm.com>
32644 PR tree-optimization/70754
32645 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
32646 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
32647 combined stmt before it if not NULL.
32648 (combine_chains): Process refs reversely and compute dominance point
32651 2017-01-23 Martin Liska <mliska@suse.cz>
32653 PR tree-optimization/79196
32654 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
32655 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
32657 (strlen_optimize_stmt): Call the renamed function.
32659 2017-01-23 Michael Matz <matz@suse.de>
32661 PR tree-optimization/78384
32662 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
32664 2017-01-23 Richard Biener <rguenther@suse.de>
32666 PR tree-optimization/79186
32667 * tree-vrp.c (register_new_assert_for): Make sure we've seen
32668 both incoming edges before moving an assert.
32670 2017-01-23 Martin Jambor <mjambor@suse.cz>
32672 * ipa-prop.c (load_from_param_1): Removed.
32673 (load_from_unmodified_param): Bits from load_from_param_1 put back
32675 (load_from_param): Removed.
32676 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
32677 with stmt. Reverted back to use of load_from_unmodified_param.
32679 2017-01-23 Martin Jambor <mjambor@suse.cz>
32682 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
32683 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
32684 field a pointer to garbage collected vector, mark lattices and
32685 ipcp_orig_node with GTY((skip)).
32686 (ipa_get_param_count): Adjust to descriptors being a pointer.
32687 (ipa_get_param): Likewise.
32688 (ipa_get_type): Likewise.
32689 (ipa_get_param_move_cost): Likewise.
32690 (ipa_set_param_used): Likewise.
32691 (ipa_get_controlled_uses): Likewise.
32692 (ipa_set_controlled_uses): Likewise.
32693 (ipa_is_param_used): Likewise.
32694 (ipa_node_params_t): Move into garbage collector. New methods insert
32696 (ipa_node_params_sum): Annotate wth GTY(()).
32697 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
32699 (ipa_load_from_parm_agg): Adjust declaration.
32700 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
32701 * ipa-profile.c (ipa_profile): Likewise.
32702 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
32703 (ipa_populate_param_decls): Make descriptors parameter garbage
32705 (ipa_dump_param): Adjust to descriptors being a pointer.
32706 (ipa_alloc_node_params): Likewise.
32707 (ipa_initialize_node_params): Likewise.
32708 (load_from_param_1): Make descriptors parameter garbage collected.
32709 (load_from_unmodified_param): Likewise.
32710 (load_from_param): Likewise.
32711 (ipa_load_from_parm_agg): Likewise.
32712 (ipa_node_params::~ipa_node_params): Removed.
32713 (ipa_free_all_node_params): Remove call to delete operator.
32714 (ipa_node_params_t::insert): New.
32715 (ipa_node_params_t::remove): Likewise.
32716 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
32717 copy known_csts and known_contexts vectors.
32718 (ipa_read_node_info): Adjust to descriptors being a pointer.
32719 (ipcp_modif_dom_walker): Make m_descriptors field garbage
32721 (ipcp_transform_function): Make descriptors variable garbage
32724 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
32726 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
32727 * config/i386/avx512dqintrin.h: Ditto.
32728 * config/i386/avx512fintrin.h: Ditto.
32729 * config/i386/i386.c: Handle new builtins.
32730 * config/i386/i386-builtin.def: Add new builtins.
32731 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
32732 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
32734 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32735 Martin Liska <mliska@suse.cz>
32737 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
32738 * asan.c (asan_expand_poison_ifn): Support stores and use
32739 appropriate ASAN report function.
32740 * internal-fn.c (expand_ASAN_POISON_USE): New function.
32741 * internal-fn.def (ASAN_POISON_USE): Declare.
32742 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
32743 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
32744 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
32745 ASAN_POISON calls w/o LHS.
32746 * tree-ssa.c (execute_update_addresses_taken): Create clobber
32747 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
32748 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
32749 * gimplify.c (asan_poison_variables): Add attribute
32750 use_after_scope_memory to variables that really needs to live
32752 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
32753 having the attribute.
32755 2017-01-23 Martin Liska <mliska@suse.cz>
32757 * asan.c (create_asan_shadow_var): New function.
32758 (asan_expand_poison_ifn): Likewise.
32759 * asan.h (asan_expand_poison_ifn): New declaration.
32760 * internal-fn.c (expand_ASAN_POISON): Likewise.
32761 * internal-fn.def (ASAN_POISON): New builtin.
32762 * sanopt.c (pass_sanopt::execute): Expand
32763 asan_expand_poison_ifn.
32764 * tree-inline.c (copy_decl_for_dup_finish): Make function
32766 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
32767 * tree-ssa.c (is_asan_mark_p): New function.
32768 (execute_update_addresses_taken): Rewrite local variables
32769 (identified just by use-after-scope as addressable) into SSA.
32771 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32773 * doc/install.texi (Specific): opensource.apple.com uses https
32774 now. Remove trailing slash.
32776 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
32778 * README.Portability: Remove note on an Irix compatibility issue.
32780 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
32782 * gcov.c (INCLUDE_ALGORITHM): Define.
32783 (INCLUDE_VECTOR): Define.
32784 No longer include <vector> and <algorithm> directly.
32786 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
32788 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
32790 * doc/invoke.texi (Code Gen Options): Ditto.
32792 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
32795 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
32797 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
32799 rtl-optimization/79125
32800 * cprop.c (local_cprop_pass): Handle cases where we make an
32801 unconditional trap.
32803 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
32807 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
32808 read from, for big endian.
32810 2017-01-20 Jiong Wang <jiong.wang@arm.com>
32812 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
32813 register pauth builtins for LP64 only.
32815 2017-01-20 Marek Polacek <polacek@redhat.com>
32818 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
32821 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32823 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
32825 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
32826 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
32827 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
32829 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32832 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32833 in big-endian mode when they are not a single duplicated value.
32835 2017-01-20 Richard Biener <rguenther@suse.de>
32837 * BASE-VER: Bump to 7.0.1.
32839 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32841 * omp-low.c (omplow_simd_context): New struct. Use it...
32842 (lower_rec_simd_input_clauses): ...here and...
32843 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32844 references to idx, lane, max_vf, is_simt.
32846 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32848 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32851 2017-01-20 Martin Jambor <mjambor@suse.cz>
32853 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32854 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32856 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32857 (GTFILES): Rename hsa.c to hsa-common.c.
32858 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32859 * hsa-dump.c: Likewise.
32860 * hsa-gen.c: Likewise.
32861 * hsa-regalloc.c: Likewise.
32862 * ipa-hsa.c: Likewise.
32863 * omp-expand.c: Likewise.
32864 * omp-low.c: Likewise.
32865 * toplev.c: Likewise.
32867 2017-01-20 Marek Polacek <polacek@redhat.com>
32870 * doc/invoke.texi: Document -Wduplicated-branches.
32871 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32872 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32873 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32874 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32875 return 0 only when not OEP_LEXICOGRAPHIC.
32876 (fold_build_cleanup_point_expr): Use the expression
32877 location when building CLEANUP_POINT_EXPR.
32878 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32879 * tree.c (add_expr): Handle error_mark_node.
32881 2017-01-20 Martin Liska <mliska@suse.cz>
32884 * tree-profile.c (init_ic_make_global_vars): Do not call
32886 (gimple_init_gcov_profiler): Likewise.
32888 2017-01-20 Martin Liska <mliska@suse.cz>
32891 * cgraph.h (maybe_create_reference): Remove argument and
32893 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
32895 * ipa-cp.c (create_specialized_node): Likewise.
32896 * symtab.c (symtab_node::maybe_create_reference): Handle
32897 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
32899 2017-01-20 Martin Liska <mliska@suse.cz>
32901 * read-rtl-function.c (function_reader::create_function): Use
32902 build_decl instread of build_decl_stat.
32904 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32906 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32907 * config/i386/avx512dqintrin.h: Ditto.
32908 * config/i386/avx512fintrin.h: Ditto.
32909 * config/i386/i386-builtin-types.def: Add new types.
32910 * config/i386/i386.c: Handle new types.
32911 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32912 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32913 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32914 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32915 (__builtin_ia32_kshiftridi): New.
32916 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32918 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32922 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32923 define to rs6000_init_stack_protect_guard.
32924 (rs6000_init_stack_protect_guard): New function.
32926 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32927 Yunqiang Su <yunqiang.su@imgtec.com>
32929 * config.gcc (supported_defaults): Add madd4.
32930 (with_madd4): Add validation.
32931 (all_defaults): Add madd4.
32932 * config/mips/mips.opt (mmadd4): New option.
32933 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32935 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32937 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
32938 (ISA_HAS_FUSED_MADD4): Likewise.
32939 * doc/invoke.texi (-mmadd4): Document the new option.
32940 * doc/install.texi (--with-madd4): Document the new option.
32942 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32944 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
32945 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
32946 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
32947 (aarch64_init_pauth_hint_builtins): New.
32948 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
32949 (aarch64_expand_builtin): Expand new builtins.
32951 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32953 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
32954 * combine-stack-adj.c (no_unhandled_cfa): Handle
32955 REG_CFA_TOGGLE_RA_MANGLE.
32956 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
32957 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
32958 info for return address signing.
32959 (aarch64_expand_epilogue): Likewise.
32961 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32963 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
32964 * config/aarch64/aarch64-protos.h
32965 (aarch64_return_address_signing_enabled): New declaration.
32966 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
32968 (aarch64_expand_prologue): Sign return address before it's pushed onto
32970 (aarch64_expand_epilogue): Authenticate return address fetched from
32972 (aarch64_override_options): Sanity check for ILP32 and ISA level.
32973 (aarch64_attributes): New function attributes for "sign-return-address".
32974 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
32975 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
32976 ("*do_return"): Generate combined instructions according to key index.
32977 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
32978 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
32980 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
32981 * config/aarch64/aarch64.opt (msign-return-address=): New.
32982 * doc/extend.texi (AArch64 Function Attributes): Documents
32983 "sign-return-address=".
32984 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32986 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32988 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
32989 overall option summary.
32991 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32993 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
32994 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
32995 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
32996 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
32998 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33000 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33001 -mpower9-minmax by default for -mcpu=power9.
33002 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33003 128-bit floating point.
33005 2017-01-20 Alan Modra <amodra@gmail.com>
33007 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33008 optimizing for size.
33010 2017-01-20 Alan Modra <amodra@gmail.com>
33013 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33014 for strcmp and strncmp from corresponding builtin decl.
33016 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33018 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33019 instead of i386/rtems-64.h.
33020 * config/i386/rtems-64.h: Remove.
33022 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33026 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33028 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33030 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33032 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33033 Change int to HOST_WIDE_INT.
33034 * config/aarch64/aarch64-protos.h
33035 (aarch64_simd_gen_const_vector_dup): Likewise.
33036 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33038 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33040 * langhooks-def.h (lhd_type_for_size): New decl.
33041 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33042 * langhooks.c (lhd_type_for_size): New function, taken from
33045 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33047 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33048 define_bypass for CR latency.
33049 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33050 (power9-alu2): Add define_bypass for CR latency.
33052 (power9-mul): Update insn latency.
33053 (power9-mul-compare): Update insn latency, bypass latency and remove
33056 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33058 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33060 * config/aarch64/aarch64.md
33061 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33062 aarch64_nopcrelative_literal_loads.
33063 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33065 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33067 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33068 TARGET_LOONGSON_3A.
33069 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33071 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33074 * config.gcc (supported_defaults): Add lxc1-sxc1.
33075 (with_lxc1_sxc1): Add validation.
33076 (all_defaults): Add lxc1-sxc1.
33077 * config/mips/mips.opt (mlxc1-sxc1): New option.
33078 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33080 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33081 __mips_no_lxc1_sxc1.
33082 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33083 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33084 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33086 2017-01-19 Richard Biener <rguenther@suse.de>
33088 PR tree-optimization/72488
33089 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33090 sure to restore SSA info.
33091 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33093 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33095 PR rtl-optimization/79121
33096 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33097 of the inner type when shifting an extended value.
33099 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33102 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33103 interposable aliases.
33105 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33108 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33109 Use the evmergelohi instruction.
33110 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33111 (mov_sitf_e500_subreg8_2_be): Likewise.
33112 (mov_sitf_e500_subreg12_2_le): Likewise.
33113 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33114 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33115 (mov_sitf_e500_subreg8_2_le): Likewise.
33116 (mov_sitf_e500_subreg12_2_be): Likewise.
33118 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33120 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33121 attribute from vecsimple to vecperm.
33122 (altivec_vbpermq2): Likewise.
33124 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33127 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33129 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33130 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33131 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33132 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33133 case where N arg is SIZE_MAX.
33134 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33135 (cmpstrsi): Add pattern.
33137 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33139 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33140 __builtin_vec_revb builtins.
33141 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33142 built-in functions to support generation of the ISA 3.0 XXBR<x>
33143 vector byte reverse instructions.
33144 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33145 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33146 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33147 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33148 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33149 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33150 (P9V_BUILTIN_VEC_REVB): Likewise.
33151 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33152 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33153 (p9_xxbrq_v16qi): Likewise.
33154 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33155 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33156 (p9_xxbrh_v8hi): Likewise.
33157 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33158 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33159 vec_revb built-in functions.
33161 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33163 PR rtl-optimization/78952
33164 * config/i386/i386.md (any_extract): New code iterator.
33165 (*insvqi_2): Use any_extract for source operand.
33166 (*insvqi_3): Use any_shiftrt for source operand.
33168 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33170 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33172 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33174 2017-01-18 Matthias Klose <doko@ubuntu.com>
33176 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33178 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33180 * config/rs6000/altivec.h (vec_bperm): Change #define.
33181 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33182 (altivec_vbpermq2): New define_insn.
33183 (altivec_vbpermd): Likewise.
33184 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33185 function interface.
33186 (VBPERMD): Likewise.
33187 (VBPERM): New polymorphic function interface.
33188 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33189 Add entries for P9V_BUILTIN_VEC_VBPERM.
33190 * doc/extend.texi: Add interfaces for vec_bperm.
33192 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33194 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33195 first letter of error messages.
33196 (s390_resolve_overloaded_builtin): Likewise.
33197 * config/s390/s390.c (s390_expand_builtin): Likewise.
33198 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33199 (s390_valid_target_attribute_inner_p): Likewise.
33200 * config/s390/s390.md ("tabort"): Likewise.
33202 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33204 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33205 (ISA_AVOID_DIV_HILO): New macro.
33206 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33207 (ISA_HAS_DDIV): Likewise.
33209 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33211 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33213 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33215 * doc/invoke.texi (fabi-version): Spelling fix.
33217 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33220 * doc/invoke.texi (fabi-version): Mention mangling fix for
33223 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33226 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33228 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33231 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33232 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33234 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33236 (stack_protect_test): Ditto.
33237 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33238 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33240 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33241 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33242 -mstack-protector-guard-offset=.
33243 (RS/6000 and PowerPC Options): Ditto.
33245 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33247 * config/i386/i386.h (MASK_CLASS_P): New define.
33248 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33249 there are no registers from different register sets also when
33250 mask registers are used. Update function comment.
33251 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33252 to (*k/*r) and (*k/*km) alternatives.
33254 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33256 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33257 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33258 (EH_RETURN_HANDLER_RTX): New define.
33259 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33260 Force frame pointer in EH return functions.
33261 (aarch64_expand_epilogue): Add barrier for eh_return.
33262 (aarch64_final_eh_return_addr): Remove.
33263 (aarch64_eh_return_handler_rtx): New function.
33264 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33266 (aarch64_eh_return_handler_rtx): New prototype.
33268 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33270 * config/rs6000/altivec.h (vec_rlmi): New #define.
33271 (vec_vrlnm): Likewise.
33272 (vec_rlnm): Likewise.
33273 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33274 (UNSPEC_VRLNM): Likewise.
33275 (VIlong): New mode iterator.
33276 (altivec_vrl<VI_char>mi): New define_insn.
33277 (altivec_vrl<VI_char>nm): Likewise.
33278 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33280 (VRLDNM): Likewise.
33281 (RLNM): New polymorphic function entry.
33282 (VRLWMI): New monomorphic function entry.
33283 (VRLDMI): Likewise.
33284 (RLMI): New polymorphic function entry.
33285 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33286 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33287 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33290 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33293 * dwarf2out.c (field_byte_offset): Restore the
33294 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33295 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33296 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33299 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33302 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33304 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33307 * dwarf2out.c (add_data_member_location_attribute): For constant
33308 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33309 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33310 DW_AT_byte_size attributes.
33312 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33314 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33315 after forcing to constant memory when the code model is medium.
33317 2017-01-17 Julia Koval <julia.koval@intel.com>
33320 * config/i386/avx512fintrin.h
33321 (_mm512_i32gather_ps): Change __addr type to void const*.
33322 (_mm512_mask_i32gather_ps): Ditto.
33323 (_mm512_i32gather_pd): Ditto.
33324 (_mm512_mask_i32gather_pd): Ditto.
33325 (_mm512_i64gather_ps): Ditto.
33326 (_mm512_mask_i64gather_ps): Ditto.
33327 (_mm512_i64gather_pd): Ditto.
33328 (_mm512_mask_i64gather_pd): Ditto.
33329 (_mm512_i32gather_epi32): Ditto.
33330 (_mm512_mask_i32gather_epi32): Ditto.
33331 (_mm512_i32gather_epi64): Ditto.
33332 (_mm512_mask_i32gather_epi64): Ditto.
33333 (_mm512_i64gather_epi32): Ditto.
33334 (_mm512_mask_i64gather_epi32): Ditto.
33335 (_mm512_i64gather_epi64): Ditto.
33336 (_mm512_mask_i64gather_epi64): Ditto.
33337 (_mm512_i32scatter_ps): Change __addr type to void*.
33338 (_mm512_mask_i32scatter_ps): Ditto.
33339 (_mm512_i32scatter_pd): Ditto.
33340 (_mm512_mask_i32scatter_pd): Ditto.
33341 (_mm512_i64scatter_ps): Ditto.
33342 (_mm512_mask_i64scatter_ps): Ditto.
33343 (_mm512_i64scatter_pd): Ditto.
33344 (_mm512_mask_i64scatter_pd): Ditto.
33345 (_mm512_i32scatter_epi32): Ditto.
33346 (_mm512_mask_i32scatter_epi32): Ditto.
33347 (_mm512_i32scatter_epi64): Ditto.
33348 (_mm512_mask_i32scatter_epi64): Ditto.
33349 (_mm512_i64scatter_epi32): Ditto.
33350 (_mm512_mask_i64scatter_epi32): Ditto.
33351 (_mm512_i64scatter_epi64): Ditto.
33352 (_mm512_mask_i64scatter_epi64): Ditto.
33353 * config/i386/avx512pfintrin.h
33354 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33355 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33356 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33357 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33358 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33359 (_mm512_prefetch_i32scatter_ps): Ditto.
33360 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33361 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33362 (_mm512_prefetch_i64scatter_pd): Ditto.
33363 (_mm512_prefetch_i64scatter_ps): Ditto.
33364 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33365 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33366 * config/i386/avx512vlintrin.h
33367 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33368 (_mm_mmask_i32gather_ps): Ditto.
33369 (_mm256_mmask_i32gather_pd): Ditto.
33370 (_mm_mmask_i32gather_pd): Ditto.
33371 (_mm256_mmask_i64gather_ps): Ditto.
33372 (_mm_mmask_i64gather_ps): Ditto.
33373 (_mm256_mmask_i64gather_pd): Ditto.
33374 (_mm_mmask_i64gather_pd): Ditto.
33375 (_mm256_mmask_i32gather_epi32): Ditto.
33376 (_mm_mmask_i32gather_epi32): Ditto.
33377 (_mm256_mmask_i32gather_epi64): Ditto.
33378 (_mm_mmask_i32gather_epi64): Ditto.
33379 (_mm256_mmask_i64gather_epi32): Ditto.
33380 (_mm_mmask_i64gather_epi32): Ditto.
33381 (_mm256_mmask_i64gather_epi64): Ditto.
33382 (_mm_mmask_i64gather_epi64): Ditto.
33383 (_mm256_i32scatter_ps): Change __addr type to void*.
33384 (_mm256_mask_i32scatter_ps): Ditto.
33385 (_mm_i32scatter_ps): Ditto.
33386 (_mm_mask_i32scatter_ps): Ditto.
33387 (_mm256_i32scatter_pd): Ditto.
33388 (_mm256_mask_i32scatter_pd): Ditto.
33389 (_mm_i32scatter_pd): Ditto.
33390 (_mm_mask_i32scatter_pd): Ditto.
33391 (_mm256_i64scatter_ps): Ditto.
33392 (_mm256_mask_i64scatter_ps): Ditto.
33393 (_mm_i64scatter_ps): Ditto.
33394 (_mm_mask_i64scatter_ps): Ditto.
33395 (_mm256_i64scatter_pd): Ditto.
33396 (_mm256_mask_i64scatter_pd): Ditto.
33397 (_mm_i64scatter_pd): Ditto.
33398 (_mm_mask_i64scatter_pd): Ditto.
33399 (_mm256_i32scatter_epi32): Ditto.
33400 (_mm256_mask_i32scatter_epi32): Ditto.
33401 (_mm_i32scatter_epi32): Ditto.
33402 (_mm_mask_i32scatter_epi32): Ditto.
33403 (_mm256_i32scatter_epi64): Ditto.
33404 (_mm256_mask_i32scatter_epi64): Ditto.
33405 (_mm_i32scatter_epi64): Ditto.
33406 (_mm_mask_i32scatter_epi64): Ditto.
33407 (_mm256_i64scatter_epi32): Ditto.
33408 (_mm256_mask_i64scatter_epi32): Ditto.
33409 (_mm_i64scatter_epi32): Ditto.
33410 (_mm_mask_i64scatter_epi32): Ditto.
33411 (_mm256_i64scatter_epi64): Ditto.
33412 (_mm256_mask_i64scatter_epi64): Ditto.
33413 (_mm_i64scatter_epi64): Ditto.
33414 (_mm_mask_i64scatter_epi64): Ditto.
33415 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
33416 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
33417 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
33418 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
33419 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
33420 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
33421 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
33422 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
33423 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
33424 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
33425 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
33426 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
33427 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
33428 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
33429 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
33430 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
33431 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
33432 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
33433 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
33434 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
33435 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
33436 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
33437 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
33438 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
33439 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
33440 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
33441 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
33442 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
33443 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
33444 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
33445 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
33446 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
33447 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
33448 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
33449 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
33450 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
33451 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
33452 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
33453 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
33454 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
33455 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
33456 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
33457 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
33458 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
33459 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
33460 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
33461 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
33462 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
33463 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
33464 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
33465 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
33466 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
33467 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
33468 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
33469 definitions accordingly.
33471 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
33472 Kuan-Lin Chen <kuanlinchentw@gmail.com>
33475 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
33478 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
33481 * ira-conflicts.c (ira_build_conflicts): Update total conflict
33482 hard regs for inner regno.
33484 2017-01-17 Martin Liska <mliska@suse.cz>
33487 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
33488 assumption and add comment.
33490 2017-01-17 Nathan Sidwell <nathan@acm.org>
33492 * ipa-visibility.c (localize_node): New function, broken out of ...
33493 (function_and_variable_visibility): ... here. Call it.
33495 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33497 PR middle-end/77445
33498 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
33499 correctly set frequency of oudgoing edge.
33500 (duplicate_thread_path): Fix profile updating.
33502 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33505 * configure.ac: Add GCC_BASE_VER.
33506 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
33507 version from BASE-VER file.
33508 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
33509 (gcc.o): Depend on $(BASEVER).
33510 * common.opt (dumpfullversion): New option.
33511 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
33512 * doc/invoke.texi: Document -dumpfullversion.
33513 * doc/install.texi: Document --with-gcc-major-version-only.
33514 * configure: Regenerated.
33516 2017-01-17 Richard Biener <rguenther@suse.de>
33518 PR tree-optimization/71433
33519 * tree-vrp.c (register_new_assert_for): Merge same asserts
33520 on all incoming edges.
33521 (process_assert_insertions_for): Handle insertions at the
33524 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
33526 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
33527 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
33529 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
33532 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
33535 2017-01-17 Alan Modra <amodra@gmail.com>
33538 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
33539 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
33540 symbolic stack limit when pic.
33542 2017-01-16 Martin Sebor <msebor@redhat.com>
33544 PR tree-optimization/78608
33545 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
33547 2017-01-16 Jeff Law <law@redhat.com>
33550 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
33551 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
33552 for several include directories that may be relative to sysroot.
33553 * config/i386/x-mingw32 (gplus_includedir): Define.
33554 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
33555 (native_system_includedir): Likewise.
33556 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
33557 override if TARGET_SYSTEM_ROOT is defined.
33558 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
33560 PR tree-optimization/79090
33561 PR tree-optimization/33562
33562 PR tree-optimization/61912
33563 PR tree-optimization/77485
33564 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
33565 and computed trims into the dump file.
33567 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33569 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
33571 2017-01-16 Jakub Jelinek <jakub@redhat.com>
33574 * gimplify.c (gimplify_init_constructor): If want_value and
33575 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
33579 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
33580 sequence. Formatting fixes.
33581 (doloop_optimize): Formatting fixes.
33584 * gcc.c (debug_level_greater_than_spec_func): New function.
33585 (static_spec_functions): Add debug-level-gt spec function.
33586 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
33588 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33589 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33590 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
33591 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
33592 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
33593 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
33595 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
33597 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
33598 QImode fixups to general and mask registers only.
33600 2017-01-16 Carl Love <cel@us.ibm.com>
33602 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
33603 for built-in functions
33604 vector signed char vec_nabs (vector signed char)
33605 vector signed short vec_nabs (vector signed short)
33606 vector signed int vec_nabs (vector signed int)
33607 vector signed long long vec_nabs (vector signed long long)
33608 vector float vec_nabs (vector float)
33609 vector double vec_nabs (vector double)
33610 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
33612 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
33613 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
33614 * doc/extend.texi: Update the documentation file for the new built-in
33617 2017-01-16 Martin Sebor <msebor@redhat.com>
33619 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
33622 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33624 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
33625 UNSPEC_VSX__XXSPLTD to require special splat handling.
33627 2017-01-16 David Malcolm <dmalcolm@redhat.com>
33630 * system.h: Poison strndup.
33632 2017-01-16 Alan Modra <amodra@gmail.com>
33635 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
33638 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
33640 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
33642 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
33644 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
33645 call recog here. Assert that INSN_CODE (insn) is non-negative.
33647 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33650 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
33652 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
33653 in the currently scheduled RTL fragment.
33655 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33657 PR rtl-optimization/78751
33658 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
33661 2017-01-14 Jeff Law <law@redhat.com>
33663 PR tree-optimization/79090
33664 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
33665 variable length stores.
33666 (compute_trims): Delete dead assignment to *trim_tail.
33667 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
33670 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
33672 PR rtl-optimization/78626
33673 PR rtl-optimization/78727
33674 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
33675 of a block, and split such blocks after everything else is finished.
33677 2017-01-14 Alan Modra <amodra@gmail.com>
33680 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
33681 target legitimate_combined_insn.
33682 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
33683 (rs6000_legitimate_combined_insn): New function.
33684 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
33686 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
33687 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
33688 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
33690 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
33692 * doc/frontends.texi (G++ and GCC): Remove references to Java.
33694 2017-01-13 Jeff Law <law@redhat.com>
33696 PR tree-optimization/33562
33697 PR tree-optimization/61912
33698 PR tree-optimization/77485
33699 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
33701 (delete_dead_assignment): Likewise.
33702 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
33703 statement to delete_dead_call and delete_dead_assignment.
33705 2017-01-13 David Malcolm <dmalcolm@redhat.com>
33708 * substring-locations.c (format_warning_va): Strengthen case 1 so
33709 that both endpoints of the substring must be within the format
33710 range for just the substring to be printed.
33712 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
33714 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
33715 * config/i386/i386.c (ix86_target_string): Add missing options
33716 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
33717 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
33718 flags_other and ix86_target_other to flags2_other. Display unknown
33720 (ix86_valid_target_attribute_inner_p): Add missing options and
33721 reorder options by implied ISAs, as in ix86_target_string.
33723 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33725 * hash-table.h (hash_table::too_empty_p): New function.
33726 (hash_table::expand): Use it.
33727 (hash_table::traverse): Likewise.
33728 (hash_table::empty_slot): Use sizeof (value_type) instead of
33729 sizeof (PTR) to convert bytes to elements. Shrink the table
33730 if the current size is excessive for the current number of
33733 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
33735 * ira-costs.c (record_reg_classes): Break from the inner loop
33736 early once alt_fail is known to be true. Update outer loop
33737 handling accordingly.
33739 2017-01-13 Jeff Law <law@redhat.com>
33741 * tree-ssa-dse.c (decrement_count): New function.
33742 (increment_start_addr, maybe_trim_memstar_call): Likewise.
33743 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
33744 when we know the partially dead statement is a mem* function.
33746 PR tree-optimization/61912
33747 PR tree-optimization/77485
33748 * tree-ssa-dse.c: Include expr.h.
33749 (maybe_trim_constructor_store): New function.
33750 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
33752 PR tree-optimization/33562
33753 PR tree-optimization/61912
33754 PR tree-optimization/77485
33755 * doc/invoke.texi: Document new dse-max-object-size param.
33756 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
33757 * tree-ssa-dse.c: Include params.h.
33758 (dse_store_status): New enum.
33759 (initialize_ao_ref_for_dse): New, partially extracted from
33761 (valid_ao_ref_for_dse, normalize_ref): New.
33762 (setup_live_bytes_from_ref, compute_trims): Likewise.
33763 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
33764 (maybe_trim_partially_dead_store): Likewise.
33765 (maybe_trim_complex_store): Likewise.
33766 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
33767 Track what bytes live from the original store. Return tri-state
33768 for dead, partially dead or live.
33769 (dse_dom_walker): Add constructor, destructor and new private members.
33770 (delete_dead_call, delete_dead_assignment): New extracted from
33772 (dse_optimize_stmt): Make a member of dse_dom_walker.
33773 Use initialize_ao_ref_for_dse.
33775 PR tree-optimization/33562
33776 PR tree-optimization/61912
33777 PR tree-optimization/77485
33778 * sbitmap.h (bitmap_count_bits): Prototype.
33779 (bitmap_clear_range, bitmap_set_range): Likewise.
33780 * sbitmap.c (bitmap_clear_range): New function.
33781 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
33783 2017-01-13 Martin Liska <mliska@suse.cz>
33786 * function.c (set_cfun): Add new argument force.
33787 * function.h (set_cfun): Likewise.
33788 * ipa-inline-transform.c (inline_call): Use the function when
33789 strict alising from is dropped for function we inline to.
33791 2017-01-13 Richard Biener <rguenther@suse.de>
33793 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
33794 for dumping GIMPLE INTEGER_CSTs.
33796 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33798 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
33799 to 201112L since C++17.
33801 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
33804 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
33805 if -fsanitize=kernel-address is present.
33807 2017-01-13 Richard Biener <rguenther@suse.de>
33809 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
33810 as _Literal ( type ) number in case usual suffixes do not
33811 preserve all information.
33813 2017-01-13 Richard Biener <rguenther@suse.de>
33815 PR tree-optimization/77283
33816 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
33817 and ssa-iterators.h.
33818 (is_feasible_trace): Implement a cost model based on joiner
33821 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
33824 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
33825 char or short to __float128/_Float128 directly.
33827 2017-01-12 Martin Sebor <msebor@redhat.com>
33829 to -Wformat-overflow.
33830 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
33831 (min_bytes_remaining): Same.
33832 (get_string_length): Same.
33833 (format_string): Same.
33834 (format_directive): Same.
33836 (pass_sprintf_length::handle_gimple_call): Same.
33838 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33840 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33841 info.nowrite calls with no lhs that can't throw. Return bool
33842 whether gsi_remove has been called or not.
33843 (pass_sprintf_length::handle_gimple_call): Return bool whether
33844 try_substitute_return_value called gsi_remove. Formatting fix.
33845 (pass_sprintf_length::execute): Don't use gsi_remove if
33846 handle_gimple_call returned true.
33849 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33850 be removed due to side-effects, don't remove following barrier nor
33851 turn the successor edge into fallthru edge.
33853 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33856 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33857 element-reversing loads and stores as not swappable.
33859 2017-01-12 Nathan Sidwell <nathan@acm.org>
33860 Nicolai Stange <nicstange@gmail.com>
33862 * combine.c (try_combine): Don't ignore result of overlap checking
33863 loop. Combine overlap & asm check into single loop.
33865 2017-01-12 Richard Biener <rguenther@suse.de>
33867 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33868 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33870 2017-01-12 Richard Biener <rguenther@suse.de>
33872 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33873 and TS_TARGET_OPTION directly derive from TS_BASE.
33874 * tree-core.h (tree_optimization_option): Derive from tree_base.
33875 (tree_target_option): Likewise.
33877 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33879 * config/i386/i386.c (memory_address_length): Increase len
33880 only when rip_relative_addr_p returns false.
33882 2017-01-11 Julia Koval <julia.koval@intel.com>
33884 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33885 (OPTION_MASK_ISA_SGX_SET): New.
33886 (ix86_handle_option): Handle OPT_msgx.
33887 * config.gcc: Added sgxintrin.h.
33888 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
33889 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
33890 * config/i386/i386.c (ix86_target_string): Add -msgx.
33892 (ix86_option_override_internal): Handle new options.
33893 (ix86_valid_target_attribute_inner_p): Add sgx.
33894 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
33895 * config/i386/i386.opt: Add msgx.
33896 * config/i386/sgxintrin.h: New file.
33897 * config/i386/x86intrin.h: Add sgxintrin.h.
33899 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33902 * fold-const.c (maybe_nonzero_address): Return 1 for function
33904 (tree_single_nonzero_warnv_p): Don't handle function local objects
33908 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33911 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33914 * opts.c: Include "spellcheck.h"
33915 (struct string_fragment): New struct.
33916 (struct edit_distance_traits<const string_fragment &>): New
33918 (get_closest_sanitizer_option): New function.
33919 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33921 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33923 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33925 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33926 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33927 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33928 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33929 for initial die_offset if dwarf_split_debug_info.
33930 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33931 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33933 (output_skeleton_debug_sections): Formatting fix. Use
33934 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
33935 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
33937 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
33939 * config/arm/cortex-a53.md: Add bypasses for
33940 cortex_a53_r2f_cvt.
33941 (cortex_a53_r2f): Only use for transfers.
33942 (cortex_a53_f2r): Likewise.
33943 (cortex_a53_r2f_cvt): Add reservation for conversions.
33944 (cortex_a53_f2r_cvt): Likewise.
33946 2017-01-11 Tamar Christina <tamar.christina@arm.com>
33948 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
33949 to all inlined functions, change static to extern.
33951 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
33954 * config/arm/arm.c (legitimize_pic_address): Handle reference to
33956 (arm_assemble_integer): Likewise.
33958 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
33960 * config.gcc: Use new awk script to check CPU, FPU and architecture
33961 parameters for --with-... options.
33962 * config/arm/parsecpu.awk: New file
33963 * config/arm/arm-cpus.in: New file.
33964 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
33966 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
33968 * config/arm/t-arm: Update dependency rules.
33969 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
33970 of processing .def files.
33971 * config/arm/genopt.sh: Deleted.
33972 * config/arm/gentune.sh: Deleted.
33973 * config/arm/arm-cores.def: Deleted.
33974 * config/arm/arm-arches.def: Deleted.
33975 * config/arm/arm-fpus.def: Deleted.
33976 * config/arm/arm-tune.md: Regenerated.
33977 * config/arm/arm-tables.opt: Regenerated.
33978 * config/arm/arm-cpu.h: New generated file.
33979 * config/arm/arm-cpu-data.h: New generated file.
33980 * config/arm/arm-cpu-cdata.h: New generated file.
33982 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33985 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
33987 (input_varpool_node): Unpack dynamically_initialized bit.
33989 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
33991 PR rtl-optimization/79032
33992 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
33993 the alignment of the adjusted memory reference against that of MODE,
33994 instead of the alignment of the original memory reference.
33996 2017-01-11 Martin Jambor <mjambor@suse.cz>
33998 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34000 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34001 decorated functions.
34003 2017-01-11 Richard Biener <rguenther@suse.de>
34005 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34006 set range/nonnull info for PHI results. Do not set it on
34007 stmts marked for removal.
34009 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34011 * expr.c (store_field): In the bitfield case, fetch the return value
34012 from the registers before applying a single big-endian adjustment.
34013 Always do a final load for a BLKmode value not larger than a word.
34015 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34018 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34019 that we correctly handle column numbers greater than
34020 LINE_MAP_MAX_COLUMN_NUMBER.
34022 2017-01-10 Martin Sebor <msebor@redhat.com>
34024 PR middle-end/78245
34025 * gimple-ssa-sprintf.c (get_destination_size): Call
34026 {init,fini}object_sizes.
34027 * tree-object-size.c (addr_object_size): Adjust.
34028 (pass_through_call): Adjust.
34029 (pass_object_sizes::execute): Adjust.
34030 * tree-object-size.h (fini_object_sizes): Declare.
34032 2017-01-10 Martin Sebor <msebor@redhat.com>
34034 PR tree-optimization/78775
34035 * builtins.c (get_size_range): Move...
34036 * calls.c: ...to here.
34037 (alloc_max_size): Accept zero argument.
34038 (operand_signed_p): Remove.
34039 (maybe_warn_alloc_args_overflow): Call get_size_range.
34040 * calls.h (get_size_range): Declare.
34042 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34044 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34045 from TI's devices.csv file as of September 2016.
34046 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34048 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34050 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34051 * doc/invoke.texi: Likewise.
34052 * doc/md.texi: Likewise.
34053 * doc/objc.texi: Likewise.
34055 2017-01-10 Joshua Conner <joshconner@google.com>
34057 * config/arm/fuchsia-elf.h: New file.
34058 * config/fuchsia.h: New file.
34059 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34060 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34062 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34064 2016-01-10 Richard Biener <rguenther@suse.de>
34066 PR tree-optimization/79034
34067 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34068 Propagate out degenerate PHIs in the joiner.
34070 2017-01-10 Martin Liska <mliska@suse.cz>
34072 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34073 (sort_congruence_classes_by_decl_uid): Likewise.
34074 (sort_congruence_class_groups_by_decl_uid): Likewise.
34075 (sem_item_optimizer::merge_classes): Sort class, groups in these
34076 classes and members in the groups by DECL_UID of declarations.
34077 This would make merge operations stable.
34079 2017-01-10 Martin Liska <mliska@suse.cz>
34081 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34082 usage of m_classes_vec.
34083 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34084 (sem_item_optimizer::get_group_by_hash): Likewise.
34085 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34086 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34087 (sem_item_optimizer::verify_classes): Likewise.
34088 (sem_item_optimizer::process_cong_reduction): Likewise.
34089 (sem_item_optimizer::dump_cong_classes): Likewise.
34090 (sem_item_optimizer::merge_classes): Likewise.
34091 * ipa-icf.h (congruence_class_hash): Rename from
34092 congruence_class_group_hash. Remove declaration of m_classes_vec.
34094 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34096 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34097 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34098 * config.gcc: Add avx512vpopcntdqintrin.h.
34099 * config/i386/avx512vpopcntdqintrin.h: New.
34100 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34101 * config/i386/i386-builtin-types.def: Add new types.
34102 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34103 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34104 __builtin_ia32_vpopcountq_v8di_mask): New.
34105 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34106 __AVX512VPOPCNTDQ__.
34107 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34108 (PTA_AVX512VPOPCNTDQ): Define.
34109 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34110 TARGET_AVX512VPOPCNTDQ_P): Define.
34111 * config/i386/i386.opt: Add mavx512vpopcntdq.
34112 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34113 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34115 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34117 PR middle-end/77484
34118 * predict.def (PRED_CALL): Set to 67.
34120 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34122 * expr.c (store_field): In the bitfield case, if the value comes from
34123 a function call and is of an aggregate type returned in registers, do
34124 not modify the field mode; extract the value in all cases if the mode
34125 is BLKmode and the size is not larger than a word.
34127 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34130 * config/i386/cpuid.h: Fix undefined behavior.
34132 2017-01-04 Jeff Law <law@redhat.com>
34134 PR tree-optimization/79007
34135 PR tree-optimization/67955
34136 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34137 conservative for pt.null when flag_non_call_exceptions is on.
34139 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34141 PR translation/79019
34142 PR translation/79020
34143 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34144 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34145 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34147 * config/avr/avr.opt (maccumulate-args): Likewise.
34148 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34149 * common.opt (freport-bug): Likewise.
34150 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34151 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34152 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34153 translatable string.
34154 * config/i386/i386.c (function_value_32): Likewise.
34155 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34156 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34158 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34159 * common/config/msp430/msp430-common.c (msp430_handle_option):
34161 * symtab.c (symtab_node::verify_base): Likewise.
34162 * opts.c (set_debug_level): Likewise.
34163 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34164 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34165 missing whitespace to translatable strings.
34166 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34167 * config/sh/superh.h: Likewise.
34168 * config/i386/xopintrin.h: Likewise.
34169 * config/i386/znver1.md: Likewise.
34170 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34171 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34172 * double-int.h (struct double_int): Likewise.
34173 * double-int.c (div_and_round_double): Likewise.
34174 * wide-int.cc: Likewise.
34175 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34176 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34177 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34178 (crossjumps_occurred): ... this.
34179 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34182 PR tree-optimization/78899
34183 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34184 returning bool return struct loop *, NULL for failure and the new
34186 (versionable_outer_loop_p): Don't version outer loop if it has
34187 dont_vectorized bit set.
34188 (tree_if_conversion): When versioning outer loop, ensure
34189 tree_if_conversion is performed also on the inner loop of the
34190 non-vectorizable outer loop copy.
34191 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34192 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34193 prevent vectorization of it.
34194 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34195 the outer loop vectorization of the non-scalar version is attempted
34196 before vectorization of the inner loop in scalar version. If
34197 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34198 vectorization of its inner loop.
34199 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34200 has 2 inner loops, rename also on edges from bb whose single pred
34201 is outer_loop->header. Fix typo in function comment.
34203 2017-01-09 Martin Sebor <msebor@redhat.com>
34206 * asan.c (asan_emit_stack_protection): Increase local buffer size
34207 to avoid snprintf truncation warning.
34209 2017-01-09 Andrew Pinski <apinski@cavium.com>
34211 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34212 to reference thunderx2t99 for the tuning structure
34213 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34215 (thunderx2t99_extra_costs): This.
34216 * config/aarch64/aarch64-tune.md: Regenerate.
34217 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34218 (vulcan_addrcost_table): This.
34219 (vulcan_regmove_cost): Rename to ...
34220 (thunderx2t99_regmove_cost): This.
34221 (vulcan_vector_cost): Rename to ...
34222 (thunderx2t99_vector_cost): this.
34223 (vulcan_branch_cost): Rename to ...
34224 (thunderx2t99_branch_cost): This.
34225 (vulcan_tunings): Rename to ...
34226 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34227 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34229 2017-01-09 Martin Jambor <mjambor@suse.cz>
34233 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34234 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34235 (propagate_vr_accross_jump_function): Use the above function for all
34236 value range computations for pass-through jump functions and type
34237 converasion from explicit value range values.
34238 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34239 parameters from TYPE_ARG_TYPES.
34240 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34241 (ipa_write_node_info): Stream type of the actual argument.
34242 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34244 2017-01-09 Martin Liska <mliska@suse.cz>
34247 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34248 (lookup_compiler): Do not show error message with have_E.
34250 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34252 PR tree-optimization/78938
34253 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34254 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34255 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34256 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34259 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34261 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34264 2017-01-09 Richard Biener <rguenther@suse.de>
34266 PR tree-optimization/78997
34267 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34268 name condition properly.
34270 2017-01-09 Richard Biener <rguenther@suse.de>
34273 * dwarf2out.c (is_cxx): New overload with context.
34274 (is_naming_typedef_decl): Use it.
34276 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34278 * invoke.texi (Option Summary): Correct spacing in option lists
34279 and add line breaks to fix over-long lines.
34281 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34283 PR middle-end/17660
34285 * extend.texi (Common Variable Attributes): Add xref to GCC
34286 Internals manual to explain mode attribute keywords.
34288 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34291 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34292 and Preprocessor Options.
34293 (Options for Linking): Document -pthread here....
34294 (RS/6000 and PowerPC Options): ...not here.
34295 (Solaris 2 Options): ...or here.
34296 * doc/cppopts.texi: Document -pthread.
34298 2017-01-08 Martin Sebor <msebor@redhat.com>
34300 PR middle-end/77708
34301 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34302 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34303 New member functions.
34304 (format_directive): Used them.
34306 (pass_sprintf_length::handle_gimple_call): Same.
34307 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34308 to avoid truncation for any argument.
34309 (extract_affine_mul): Same.
34310 * tree.c (get_file_function_name): Same.
34312 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34314 PR middle-end/77484
34315 * predict.def (PRED_INDIR_CALL): Set to 86.
34317 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34319 PR preprocessor/54124
34320 * doc/cppopts.texi: Reformat -d subtable to list the full name
34321 of the options. Add cross-reference to the docs for the general
34322 compiler -d options.
34323 * doc/invoke.texi (Developer Options): Add cross-reference to the
34324 preprocessor-specific -d option documentation.
34326 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34328 PR preprocessor/13498
34329 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34330 redudant material, and reflect new command-line options.
34331 (System Headers): Likewise.
34333 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34335 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34336 -isystem, and -idirafter. Copy-edit.
34337 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34338 default for -ftrack-macro-expansion. Delete obsolete and
34339 badly-formatted implementation details about -fdebug-cpp output.
34340 * doc/cppwarnopts.texi: Copy-edit.
34342 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34345 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34346 that the transition from a max line width >= 1<<10 to narrower
34347 lines works correctly.
34349 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34351 * doc/options.texi (PerFunction): New.
34352 * opt-functions.awk (switch_flags): Map both Optimization and
34353 PerFunction to CL_OPTIMIZATION.
34354 * opth-gen.awk: Test for PerFunction flag along with
34356 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34357 it only when the latter is present. Skip those that don't in
34358 the hash function generator.
34359 * common.opt (fvar-tracking): Mark as PerFunction instead of
34361 (fvar-tracking-assignments): Likewise.
34362 (fvar-tracking-assignments-toggle): Likewise.
34363 (fvar-tracking-uninit): Likewise.
34365 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34367 PR translation/79018
34368 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34371 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34374 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34375 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34376 TARGET_LONG_JUMP_TABLE_OFFSETS.
34377 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34378 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34379 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34380 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34381 * config/m68k/m68k.md (tablejump expander): Likewise.
34382 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34383 TARGET_LONG_JUMP_TABLE_OFFSETS.
34384 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
34385 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
34387 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
34388 David Holsgrove <david.holsgrove@xilinx.com>
34390 * common/config/microblaze/microblaze-common.c
34391 (TARGET_EXCEPT_UNWIND_INFO): Remove.
34392 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
34394 * config/microblaze/microblaze.c (microblaze_must_save_register)
34395 (microblaze_expand_epilogue, microblaze_return_addr): Handle
34397 (microblaze_eh_return): New function.
34398 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
34399 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
34400 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
34401 * config/microblaze/microblaze.md (eh_return): New pattern.
34403 2017-01-06 Jakub Jelinek <jakub@redhat.com>
34405 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
34406 GCC_DIAGNOSTIC_STRINGIFY): Define.
34408 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
34410 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34412 * config/arm/arm.md (<mcrr>): New.
34414 * config/arm/arm.c (arm_arch5te): New.
34415 (arm_option_override): Set arm_arch5te.
34416 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
34418 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
34419 (arm_mcrr_qualifiers): ... this. New.
34420 (MRRC_QUALIFIERS): Define to...
34421 (arm_mrrc_qualifiers): ... this. New.
34422 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
34424 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
34425 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
34426 (MRRCI, mrrc, MRRC): New.
34427 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
34428 VUNSPEC_MRRC2): New.
34430 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34432 * config/arm/arm.md (<mcr>): New.
34434 * config/arm/arm.c (arm_coproc_builtin_available): Add
34435 support for mcr, mrc, mcr2 and mrc2.
34436 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
34437 (arm_mcr_qualifiers): ... this. New.
34438 (MRC_QUALIFIERS): Define to ...
34439 (arm_mrc_qualifiers): ... this. New.
34440 (MCR_QUALIFIERS): Define to ...
34441 (arm_mcr_qualifiers): ... this. New.
34442 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
34444 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
34445 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
34446 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
34447 VUNSPEC_MRC2): New.
34449 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34451 * config/arm/arm.md (*ldc): New.
34455 * config/arm/arm.c (arm_coproc_builtin_available): Add
34456 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
34457 (arm_coproc_ldc_stc_legitimate_address): New.
34458 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
34459 'qualifier_const_pointer'.
34460 (LDC_QUALIFIERS): Define to...
34461 (arm_ldc_qualifiers): ... this. New.
34462 (STC_QUALIFIERS): Define to...
34463 (arm_stc_qualifiers): ... this. New.
34464 * config/arm/arm-protos.h
34465 (arm_coproc_ldc_stc_legitimate_address): New.
34466 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
34467 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
34468 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
34469 stc2, stcl, stc2l): New.
34470 * config/arm/constraints.md (Uz): New.
34471 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
34472 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
34473 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
34474 VUNSPEC_STC2L): New.
34476 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34478 * config/arm/arm.md (<cdp>): New.
34479 * config/arm/arm.c (neon_const_bounds): Rename this ...
34480 (arm_const_bounds): ... this.
34481 (arm_coproc_builtin_available): New.
34482 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
34483 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
34484 (CDP_QUALIFIERS): Define to...
34485 (arm_cdp_qualifiers): ... this. New.
34487 (arm_expand_builtin_args): Add case for 6 arguments.
34488 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
34489 (arm_const_bounds): ... this.
34490 (arm_coproc_builtin_available): New.
34491 * config/arm/arm_acle.h (__arm_cdp): New.
34493 * config/arm/arm_acle_builtins.def (cdp): New.
34495 * config/arm/iterators.md (CDPI,CDP,cdp): New.
34496 * config/arm/neon.md: Rename all 'neon_const_bounds' to
34497 'arm_const_bounds'.
34498 * config/arm/types.md (coproc): New.
34499 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
34500 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
34501 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
34502 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
34504 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34506 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
34507 (UBINOP_QUALIFIERS): New.
34509 (acle_builtin_data): New. Change comment.
34510 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
34511 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
34512 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
34513 arm_acle_builtins.def.
34514 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
34515 (arm_init_acle_builtins): New.
34516 (CRC32_BUILTIN): Remove.
34517 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
34518 crc32cb, crc32ch and crc32cw.
34519 (arm_init_crc32_builtins): Remove.
34520 (arm_init_builtins): Use arm_init_acle_builtins rather
34521 than arm_init_crc32_builtins.
34522 (arm_expand_acle_builtin): New.
34523 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
34524 * config/arm/arm_acle_builtins.def: New.
34526 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34528 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
34529 (arm_builtin_datum): ... this.
34530 (arm_init_neon_builtin): Rename to ...
34531 (arm_init_builtin): ... this. Add a new parameters PREFIX
34532 and USE_SIG_IN_NAME.
34533 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
34534 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
34535 'arm_builtin_datum'.
34536 (arm_init_vfp_builtins): Likewise.
34537 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
34538 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
34539 (arm_expand_neon_args): Rename to ...
34540 (arm_expand_builtin_args): ... this. Rename builtin_arg
34541 enum values and differentiate between ARG_BUILTIN_MEMORY
34542 and ARG_BUILTIN_NEON_MEMORY.
34543 (arm_expand_neon_builtin_1): Rename to ...
34544 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
34545 values, arm_expand_builtin_args and add bool parameter NEON.
34546 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
34547 (arm_expand_vfp_builtin): Likewise.
34548 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
34550 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34552 PR middle-end/77484
34553 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
34554 * predict.c (tree_estimate_probability_bb): Reverse direction of
34555 polymorphic call predictor.
34557 2017-01-06 David Malcolm <dmalcolm@redhat.com>
34559 * passes.c (execute_one_pass): Split out pass-skipping logic into...
34560 (determine_pass_name_match): ...this new function and...
34561 (should_skip_pass_p): ...this new function.
34563 2017-01-06 Nathan Sidwell <nathan@acm.org>
34565 * ipa-visibility.c (function_and_variable_visibility): Reformat
34566 comments and long lines. Remove extrneous if.
34567 * symtab.c (symtab_node::make_decl_local): Fix code format.
34568 (symtab_node::set_section_for_node): Fix comment typo.
34570 2017-01-06 Martin Liska <mliska@suse.cz>
34573 * lra-constraints.c: Rename invariant to lra_invariant.
34574 * predict.c (set_even_probabilities): Initialize e to NULL.
34576 2017-01-05 Martin Sebor <msebor@redhat.com>
34578 PR tree-optimization/78910
34579 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
34580 (format_integer): Correct off-by-one error in the handling
34581 of precision with negative numbers in signed conversions..
34583 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
34585 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
34587 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34589 PR tree-optimization/71016
34590 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
34591 factor_out_conditional_conversion. Formatting fix.
34592 (factor_out_conditional_conversion): Add cond_stmt argument.
34593 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
34594 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
34597 2017-01-05 David Malcolm <dmalcolm@redhat.com>
34599 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
34600 read-rtl-function.o, and selftest-rtl.o.
34601 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
34602 (selftest::aarch64_test_loading_full_dump): New function.
34603 (selftest::aarch64_run_selftests): New function.
34604 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
34605 selftest::aarch64_run_selftests.
34606 * config/i386/i386.c
34607 (selftest::ix86_test_loading_dump_fragment_1): New function.
34608 (selftest::ix86_test_loading_call_insn): New function.
34609 (selftest::ix86_test_loading_full_dump): New function.
34610 (selftest::ix86_test_loading_unspec): New function.
34611 (selftest::ix86_run_selftests): Call the new functions.
34612 * emit-rtl.c (maybe_set_max_label_num): New function.
34613 * emit-rtl.h (maybe_set_max_label_num): New decl.
34614 * function.c (instantiate_decls): Guard call to
34615 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
34616 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
34618 * gensupport.c (gen_reader::gen_reader): Pass "false"
34619 for new "compact" param of rtx_reader.
34620 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
34621 rather than an empty string for NULL strings.
34622 * read-md.c: Potentially include config.h rather than bconfig.h.
34623 Wrap include of errors.h with #ifdef GENERATOR_FILE.
34624 (have_error): New global, copied from errors.c.
34625 (md_reader::read_name): Rename to...
34626 (md_reader::read_name_1): ...this, adding "out_loc" param,
34627 and converting "missing name or number" to returning false, rather
34629 (md_reader::read_name): Reimplement in terms of read_name_1.
34630 (md_reader::read_name_or_nil): New function.
34631 (md_reader::read_string): Handle "(nil)" by returning NULL.
34632 (md_reader::md_reader): Add new param "compact".
34633 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
34634 (md_reader::read_file): New method.
34635 * read-md.h (md_reader::md_reader): Add new param "compact".
34636 (md_reader::read_file): New method.
34637 (md_reader::is_compact): New accessor.
34638 (md_reader::read_name): Convert return type from void to file_location.
34639 (md_reader::read_name_or_nil): New decl.
34640 (md_reader::read_name_1): New decl.
34641 (md_reader::m_compact): New field.
34642 (noop_reader::noop_reader): Pass "false" for new "compact" param
34644 (rtx_reader::rtx_reader): Add new "compact" param.
34645 (rtx_reader::read_rtx_operand): Make virtual and convert return
34646 type from void to rtx.
34647 (rtx_reader::read_until): New decl.
34648 (rtx_reader::handle_any_trailing_information): New virtual function.
34649 (rtx_reader::postprocess): New virtual function.
34650 (rtx_reader::finalize_string): New virtual function.
34651 (rtx_reader::m_in_call_function_usage): New field.
34652 (rtx_reader::m_reuse_rtx_by_id): New field.
34653 * read-rtl-function.c: New file.
34654 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
34655 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
34656 (selftest::verify_three_block_rtl_cfg): New decl.
34657 * read-rtl-function.h: New file.
34658 * read-rtl.c: Potentially include config.h rather than bconfig.h.
34659 For host, include function.h, memmodel.h, and emit-rtl.h.
34660 (one_time_initialization): New function.
34661 (struct compact_insn_name): New struct.
34662 (compact_insn_names): New array.
34663 (find_code): Handle insn codes in compact dumps.
34664 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
34665 (bind_subst_iter_and_attr): Likewise.
34666 (add_condition_to_string): Likewise.
34667 (add_condition_to_rtx): Likewise.
34668 (apply_attribute_uses): Likewise.
34669 (add_current_iterators): Likewise.
34670 (apply_iterators): Likewise.
34671 (initialize_iterators): Guard usage of apply_subst_iterator with
34672 #ifdef GENERATOR_FILE.
34673 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
34674 (md_reader::read_mapping): Likewise.
34675 (add_define_attr_for_define_subst): Likewise.
34676 (add_define_subst_attr): Likewise.
34677 (read_subst_mapping): Likewise.
34678 (check_code_iterator): Likewise.
34679 (rtx_reader::read_rtx): Likewise. Move one-time initialization
34681 (one_time_initialization): New function.
34682 (rtx_reader::read_until): New method.
34683 (read_flags): New function.
34684 (parse_reg_note_name): New function.
34685 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
34686 Handle reuse_rtx ids.
34687 Wrap iterator lookup within #ifdef GENERATOR_FILE.
34688 Add parsing support for RTL dumps, mirroring the special-cases in
34689 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
34690 values, and calling handle_any_trailing_information.
34691 (rtx_reader::read_rtx_operand): Convert return type from void
34692 to rtx, returning return_rtx. Handle case 'e'. Call
34693 finalize_string on XSTR and XTMPL fields.
34694 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
34695 "(nil)" values were omitted. Call the postprocess vfunc on the
34697 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
34698 class ctor. Initialize m_in_call_function_usage. Call
34699 one_time_initialization.
34700 * rtl-tests.c (selftest::test_uncond_jump): Call
34701 set_new_first_and_last_insn.
34702 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
34703 * selftest-rtl.c: New file.
34704 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
34705 (selftest::get_insn_by_uid): New decl.
34706 * selftest-run-tests.c (selftest::run_tests): Call
34707 read_rtl_function_c_tests.
34708 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
34709 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
34712 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
34714 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
34715 operands in a special way. Assert that pos+len <= mode precision.
34717 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34719 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
34720 3 argument Alias with unlimited for the negative form.
34721 (fno-vect-cost-model): Removed.
34723 2017-01-05 Martin Liska <mliska@suse.cz>
34725 * hsa-gen.c (gen_hsa_divmod): New function.
34726 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
34728 2017-01-05 Martin Liska <mliska@suse.cz>
34731 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
34734 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34736 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
34737 small constant length operands.
34739 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34741 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
34742 between loop iterations.
34744 2017-01-05 Martin Liska <mliska@suse.cz>
34747 * gimplify.c (gimplify_decl_expr): Compare to
34748 asan_poisoned_variables instread of checking flags.
34749 (gimplify_target_expr): Likewise.
34750 (gimplify_expr): Likewise.
34751 (gimplify_function_tree): Conditionally initialize
34752 asan_poisoned_variables.
34754 2017-01-04 Jeff Law <law@redhat.com>
34756 PR tree-optimizatin/78812
34757 * rtl.h (contains_mem_rtx_p): Prototype.
34758 * ifcvt.c (containts_mem_rtx_p): Move from here to...
34759 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
34760 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
34761 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
34762 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
34764 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34766 * input.c (assert_char_at_range): Default-initialize actual_range.
34768 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34770 * df-scan.c (df_ref_create_structure): Make regno unsigned,
34771 to match the caller.
34773 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34775 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
34776 insns after final jump in test to emit dummy move.
34778 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34780 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
34781 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
34783 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
34785 * multiple_target.c (create_dispatcher_calls): Init e_next.
34786 * tree-ssa-loop-split.c (split_loop): Init border.
34787 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
34790 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
34795 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
34796 (altivec_register_operand): Do not return true if the operand
34797 contains a SUBREG mixing SImode and SFmode.
34798 (vsx_register_operand): Likewise.
34799 (vsx_reg_sfsubreg_ok): New predicate.
34800 (vfloat_operand): Do not return true if the operand contains a
34801 SUBREG mixing SImode and SFmode.
34802 (vint_operand): Likewise.
34803 (vlogical_operand): Likewise.
34804 (gpc_reg_operand): Likewise.
34805 (int_reg_operand): Likewise.
34806 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
34807 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
34808 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
34810 (rs6000_emit_move_si_sf_subreg): New helper function.
34811 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
34812 fixup SUBREGs involving SImode and SFmode.
34813 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
34814 numbers for the new peephole2 optimization.
34815 (peephole2 for SFmode unions): New peephole2 to optimize cases in
34816 the GLIBC math library that do AND/IOR/XOR operations on single
34817 precision floating point.
34818 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
34819 target macros to say whether we need to avoid SUBREGs mixing
34821 (TARGET_ALLOW_SF_SUBREG): Likewise.
34822 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
34823 (UNSPEC_SI_FROM_SF): Likewise.
34824 (iorxor): Change spacing.
34825 (and_ior_xor): New iterator for AND, IOR, and XOR.
34826 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
34827 (movdi_from_sf_zero_ext): Likewise.
34828 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
34829 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
34830 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
34831 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34832 (fms<mode>4): Likewise.
34833 (fnma<mode>4): Likewise.
34834 (fnms<mode>4): Likewise.
34835 (nfma<mode>4): Likewise.
34836 (nfms<mode>4): Likewise.
34838 2017-01-04 Marek Polacek <polacek@redhat.com>
34841 * doc/invoke.texi: Document -Wpointer-compare.
34843 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34845 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34848 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34849 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34852 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34855 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34856 documentation of the powerpc_popcntb_ok attribute.
34857 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34858 code to issue warning messages if a requested CPU configuration is
34859 not supported by the binary (assembler and loader) toolchain.
34860 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34861 made to define a built-in function that has been disabled.
34862 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34863 made to define a built-in function that has been disabled.
34864 (altivec_init_builtins): Add comment explaining why definition
34865 of the DST built-in functions is not preceded by an assertion
34866 check. Add assertions to prevent ICE if attempts are made to
34867 define an altivec predicate or an abs* built-in function that has
34869 (htm_init_builtins): Add comment explaining why definition of the
34870 htm built-in functions is not preceded by an assertion check.
34872 2017-01-04 Jeff Law <law@redhat.com>
34874 PR tree-optimizatin/67955
34875 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34876 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34877 the points-to solution does not include pt_null. Use DECL_PT_UID
34880 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34882 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34883 Use gen_int_mode instead of gen_lopwart for const_int operands.
34885 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34887 PR tree-optimization/71563
34888 * match.pd: Simplify X << Y into X if Y is known to be 0 or
34889 out of range value - has low bits known to be zero.
34891 2017-01-04 Alan Modra <amodra@gmail.com>
34893 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
34894 * configure: Regenerate.
34895 * config.in: Regenerate.
34897 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34900 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34901 a substring of the message, but strcmp with the whole message. Ifdef
34902 ENABLE_NLS, translate the message first using dgettext.
34904 2017-01-03 Jeff Law <law@redhat.com>
34906 PR tree-optimizatin/78856
34907 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34908 (mark_threaded_blocks): Remove code to truncate thread paths that
34909 cross multiple loop headers. Instead invalidate the cached loop
34910 iteration information and handle case of a thread path walking
34911 into an irreducible region.
34913 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34916 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34917 assertions. Add support for doing the signbit if the IEEE 128-bit
34918 floating point value is in a GPR.
34919 * config/rs6000/rs6000.md (Fsignbit): Delete.
34920 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34921 Update the length attribute if the value is in a GPR.
34922 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34923 the sign or zero extension instruction, since the value is always 0/1.
34924 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34927 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34928 extracting SImode to a GPR register so that we can generate a
34929 store, limit the vector to be in a traditional Altivec register
34930 for the vextuwrx instruction.
34932 2017-01-03 Ian Lance Taylor <iant@google.com>
34934 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
34936 2017-01-03 Martin Sebor <msebor@redhat.com>
34938 PR tree-optimization/78696
34939 * gimple-ssa-sprintf.c (format_floating): Correct handling of
34940 precision. Use MPFR for %f for greater fidelity. Correct handling
34942 (pass_sprintf_length::compute_format_length): Set width and precision
34943 specified by asrerisk to void_node for vararg functions.
34944 (try_substitute_return_value): Adjust dump output.
34946 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
34948 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
34950 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
34952 * doc/invoke.texi (SPARC options): Document -mlra as the default.
34953 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
34954 -mlra/-mno-lra was passed to the compiler.
34956 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
34958 PR rtl-optimization/65618
34959 * emit-rtl.c (try_split): Move initialization of "before" and
34960 "after" to just before the call to emit_insn_after_setloc.
34962 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
34964 * doc/md.texi (Standard Names): Remove reference to Java frontend.
34966 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
34968 * dwarf2out.c (gen_enumeration_type_die): When
34969 -gno-strict-dwarf, add a DW_AT_encoding attribute.
34971 2017-01-03 Jakub Jelinek <jakub@redhat.com>
34973 PR tree-optimization/78965
34974 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
34975 Change first argument from const call_info & to call_info &. For %n
34976 set info.nowrite to false.
34978 PR middle-end/78901
34979 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
34980 possibly throwing calls.
34982 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34983 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34984 and fns handling, rather than in a separate case SSA_NAME.
34986 2017-01-02 Jeff Law <law@redhat.com>
34988 * config/darwin-driver.c (darwin_driver_init): Const-correctness
34989 fixes for first_period and second_period variables.
34991 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
34994 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
34995 (*insvqi_1): New insn pattern.
34996 (*insvqi_1_mem_rex64): Ditto.
34997 (*insvqi_2): Ditto.
34998 (*insvqi_3): Rename from *insvqi.
35000 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35002 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35004 * doc/cfg.texi (Edges): Remove reference to Java.
35005 (Maintaining the CFG): Ditto.
35007 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35009 PR middle-end/77674
35010 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35011 transparent aliases.
35013 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35015 PR middle-end/77484
35016 * predict.def (PRED_CALL): Update hitrate.
35017 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35018 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35019 into direct/indirect/polymorphic variants.
35021 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35023 Update copyright years.
35025 * gcc.c (process_command): Update copyright notice dates.
35026 * gcov-dump.c (print_version): Ditto.
35027 * gcov.c (print_version): Ditto.
35028 * gcov-tool.c (print_version): Ditto.
35029 * gengtype.c (create_file): Ditto.
35030 * doc/cpp.texi: Bump @copying's copyright year.
35031 * doc/cppinternals.texi: Ditto.
35032 * doc/gcc.texi: Ditto.
35033 * doc/gccint.texi: Ditto.
35034 * doc/gcov.texi: Ditto.
35035 * doc/install.texi: Ditto.
35036 * doc/invoke.texi: Ditto.
35038 Copyright (C) 2017 Free Software Foundation, Inc.
35040 Copying and distribution of this file, with or without modification,
35041 are permitted in any medium without royalty provided the copyright
35042 notice and this notice are preserved.